Interplay between Collective and Localized Effects of Point Defects on Photoelectrochemical Performance of TiO2 Photoanodes for Oxygen Evolution

Among the various photoanode materials investigated for photoelectrochemical water splitting cells, TiO2 stands out due to its abundance, stability, and favorable valence band edge for water oxidation. In this study, the importance of introducing and combining oxygen and titanium vacancy point defects in anatase TiO2 photoanodes to improve their performance is unveiled, achieving a photocurrent density of 0.73 (±0.015) mA cm−2 at +1.23 VRHE under 100 mW cm−2 of simulated sunlight or 26.4 mA cm−2 at +1.23 VRHE under 100 mW cm−2 of 365 nm light. The characterization by X‐ray photoelectron spectroscopy, surface photovoltage, and electron paramagnetic resonance demonstrates that these oxygen and titanium vacancies can have both collective and localized positive effects on the material, leading to a narrowing of the bandgap, an increase in donor density, and an increase in hydroxyl groups on the surface of TiO2. These result in enhanced light absorption, conductivity, and photovoltage, as well as a more negative flat‐band potential and increase in hole flux to the semiconductor–electrolyte interface. These findings provide valuable insights into the role of point defects in modulating the properties of TiO2 and have important implications for the development of high‐performance TiO2‐based devices.


Introduction
Photoanodes play a critical role in photoelectrochemical (PEC) water splitting, which is a promising technology for producing sustainable and clean hydrogen fuel.Suitable n-type semiconductors are required to construct photoanodes and generate holes under illumination to efficiently oxidize water. [1][4] Different phases of TiO 2 , such as rutile, anatase, and brookite, have been extensively studied, but typically anatase shows superior performance due to lower charge carrier recombination and higher adsorption strength of hydroxyl groups.We recently developed an aerosol-assisted chemical vapor deposition (AACVD) of Ti 16 O 16 (OEt)  32  clusters that leads to a nanostructured anatase TiO 2 with exposed {0 1 0} facets.The morphology of this TiO 2 resembles the crystals of gypsum, sand and water found in nature, also known as desert roses. [5]his TiO 2 achieves high incident photon-to-current efficiency (IPCE) close to 100% at 1.23 V versus the reversible hydrogen electrode (V RHE ).This high IPCE is, however, achieved at 350 nm due to the TiO 2 bandgap of 3.2 eV, which limits the wide spectral utilization of solar light.Therefore, it is imperative to develop strategies to enhance light utilization in TiO 2 photoelectrodes.
Band structure engineering is a widely used approach to enhance the electrical and optical properties of semiconductors.Among different band structure engineering approaches, point defect introduction is becoming increasingly important for improving and tailoring bulk semiconductors and their surfaces. [6,7]This approach has been shown to improve the solar light harvesting capability, charge separation, and charge transfer.For instance, oxygen vacancies (V O ) are shallow donors that increase the charge carrier concentration, resulting in improved performance. [8,9]Recently, TiO 2 with a continuous oxygen-vacancy gradient was reported to exhibit high light absorption and considerable photoelectrochemical performance improvement. [10]In addition to V O , Ti vacancies (V Ti ) also play a crucial role in improving the oxygen evolution reaction (OER) catalytic activity of TiO 2 .However, the coexistence of V O and V Ti in TiO 2 has been rarely reported.This is because, in synthesized TiO 2 materials, V O is responsible for intrinsic n-type conductivity and V Ti for intrinsic p-type conductivity.It is challenging to simultaneously observe atomic-/nano-scaled V Ti in the macro-scaled O-vacancy phases. [11]erein, we report on the successful introduction of different point defects, namely V O and V Ti , in TiO 2 photoanodes through a chemical treatment process at room temperature, utilizing NaBH 4 as a reductant, followed by HCl washing.A range of advanced techniques were employed to investigate the intrinsic band energy structure of the synthesized TiO 2 photoanodes.The results of ex situ and in situ characterization reveal that the formation of V O and V Ti can result in favorable effects on photoanodes, leading to an enhancement in the PEC performance.The enhanced PEC performance was attributed to the interplay of collective and localized effects of point defects on TiO 2 .For the collective effects, we found that the increased point defects could lead to increased light absorption, enhanced donor density, conductivity and photovoltage, as well as a more negative flat-band potential.Additionally, our results show that the presence of localized surface V Ti surrounded by O − can increase the amount of surface hydroxyls and create a more basic environment for Ti─OH species.This, in turn, facilitates the water oxidation process by serving as trapping sites and self-reduction sites during OER.Overall, our findings shed light on the role of point defects in the water oxidation process on anatase TiO 2 and provide valu-able insights into the development of high-performance photoanodes for sustainable energy conversion applications.

Structural Characterization
TiO 2 photoanodes were synthesized through aerosol-assisted chemical vapor deposition (AACVD) at 450 °C using toluene solutions of Ti 7 O 4 (OEt) 20 precursor, nitrogen gas carrier, and fluorine-doped tin oxide (FTO) coated aluminoborosilicate glass substrate. [5]After the synthesis, the photoanodes were annealed in air at 800 °C to remove carbon residues and achieve optimal TiO 2 crystallization.Some of these photoanodes were subsequently treated using 0.1, 0.2, or 0.3 m NaBH 4 ethanolic solutions and 1 m HCl aqueous solution and named Treated TiO 2 -1, −2, or −3, respectively (Figure 1a).The phase of the synthesized photoanodes was characterized by measuring XRD diffractograms of obtained samples before and after chemical treatment.The peaks in the XRD diffractograms at 2 of 25.5, 37.9, and 47.9°correspond to (101), (004), and (200), respectively, indicating that the TiO 2 films are the anatase crystalline phase (Figure 1b; Figure S1, Supporting Information).The morphology of the resulting films was inspected using scanning electron microscopy (SEM).The SEM micrographs show a homogeneous coverage of TiO 2 on the FTO layer for both pristine TiO 2 and treated TiO 2 photoanodes in Figure 1c-f and Figures S2 and S3 (Supporting Information).The thickness of the TiO 2 films is ≈1.5 μm (Figure 2c,e inset), and the morphology was preserved after additional chemical treatment.TiO 2 grains are grown on the FTO with a good coalescence and top layers are aggregations, clusters of perpendicular faceted grains, similar to so-called "desert roses" of gypsum or baryte.Transmission electron microscope (TEM) micrographs on a scraped piece of treated TiO 2 sample show TiO 2 grains of plate shape with lengths of ≈150 nm (Figure 1g) and thicknesses of ≈10 nm (Figure 2f).The high-resolution TEM (HRTEM) micrograph in Figure 1h shows stacked grains and lattice fringes with interplanar distances of 3.35, 2.35, and 1.91 Å that are indexed to (101), (004), and (200) planes of TiO 2 , respectively.A scanning TEM (STEM) micrograph of Treated TiO 2 -2 and its corresponding energy dispersive X-ray spectroscopy (EDX) elemental mapping micrographs show a uniform distribution of Ti and O elements with no other elements present (Figure 1i-l).Overall, the results indicate that the synthesized TiO 2 photoanodes have a well-defined morphology and crystalline structure.

Photoelectrochemical Characterization
The performance of TiO 2 photoanodes was evaluated using various PEC techniques.Figure 2a shows the current-voltage curves under chopped simulated sunlight (1 sun, 100 mW cm −2 ) in 1 m NaOH aqueous solution (pH 13.7).All the TiO 2 photoanodes exhibited positive photocurrents during anodic sweeps at a scan rate of 10 mV s −1 , with an onset potential around 0 V RHE .The treated TiO 2 photoanodes showed higher photocurrents, with the optimal results achieved for Treated TiO 2 -2 sample.The current-voltage results for three series of samples were obtained, confirming reproducibility (Figure S4, Supporting Information).The optimal Treated TiO 2 -2 photoanode showed a photocurrent of 0.54 (± 0.027) mA cm −2 at +0.34 V RHE and 0.73 (± 0.015) mA cm −2 at +1.23 V RHE , compared with 0.34 (± 0.044) mA cm −2 at +0.34 V RHE and 0.53 (± 0.015) mA cm −2 at +1.23 V RHE for pristine TiO 2 photoanode.Moreover, the photocurrent of Treated TiO 2 -2 photoanode under UV light (365 nm, 100 mW cm −2 ) was 26.4 mA cm −2 at +1.23 V RHE as shown in Figure 2b, which was much higher than that of pristine TiO 2 (14.9 mA cm −2 at +1.23 V RHE ).Control tests were conducted on samples treated with 0.2 m NaBH 4 ethanolic solution and with only 1 m HCl at room temperature (Figure S5, Supporting Information).After only NaBH 4 treatment, the photocurrent of the sample increased from 0.53 mA cm −2 to 0.67 mA cm −2 at +1.23 V RHE , which was higher than that of pristine TiO 2 but slightly less than that of fully Treated TiO 2 -2 (0.73 mA cm −2 at +1.23 V RHE ).The limited improvement might be caused by the presence of coated byproducts during the treatment, such as B 2 O 3 or other boron oxide species, which are only washed away upon HCl acid wash. [12]On the other hand, after only HCl washing, the photocurrent plateau did not change, but slightly larger photocurrents were achieved at applied potentials below +0.6 V RHE (Figure S5a, Supporting Information).The photocurrent analysis confirmed that Treated TiO 2 -2 had the best PEC performance.Therefore, we focused our subsequent characterization efforts on this sample and pristine TiO 2 for comparison.
To characterize the translation of light absorption to final PEC performance, we conducted measurements of ultraviolet-visible (UV-vis) spectroscopy and IPCE for both pristine and treated TiO 2 photoanodes.As shown in Figure 2c, both samples exhibit an absorption edge corresponding to the bandgap of anatase TiO 2 of 3.2 eV.However, the absorptance of the Treated TiO 2 -2 sample is slightly red-shifted by 10-20 nm, indicating a greater ability to absorb the solar spectrum.Similarly, the IPCE curves show an IPCE edge assigned to the bandgap of anatase TiO 2 , with IPCE values approaching 100% in the UV range.The Treated TiO 2 -2 sample also exhibits a red-shifted IPCE, achieving higher values than the pristine TiO 2 sample over a broader range.This redshifted IPCE provide evidence for the treated photoanodes absorbing and converting more UV and violet light into photocurrent.To confirm that the enhanced photocurrents observed in the treated samples result in oxygen evolution reaction (OER), we conducted experiments using a gas-tight PEC cell at 1.23 V RHE for 2 h, measuring the O 2 content inside the cell with a fiber-optic oxygen meter (Figure 2d).The results indicate that the Faradaic efficiency for OER was practically 100% for the first hour and 85-90% for the second hour.We attribute the decrease in Faradaic efficiency during the second hour to diffusion or small leakage of O 2 through rubber fittings, O 2 reduction in the counter electrode, or bubble trapping at the reactor walls, resulting in a reduced O 2 amount detected by the sensor.The high Faradaic efficiency of the Treated TiO 2 -2 photoanodes confirms that the increased photocurrent results from water oxidation rather than oxidation of the TiO 2 materials.To test the stability of the samples, we conducted photocurrent measurements for 40 h under continuous simulated sunlight and 1.23 V RHE with a 30-min dark interval halfway through the experiment (Figure S6, Supporting Information).After 20 h, the Treated TiO 2 -2 sample retained 79% of its initial photocurrent, with half of the loss attributed to bubble formation on the photoanode surface and consequent blockage of the electroactive surface area.During the 30-min dark interval, we removed the bubbles by means of electrode agitation, and the photocurrent recovered 90% of its initial value.The final photocurrent after the 40-hour stability test was 76% of its initial value, indicating reasonable stability of the sample.

Energetics and Defect States of TiO 2
To understand the origin of improved PEC performance, we investigated the energy levels and in particular the defect states of the treated and pristine TiO 2 .Surface photovoltage (SPV) signals were recorded for pristine TiO 2 and Treated TiO 2 -2 by measuring the surface potential in the dark (WF dark ) and under illumination (WF light ).The magnitude and sign of SPV is defined here as the change in surface potential: SPV = WF light −WF dark .The negative sign of SPV under white light illumination (Figure 3a) is consistent with an increased density of holes at the surface of the n-type TiO 2 , which could result from photoexcitation of electrons from both the valence band as well as electronic trap states within the bandgap, as shown in Figure S7 (Supporting Information). [13,14]he larger SPV magnitude and the slower SPV rise/decay (i.e., slower trapping/de-trapping) under white light illumination suggest the presence of higher density of electronic trap states in the Treated TiO 2 -2 compared to TiO 2 .SPV spectra using a monochromator in Figure 3b show significant magnitude of SPV even under sub-bandgap illumination (i.e., lower energy compared to the bandgap of 3.2 eV), providing evidence for trap states filled with electrons within the bandgap of TiO 2 .Importantly, the larger magnitude of sub-bandgap SPV signal and a clear redshift (from 550 to 600 nm) in the SPV onset for the Treated TiO 2 -2 compared to the reference TiO 2 confirm an increased density of trap states 2.1 eV (600 nm illumination) below E c .We assign these to Ti vacancies (V Ti ) in the TiO 2 crystal structure, which were calculated to be also at 2.1 eV below E c . [15]Similar trap states related SPV signals were also recorded at lower light intensities using different monochromatic LEDs of wavelengths between 800 and 370 nm (Figure 3c).
To gain further understanding of the trap states present in the samples, EPR spectra were recorded under low temperature and with/without UV light illumination, which are depicted in Figure 3d.We compared the g values of TiO 2 and Treated TiO 2 -2, which are highly dependent on the crystallographic geometry of compounds and other magnetic couplings such as spin-orbit. [16]ompared with the EPR data with/without UV illumination, the EPR spectra with UV illumination of TiO 2 and Treated-TiO 2 -2 exhibit two sets of signals, denoted as A and B, which arise from electrons and holes, respectively.The sharp signal observed at g A1 = 1.987 and g A2 = 1.956 is attributed to electron trapping sites, indicating the presence of Ti 3+ in solid TiO 2 samples.The EPR signals with g factors g B1 = 2.014, g B2 = 2.003 and g B3 = 1.996 correspond to hole trapping sites, which are assigned to O − stabilized by surface hydroxyls in solid TiO 2 samples. [17,18]O − radicals are generated by photogenerated holes that are trapped by lowcoordinate bridging O 2− anions or pre-existing O − .Upon chemical treatment, the Treated TiO 2 -2 sample exhibits increased signal intensities for both A and B, indicating an increase in the amounts of Ti 3+ and O − species.These species have the capability to compensate for O 2− deficiencies (V O ) and Ti deficiencies (V Ti ) resulting from the chemical treatment.
To investigate the composition and surface chemical environment of TiO 2 and Treated TiO 2 samples, X-ray photoelectron spectroscopy (XPS) analysis was carried out.The survey XPS spectra in Figure S8 (Supporting Information) reveal that both TiO 2 and Treated TiO 2 -2 consisted of the same elements, including Ti, O and adventitious C. The high-resolution XPS spectra of Ti 2p and O 1s regions of samples are presented in Figure 3e,f.The Ti 2p spectrum of TiO 2 showed peaks at 464.2 and 458.5 eV in Ti 4+ state, corresponding to Ti 2p1/2 and 2p3/2, respectively.In contrast, the Ti 2p spectrum in Treated TiO 2 samples (in the Ti 4+ state) showed a small shift to lower binding energy, and peaks of Ti 3+ were observed at 463.2 and 457.5 eV, which is assigned to the reducing effect of NaBH 4 . [19]The O 1s XPS spectra were identified via two peaks in Figure 3f.The peak at 529.8 eV was allocated to lattice oxygen (Ti-O band), where a small shift to lower binding energy occurred in Treated TiO 2 samples, indicating lower Ti coordination due to the increased amount of V O in TiO 2 as shown in EPR results.The other peak is associated with chemisorbed O, of which the intensity increased significantly in Treated TiO 2 -2 after chemical treatment.The formation of O 1s peak about 529-532 eV is quite complex and can be deconvoluted to different oxygen species.According to previous studies, this peak could be allocated to hydroxy species of surface-adsorbed water molecules (∼532 eV) and defect sites with low oxygen coordination (∼531 eV). [20] Therefore, the increased peak intensity (529-532 eV) can be attributed to the influence of these increased oxygen species simultaneously.
The valence band density of states (DOS) of TiO 2 and Treated TiO 2 -2 were also measured using valence band XPS and are presented in Figure 4a.The valence band DOS of pristine TiO 2 exhibited typical characteristics of TiO 2 , with the valence band edge (E V ) 2.85 eV below the Fermi level.In contrast, the E V of Treated TiO 2 -2 was 0.12 eV shifted to lower binding energy, which is attributed to the increased amount of V Ti .These observations are consistent with the SPV results, indicating the introduction of V Ti in the sample after chemical treatment.
Work function measurements show that the Fermi level of the Treated TiO 2 -2 (-4.40 eV) is shallower than of the pristine TiO 2 (−4.66 eV) (Figure 4b).The optical bandgap values of TiO 2 and Treated TiO 2 -2 (Figure S9d, Supporting Information) and the E V values allow the calculation of the E C values at -4.18 and -3.98 eV, respectively.These energy levels confirm a stronger n-type characteristic of the Treated TiO 2 -2, which is typically related to the V O acting as electron donors. [23]Consequently, the shallower Fermi level of the Treated TiO 2 -2 can be linked to an increased density of V O , which is in good agreement with the XPS and EPR results.

Mechanism of Point Defect Formation
Based on the analysis conducted above, a higher concentration of Ti 3+ and O − was observed in Treated TiO 2 samples.In this section, we propose a possible mechanism for the formation of these point defects in TiO 2 .During the chemical treatment process, Ti 4+ underwent reduction to Ti 3+ along with oxygen-vacancy sites [V O • Ti 3+ ], while V Ti was produced along with O − pieces [V Ti • O − ] due to the O-rich microenvironment formed by the escaping oxygen from V O . [11]nitially, Ti 4+ could be reduced to Ti 3+ by NaBH 4 , following the reaction along with the generation of oxygen vacancy sites [V O • Ti 3+ ]. [12] The formation of V O and V Ti can be demonstrated using the reaction equations using Kröger-Vink notation: [11,16,24] V O formation : where O O represents lattice oxygen, Ti Ti lattice titanium, V   4d.The introduction of V Ti in Treated TiO 2 -2 results in a higher E V , while V O causes a higher Fermi level and smaller energy bandgap.

Effects of the Point Defects
The influence of point defects on light absorption was examined using UV-vis spectroscopy measurements for TiO 2 and Treated TiO 2 -2, which are shown in Figure S9 (Supporting Information).To exclude the influence of acid wash on light absorption, the UV-vis absorbance spectrum of TiO 2 washed by HCl was also measured (Figure S9b, Supporting Information), indicating that acid wash did not affect the light absorption of TiO 2 samples.Compared to TiO 2 , a shift in the reflectance curve edge toward higher wavelengths was observed after chemical treatment, indicating higher light absorption in Treated TiO 2 -2.The Kubelka-Munk function of optical reflectance curves for pristine TiO 2 and Treated TiO 2 -2 in Figure S9d (Supporting Information) reveals that the bandgap of pristine TiO 2 and Treated TiO 2 -2 is ≈3.27 and 3.19 eV, respectively.Band tail states were also observed after chemical treatment.The decreased bandgap and band tail states as well as the increased light absorption after chemical treatment are assigned to the increased presence of point defects, which is consistent with the SPV and EPR results.
To further investigate the bulk and surface properties of the photoanodes, we conducted electrochemical impedance spectroscopy (EIS) measurements in the dark.The obtained EIS data are presented in the form of Mott-Schottky (M-S) plots at 500, 1000, and 2000 Hz, as illustrated in Figure S10 (Supporting Information).The frequency dispersion of the Mott-Schottky plots originates from current and potential distributions due to interface inhomogeneities, which might be attributed to surface disorder, roughness, and electrode porosity. [25]The M-S plots revealed that the synthesized photoanodes exhibited a positive slope, which is a typical characteristic of n-type semiconductors.
To determine flat-band potential (E fb ), we analyzed the M-S relationship by considering the capacitance of the space charge layer (C sc ) of the semiconductor electrode. [26]The E fb values observed in Figure S10c (Supporting Information) exhibited a small variation of only 0.04 V between different frequencies, indicating a reasonable level of accuracy in the measured values for E fb .The E fb was estimated to be +0.19 and +0.14 V RHE for pristine TiO 2 and Treated TiO 2 -2, respectively, which is consistent with previous energy level measurements (Figure 4).Theoretically the E fb of photoanodes in a certain electrolyte is dependent on the bulk Fermi level of the semiconductor and the potential drop of Helmholtz layer.The doping concentration affects the bulk Fermi level of the semiconductor, while surface conditions influence the potential drop of the Helmholtz layer.Therefore, the decrease in E fb observed after chemical treatment may be attributed to an increase in donor density caused by point defects in the photoanodes.Additionally, the smaller slope values of M-S from Treated TiO 2 -2 indicate an increased donor density, which promotes band bending and facilitate charge transport and charge injection, [27] consistent with the enhanced SPV intensity after chemical treatment, as well as the observed higher PEC photocurrent and higher IPCE efficiency.
To further explore the influence of chemical treatment on the photoanode performance, the contact angles of pristine TiO 2 and Treated TiO 2 -2 with water were also measured and presented in Figure S11 (Supporting Information).The contact angle of Treated TiO 2 -2 was found to be much smaller than that of pristine TiO 2 (10°vs 40°), indicating that the surface of Treated TiO 2 -2 with its higher presence of point defects is more hydrophilic.A more hydrophilic photoanode surface is expected to improve the first OER steps of hydroxyl adsorption, thereby improving photoanode currents.To better understand the effects of the chemical treatment, proton magic angle spinning nuclear magnetic resonance ( 1 H MAS NMR) spectra of the samples were collected.The results of 1 H MAS NMR are shown in Figure 5a.Two peaks, symmetrical about the centre, at 25.9 and -11.7 ppm could be observed, which are spinning side bands (SSB) that arise due to incomplete averaging of the homonuclear dipole-dipole interaction which predominates in 1 H NMR spectra.Both samples were evacuated to remove excess water.The pristine TiO 2 and Treated TiO 2 -2 samples show an intense peak ≈7.2 ppm (peak A) and two lower intensity peaks at 2.6 ppm (peak B) and -0.6 ppm (peak C), indicating that there are at least three different types of chemical shift site for protons.The intense peak A is assigned to protons of residual adsorbed water.The linewidth of peak A is dominated by water mobility effects. [28]It is slightly narrower in pristine TiO 2 than Treated TiO 2 -2.For Treated TiO 2 -2, peak A is wider and increased in intensity, due to the more strongly adsorbed water protons of Treated TiO 2 -2.This may be due to the increased number of potential hydrogen bonding sites present in Treated TiO 2 -2.Peak B and C originate from hydroxyl protons (Ti─OH), with peak C being assigned to a more basic-type chemical environment. [29]For peak B, the major change is in both the peak shape, the intensity and a slight shift toward the right-hand side of the spectrum, as it is now much broader than the pristine TiO 2 peak.The increase in intensity is due to the higher concentration of hydroxyl groups of Treated TiO 2 -2 compared to pristine TiO 2 and may indicate a higher concentration of oxygen deficiencies where adsorbed water can be dissociated. [30]The increase in basic character of the Ti-OH, for both peak B and C, causes a movement toward the right-hand side of the spectrum. [28]In Treated TiO 2 -2, this indicates that part of absorbed groups is more basic (more shielded).This agrees with the analysis of SPV, EPR, and XPS, where it was observed that there was an increase in the concentration of O − species, which in turn generated a more basic environment.It has been demonstrated that hydroxyls on a catalyst surface can significantly enhance its catalytic activity. [31]irst, they can act as trapping sites, facilitating rapid proton exchange between the defect-involved bridging hydroxyl groups and the adsorbed molecules. [32]Second, the self-reduction of surface hydroxyl into water can significantly reduce the overpotential for OER, thereby boosting the OER activity of catalyst. [33]herefore, the improved PEC performance of Treated TiO 2 -2 may also be attributed to the increased water adsorption and hydroxyls on the surface after chemical treatment.
Open circuit potential (OCP) measurements were also conducted to further research and compare the PEC response of these TiO 2 samples.Figure S12a,b (Supporting Information) shows the OCP curves for both pristine TiO 2 and Treated TiO 2 -2 photoanodes measured under intermittent visible light at different light intensities, with the OCP response to illumination shifting to smaller potential.The initial negative shift in OCP was due to the transfer of the excited photoelectrons from the bulk to the surface. [34]The relatively stable OCP values observed thereafter can be attributed to the balance between the generation and recombination of photoexcited electrons. [35]Upon switching off the irradiation, the OCP values for Treated TiO 2 -2 were observed to return to their original values at a slower speed compared to TiO 2 .This phenomenon may be due to the electron pool effect of Treated TiO 2 -2, which is enhanced by a higher donor intensity after chemical treatment.The electron pool effect allows for the storage of photoinduced electrons under light irradiation and their slow release in the absence of light.Figure S12c (Supporting Information) displays the OCP values of both pristine TiO 2 and Treated TiO 2 -2 photoanodes, with the latter exhibiting a higher donor intensity following chemical treatment.Additionally, the illuminated OCP can be utilized to estimate the E fb when the illumination intensity exceeds 300 mW cm −2 , which is sufficiently intense to fully eliminate pre-existing band bending at the surface. [26,36]Under light irradiation, the OCP value for TiO 2 was approximately +0.14 V RHE , while the OCP values for Treated TiO 2 -2 reached +0.12 V RHE , indicating that Treated TiO 2 -2 photoanode has a smaller E fb and exhibits the same trend as the M-S measurement results.The photovoltages (V ph ) of the TiO 2 and Treated TiO 2 -2 photoanodes were calculated and presented in Figure S12d (Supporting Information).The Treated TiO 2 -2 photoanode possesses a higher V ph value than pristine TiO 2 , indicating a greater driving force for the OER reaction and higher internal electric fields that can minimize charge recombination.

OER Kinetics
To characterize the OER kinetics of these photoanodes, we used photoelectrochemical impedance spectroscopy (PEIS) and intensity modulated photocurrent spectroscopy (IMPS).PEIS is a powerful method for analyzing charge transfer and recombination processes at semiconductor-electrolyte interfaces, and we conducted it at various applied biases.Figure 5b shows typical Nyquist plots with the equivalent circuit obtained in 1.0 m NaOH aqueous solution for the illuminated photoanodes.The equivalent circuit includes two resistances, one attributed to the electrolyte and conductive substrate layer (R s ), and the other attributed to the TiO 2 -electrolyte junction (R ct ).A constant phase element (CPE) was applied to account for the distribution of the time constants in porous and defective photoanodes. [21]The R ct values as a function of applied bias are shown in Figure 5c S13 (Supporting Information).The results show that the R ct and C s values of the FTO base did not change with/without light and were lower than the TiO 2 samples, which also shows that the behavior of extracted capacitance is from TiO 2 samples rather than FTO base.In addition, electrochemically active surface area (ECSA) measurements and corresponding CV curves are shown in Figure S14 (Supporting Information).In such measurements, the slope of the current density versus scan rate can be related to the double layer capacitance (C dl ), which is directly proportional to the ECSA.Based on the obtained results, Treated TiO 2 -2 possesses a slightly higher surface area (C dl = 0.037 mF cm −2 ) than pristine TiO 2 (C dl = 0.028 mF cm −2 ), suggesting that the chemical treatment has a small influence on the surface morphology of TiO 2 .This finding contrasts with the smaller C s values obtained on Treated TiO 2 -2 via PEIS, which can be attributed to the fact that the ECSA measurements were performed in the dark, whereas the PEIS measurements were conducted under light illumination.We attribute the slightly smaller Cs value of Treated TiO 2 -2 to the presence of occupied surface states resulting from an increased number of potential hydrogen and hydroxyl bonding sites under light illumination.IMPS was conducted to investigate the reaction kinetics on pristine TiO 2 and treated TiO 2 -2 photoanodes.Similar to PEIS plots, depressed semicircles were observed owing to distributed time constants.Typically, the IMPS spectrum for an ntype photoanode consists of two semicircles located in the first and fourth quadrant. [37]The high frequency semicircle starts in the fourth quadrant, representing the attenuation by the total resistance of the cell and combined space charge capacitance and Helmholtz layer capacitance.By decreasing the frequency of perturbation surface recombination and charge transfer processes dominate, giving another semicircle in the first quadrant. [38,39]n Figure 5e,f, only semicircles in the fourth quadrant are observed in the entire applied bias range.This suggests negligible surface recombination on pristine TiO 2 and Treated TiO 2 -2 photoanodes, which is consistent with the high IPCE values.The high frequency intercept point of Treated TiO 2 -2 is notably higher than that of the TiO 2 sample, indicating a higher hole flux at the surface, consistent with the larger photovoltage in OCP measurements (Figure S12d, Supporting Information).
We propose a mechanism to explain the enhanced PEC performance upon chemical treatment with NaBH 4 and HCl wash.The efficiency of a photoanode can be evaluated by considering the following three consecutive steps: light absorption ( abs ), separation of electrons and holes to the current collector and electrode surface, respectively, ( sep ), and gas evolution reaction ( rxn ), as shown in Equation ( 7). [40] cell =  abs ×  sep ×  rxn (7)   First, the presence of point defects in TiO 2 narrows the bandgap and improves  abs .Second, the introduction of point defects increases the donor density of TiO 2 , which enhances band bending and facilitates charge transport and charge injection, leading to improved  sep .Additionally, the chemical treatment process results in the formation of V Ti due to the O-rich microenvironment created by the oxygen escaping from V O .The presence of surface V Ti surrounded by O − can enhance the number of hydroxyls on the surface, leading to the production of a more basic environment for Ti─OH.The increased hydroxyl density, in turn, facilitates the water oxidation process because they are adsorption sites and intermediates for OER, [32,33] ultimately improving the reaction efficiency  rxn .Thus, due to the interplay of collective and localized effects of point defects, the PEC performance of TiO 2 is enhanced upon chemical treatment.

Conclusion
Our study has demonstrated the successful introduction of point defects (Vo and V Ti ) in AACVD nanostructured TiO 2 photoanodes through chemical treatment involving NaBH 4 and HCl.This treatment led to enhanced photoelectrochemical water oxidation in alkaline electrolyte.Treated TiO 2 photoanodes exhibited a photocurrent density of 0.54 mA cm −2 at +0.34 V RHE and 0.73 mA cm −2 at +1.23 V RHE with a high IPCE efficiency near 100%, surpassing the performance of pristine TiO 2 .Various spectroscopic techniques confirm the introduction of point defects and the interplay of their collective and localized effects which lead to increased light absorption, donor density, conductivity and photovoltage, and reduced flat-band potential, as well as increased hole flux at the TiO 2 surface.Our findings demonstrate a straightforward pathway to introduce point defects on oxide semiconductor surfaces and emphasize their importance and complex role to advance photoelectrodes for solar fuel technologies.

Figure 2 .
Figure 2. a) Current-voltage curves of TiO 2 and Treated TiO 2 photoanodes under 1 sun (Xe source, AM 1.5G filter, 100 mW cm −2 ) chopped illumination at a scan rate of 10 mV s −1 .b) Current-voltage curves of photoanodes under UV light of 365 nm (100 mW cm −2 ) and in the dark.c) IPCE spectra (solid line, left) at 1.23 V RHE and UV-vis absorptance (dash line, right).d) Amount of measured and theoretical oxygen evolution and calculated Faradaic efficiency of Treated TiO 2 -2 at an applied bias of 1.23 V RHE .All these measurements were carried out in 1 m NaOH aqueous solution.

Figure 3 .
Figure 3. a) Surface photovoltage (SPV) signals of TiO 2 and Treated TiO 2 -2 with white light illumination (20 mW cm −2 ).Inset: normalized SPV rise and decay curves.b) SPV spectra recorded in the wavelength range of 500-800 nm with 5 nm step size.c) SPV signals with monochromatic LED illumination with wavelengths between 370 and 800 nm.d) EPR spectra at low temperature (4.2 K) under UV light of 360 nm and in the dark.e) High-resolution Ti 2p XPS spectra and f) O 1s XPS spectra.

Figure 4 .
Figure 4. a) Density of states of the valence band of TiO 2 and Treated TiO 2 -2.b) Fermi level energy of TiO 2 and Treated TiO 2 -2 calculated from the contact potential difference (CPD) between tip and samples.c) Atomic structure diagram before and after chemical treatment.d) Schematic energy diagram of TiO 2 and Treated TiO 2 -2.

Figure 5 .
Figure 5. a) 1 H MAS-NMR spectra of TiO 2 and Treated TiO 2 -2 at room temperature.b) Equivalent circuit and Nyquist plots at +0.83 V RHE in 1 m NaOH solution with pH 13.7 under simulated sunlight (Xe source, AM 1.5G, 100 mW cm −2 ).c,d) R ct and C s as a function of applied bias between +0.20 and +1.3 V RHE in 1 m NaOH solution with pH 13.7 under simulated sunlight.e,f) IMPS spectra as a function of applied bias between +0.1 and +1.3 V RHE in 1 m NaOH solution under 365 nm LED (illumination (37.5 mW cm −2 ) with a modulation of 50% in light intensity, over a frequency range from 10 3 to 0.1 Hz at each potential step.
. The R ct values of the treated TiO 2 photoanodes were smaller than those of pristine TiO 2 , indicating a more effective separation and easier charge transfer of photogenerated electron-hole pairs across the TiO 2 -electrolyte junction.The results show that the introduced point defects must serve as intermediate surface states (i-ss) on photoanodes, facilitating the interfacial charge transfer process.Capacitance (C s ) values were calculated from CPE values and shown in Figure 5d.The C s in treated TiO 2 -2 was close to pristine TiO 2 , and the slightly smaller C s might be attributed to the occupied surface states by absorbed water and hydroxyls.To exclude the influence on the behavior of R ct and Cs of the FTO base on the results, the R ct and C s values of a bare FTO base were measured and added to Figure 5c,d.The Rct and C s values were much smaller than those of TiO 2 and Treated TiO 2 -2 samples, showing the small influence of FTO on the measurement.The R ct and C s values of the FTO base, TiO 2 and Treated TiO 2 samples were also calculated at +0.83 V RHE under both light and dark in 1 m NaOH, as shown in Figure