Decoupling of Light and Dark Reactions in a 2D Niobium Tungstate for Light-Induced Charge Storage and On-Demand Hydrogen Evolution

The direct coupling of light harvesting and charge storage in a single material opens new avenues to light storing devices. Here we demonstrate the decoupling of light and dark reactions in the two-dimensional layered niobium tungstate (TBA)+(NbWO6)− for on-demand hydrogen evolution and solar battery energy storage. Light illumination drives Li+/H+ photointercalation into the (TBA)+(NbWO6)− photoanode, leading to small polaron formation assisted by structural distortions on the WOx sublattice, along with a light-induced decrease in material resistance over 2 orders of magnitude compared to the dark. The photogenerated electrons can be extracted on demand to produce solar hydrogen upon the addition of a Pt catalyst. Alternatively, they can be stored for over 20 h under oxygen-free conditions after 365 nm UV illumination for only 10 min, thus featuring a solar battery anode with promising capacity and long-term stability. The optoionic effects described herein offer new insights to overcome the intermittency of solar irradiation, while inspiring applications at the interface of solar energy conversion and energy storage, including solar batteries, “dark” photocatalysis, solar battolyzers, and photomemory devices.


NbWO6 nanosheets synthesis
Layered α-LiNbWO6 powder was synthesized as reported in the literature. 1 Briefly, Li2CO3 (99.999%,Acros), Nb2O5 (>99.9%,Roth) and WO3 (Aldrich) in a molar ratio of 1:1:2 were mixed thoroughly by grinding the mixture for around 15 min.The mixture was calcined at 760 °C for 24 h in the air.The protonated HNbWO6•xH2O was obtained by treating bulk α-LiNbWO6 powder with 0.1 M HCl solution for 3 days and replacing new acid solution every day, followed by washing thoroughly with water and drying at room temperature.The exfoliated NbWO6 nanosheets were obtained by adding an equimolar amount of tetrabutylammonium hydroxide (TBAOH, 40 wt%, Acros) solution into HNbWO 6 •H 2 O suspension, followed by stirring for 7 days at room temperature.The suspension was centrifuged at 2 000 rpm for 10 min and collected the top suspension for future experiments.

Photoanode fabrication
The photoanodes were prepared by drop casting 50 µL 5 mg/mL NbWO6 nanosheets suspension on oxygen plasma cleaned fluorine-doped tin oxide glass (FTO, Sigma-Aldrich)   substrates with size of 10 × 12 mm 2 , and dried on a hot plate at 60 °C for 15 min, followed by annealing at 200 °C in the air for 1 h.To complete the photoanodes fabrication, about 12 mm long copper wire was connected to FTO layer by silver paste.The contact area was sealed with epoxy glue (DP410, 3M Scotch-Weld), leaving an active electrode area of approximately 10 × 10 mm 2 .The photoanode thickness is around 850 nm, the mass loading is around 0.20 mg.

Characterizations
Atomic force microscopy (AFM) was performed on Bruker Dimension ICON under Peak Force Tapping mode.The AFM data were analyzed by Gwyddion (version 2.59) software.Scanning electron microscopy (SEM) was conducted on Zeiss Merlin.Transmission Electron Microscopy (TEM) was performed on a Philips CM 30 ST microscope (300 kV, LaB6 cathode).Images were taken with a TVIPS TemCam-F216 CMOS Camera.The program EM-Menu 4.0 Extended was used to perform Fast Fourier Transformations (FFT).Powder X-ray diffraction (PXRD) was conducted on a STOE Stadi P diffractometer (Ag Kα1, Johann-type Ge111 monochromator, triple array of Mythen (Dectris) detectors) in a Debye-Scherrer configuration.The samples were sealed in 0.5 mm borosilicate glass capillaries (Hilgenberg, glass No. 14), which were spun during the measurements applying a total scan time of 3 hours.Temperature dependent in situ PXRD measurements were performed using the same device.A capillary of HNbWO6•xH2O was heated with a hot air blower (Large Hot Air Gas Blower DGB0001 FMB Oxford).The sample was heated from 25 °C to 300 °C in 25 K steps applying a heating rate of 3K/min.During isothermal hold periods, RXRD patterns were recorded applying a total scan time of one hour and an isothermal delay of 2 minutes prior to every measurement for ensuring thermal equilibration.
X-ray photoelectron spectroscopy (XPS) was conducted on a Kratos Axis Ultra system with monochromated Al Kα x-ray source (hν = 1486.6eV) under a base pressure better than 5x10 -5 mbar.A charge neutralizer was used to compensate for the sample charging.For binding energy calibration, the main C 1s peak was set to 284.8 eV (adventitious carbon). 2 Casa XPS was used to analyze the data: After subtracting a Shirley background, the peaks were fitted with LA lineshapes, which is a numerical convolution of a Lorentzian and Gaussian function. 3r the W 4f7/2 and W4f5/2 the binding energy and area ratio were constrained to 2.18 eV and 4:3, respectively.We observed an increase of the W 5+ amount and the appearance of some Nb 4+ during longer measurement times.We relate this to some reduction of the NbWO6 probably due to the exposure of the sample to x-rays and low energy electrons by the charge neutralizer.Therefore, the data shown in the main manuscript consist of single Nb 3d and W 4f scans on a fresh sample, which was not exposed to any x-rays or electrons before this measurement started.
Inductively coupled plasma-optical emission spectrometry (ICP-OES) was performed on Varian Vista-PRO (simultaneous ICP-OES spectrometer with axial plasma (Fa.Varian Darmstadt)).The sample was dissolved in HNO3 (65%), HF (40%) and H3PO4 at 165°C for 35min which was diluted with double distilled water.The microwave digestion with Discover SP-D is from CEM GmbH.ICP-Expert software was employed to analyze the data.(In-situ) Ultraviolet-visible (UV-Vis) spectroscopy was performed on Agilent Cary 60 spectrophotometer in transmission and absorption modes.To perform in-situ UV-vis absorption measurement on suspension, the 365 nm UV light was illuminated from the top of a quartz cuvette with a distance of 20 cm.In-situ UV-Vis transmission under different applied bias potentials was conducted by connecting the potentiostat (Autolab PGSTAT302N, Metrohm) with the electrode, which was immersed in oxygen-free 1 M LiCl electrolyte.Ultraviolet-Visible-Near Infrared (UV-Vis-NIR) spectra was conducted on Cary 5000 UV-Vis -NIR (Agilent Technologies).

PEC measurement
All (photo)electrochemistry measurements were performed in a home-made closed glass reactor equipped with a quartz window for light illumination.An Ag/AgCl electrode (saturated KCl, RE-1CP) was used as reference electrode and an Au foil was used as counter electrode.
Unless otherwise specified, all potentials in this work were measured versus Ag/AgCl.Oxygenfree 1M LiCl (Roth) aqueous solution was used as electrolyte.Methanol, 4-methylbenzyl alcohol (4-MBA) and water were used as donor.The electrolyte was purged with Ar for at least 30 min prior to every measurement through a porous glass frit to remove dissolved oxygen.
Artificial sunlight was provided by a Sciencetech LightLine A4 solar simulator (class AAA) fitting the ASTM standard G138 (AM 1.5G).The 365 nm UV illumination was provided by Thorlabs M365LP1-C4 lamp at an operating current of 1700 mA.The sunlight intensity was measured by a calibrated Thorlabs S130C/PM100D thermal power meter.The light intensity for 365 nm UV was measured by a calibrated Thorlabs S120VC standard photodiode power sensor and PM100D Thorlabs power meter.The solar-to-output efficiency was calculated from photocharging and electric discharging measurements.The theoretical capacity is calculated by:  theoretical =  3.6 = 1 × 96,485.3C mol −1 3.6  mA −1 h −1 × 372.74 g mol −1 = 71.9mAh g −1 Where n is the number of electrons transferred per formula unit, we only count W here.F is Faraday's constant, 3.6 is a conversion factor between coulombs and mAh, and m is the molar mass per formula unit.
(Photo)electrochemical measurements were performed on a multichannel potentiostat (Autolab M204, Metrohm).The dark electrochemical impedance spectroscopy (EIS) was performed under different applied bias potentials from 10 kHz to 0.1 Hz.The light EIS was conducted under 340 nm UV LED and 1 sun irradiation for different time from 10 kHz to 0.1 Hz.
The real capacitance C' is defined as 4 : Where -Z'' is the imaginary impedance, ω is the frequency, Z is the electrochemical impedance.

Dark photocatalysis
The dark hydrogen evolution experiments were conducted in a home-made glass reactor with quartz window on top for illumination and thermostated at 25 °C as previous described 5,6 .
The reactor was soaked in aqua regia overnight before dark photocatalysis experiment.The 365 nm UV illumination was provided by Thorlabs M365LP1-C4 lamp at an operating current of 1700 mA.The NbWO6 nanosheets (95 mg) were dispersed in water (8 mL) with the presence of MeOH (1 mL).The headspace of the reactor was evacuated and argon backfilled several times to remove the air.The suspension was stirred at 400 rpm during illumination.
For the dark hydrogen evolution experiments, the platinum nanoparticles catalyst (1 mL of 1000 ppm aqueous colloidal solution, Aldrich) was injected into the photocharged suspension with different delay time.To measure the amount of hydrogen evolution upon adding the Pt catalyst, the headspace of the reactor was periodically sampled and the amount of evolved gases was quantified by gas chromatography (Shimadzu GC-2030).This gas chromatograph is equipped with a Barrier Discharge Ionization detector (BID) and Thermal conductivity detector (TCD) using argon as the carrier gas.Unless stated differently, evolution of gases in the headspace is measured in a closed system (i.e., batch measurements).Control experiments confirmed that no hydrogen was generated in the absence of NbWO6, light or Pt catalyst.

Computational details
We conducted electronic structure calculations utilizing the FHI-aims program package.All structural optimizations employed the HSE06 hybrid functional to ensure precise electronic structures.Owing to the size of the models, numerical convergence was attained with a light basis set.The reciprocal space was sampled through a 2 × 2 × 1 Monkhorst-Pack k-point grid.The structures were utterly relaxed until the forces fell below 5 × 10 −2 /Å.
To identify potential polaron formation sites, we initially executed relaxation by applying PBE+U and implementing a Hubbard U correction on a single atom of the chosen element.In this particular case, we determined that  = 1.2 was sufficient for pre-relaxing the structure towards the desired polaron formation, using a tight basis set.Subsequently, the pre-relaxed structure was subjected to a comprehensive relaxation protocol at the HSE06 level without the +U correction, yielding the final geometry and electronic structure.

Models
The single-layer NbWO 6 ⬚ / NbWO 6 − structures were derived from the pristine  6 structure obtained from the Materials Project.[10.17188/1278002]The two-hydroxyl model was constructed by incorporating two hydrogen atoms symmetrically onto oxygen atoms in the single layer.The oxygen vacancy model was generated by eliminating one oxygen atom bound to one of the tungsten atoms.

Virtual crystal approach
In order to model the negatively charged single-layer NbWO 6 − system under periodic boundary conditions, we employed the virtual crystal approach (VCA) by adjusting the charge of atomic nuclei.A minuscule extra charge   was incorporated into each nucleus.The   value is system-dependent and conforms to .species, we only modeled the negatively charged single layer within the simulation box.The models and their corresponding additional   values are presented in Table S1.
Table 1.The added nuclear charge,   , for each model.

XRPD data analyses and crystal structure refinement
The program TOPAS 6.0 (ref.S2, second column).However, they point out the possibility of a cation disorder, which they could not rule out due to the limited quality of the sample and the X-ray and neutron scatting data.Using a structure model of α-LiNbWO6 with ordered cations only led to a very poor fit of our diffraction data (Fig. S2a).An inspection of the Fourier map revealed considerable positive residual electron density at the lithium site (=metal( 1)), slight positive residual electron density at the niobium sites (= metal( 2)) and considerable negative residual electron density at the tungsten site (= metal( 3)) indicating occupational disorder among the cations.In the first attempt we tried to model the disordered cation substructure isotropically, i.e. the lithium site is partially substituted by equivalent amounts of tungsten and niobium, the niobium site is partially substituted by equivalent amounts of tungsten and lithium and so on (Table S2, third column).The led to a significant improvement in the refinement (Fig. S2b) and yielded an acceptable R-wp value (4.03 %).
However, there is still some misfit and this model does not properly account for what was observed in the residual electron density map.In X-ray diffraction the scattering power roughly scales with the number of electrons.For the cations we consider: Li + = 2 electrons, Nb 5+ = 36 electrons and W6+ = 68 electrons.The significant positive residual electron density observed for the lithium position in the refinement using a model with ordered cations, can be explained by the presence of niobium and/or tungsten on this position.The negative residual electron density observed for the tungsten position can be explained by the presence of niobium and/or lithium on this position, with lithium having the stronger impact.For the niobium position a slightly positive residual electron density was observed.If niobium cations (36 electrons) were substituted isotropically by lithium and tungsten cations ((2 electrons + 68 electrons)/2 = 35 electrons) this would yield in hardly detectable negative residual electron density.Only an excess of tungsten can lead to a slightly positive residual electron density.
Hence, the presence of an anisotropically disordered cation substructure appears to be more suitable.For testing this, we refined a structure model in which lithium and niobium are partially replaced by tungsten and tungsten is partially replaced by both lithium and niobium (Table S2, fourth column).This led to an additional improvement of the fit (Fig. S2c) and a lower R-wp value (3.74 %).Since we cannot determine the distribution of the cations among the metal positions and as the refined atomic coordinates do not differ significantly from the dataset published by Fourquet et al. 1 we did not deposit a new crystal structure dataset into the databases.
Table S2.LiNbWO6 in which all metal cations are occupationally disordered lithium cations cannot be exchanged by protons, we believe that the proton exchange is incomplete in the LiNbWO6.
An intergrowth of proton-exchange and non-exchanged layers in the material may lead to pronounced structural disorder.Temperature dependent in situ PXRD measurements (Fig. S4) give insights into the HNbWO6•xH2O material.Moderate heating at 75 °C leads to a significant change in the diffraction pattern (Fig. S4a, i and ii).The basal reflection corresponding to a d-spacing of 13.01 Å is shifted towards 10.7 Å (Fig. S4b), i.e. the interlayer spacing is significantly contracted.This points to the presence of loosely bound water molecules in-between the layers.In addition, broad peaks situated at 7.8 and 11.2 ° 2θ disappear indicating a reduction of structural disorder by the release of water molecules.Further heating above 150 °C leads to a gradual broadening of the basal reflection, which eventually disappears, whereas all other peaks remain sharp.This is most likely attributed to either the gradual release of additional water molecules or the dehydration of hydroxide groups, which leads to an increasing modulation of the interlayer distance and at the end of the process the material loses its layered character.Neither the reflections of hydrated HNbWO6 nor of dehydrated HNbWO6 can be assigned to any known compound.Peaks attributed to the minor impurities LiNbWO6 and LiNb3O8 (Fig. 4b, magenta and green tick marks) can be observed throughout the entire temperature range.(Fig. S8d), which are 70 W, 1.2 A, 1.5 V and 140 W, 2A, 3.5 V, do not show significantly different W(VI): W(V) ratios, which are around 7:1 and 6:1, respectively.However, we observed an increase in the fraction of W(V) over the scanning time (Fig. S8e).The last spectrum in Fig. 8e was taken approx.2 h 15 min after the first one.Therefore, we used the first scanning data, which we approximate to the "pristine state" in the manuscript.We measured UV-Vis absorption spectra to quantify the blue color change after light illumination.The absorbance at a wavelength of 575 nm in the presence of MeOH donor under 365 nm UV illumination is higher than for 4-MBA in water, while the absorption intensity in water is the lowest.These results further confirm that MeOH shows higher efficiency for hole quenching than 4-MBA and water.The electrode shows high transparence at OCP conditions.There is no color change when charging to -0.5 V.When further charging the electrode to -0.65 V, the electrode exhibits light blue color.A significant color change is observed when charging the electrode to -0.7 V.The electrode shows similar light blue color when discharging it to -0.65 V.The electrode becomes highly transparent when discharging to -0.5 V, indicating a highly reversible charge and discharge processes.As shown in Fig. S26, the NbWO6 electrode in oxygen-free 1M LiCl changes to blue color under the electric charge and discharge processes.To quantify the relationship between the intensity of the blue color and applied potential, we performed in-situ UV-vis transmittance spectra.As shown in Fig. S27a-b, the transmittance starts to decrease and the absorption starts to increase when the electrode is charged to -0.6 V from OCP, indicating charge storage.
The transmittance continuously decreases when the electrode is charged to more negative potential, suggesting charge accumulation in the electrode.Upon discharge, the transmittance starts to increase when the electrode is discharged to 0 V.A summary of absorption intensity at 575 nm upon electric charge and discharge is shown in Fig. S27c.It is noteworthy that the electrode absorbance intensity exhibits high reversibility.pseudocapacitive behavior, which is typical for 2D materials. 14 gain insights into the nature of charge transfer and mechanism of charge storage in the NbWO6 thin film electrode, electrochemical impedance spectroscopy (EIS) under different applied bias potentials were performed in the dark (Fig. S28e).All Nyquist plots show similar series resistance R C of approximately 16 Ω.However, there is a difference in the charge transfer mechanism under different applied bias potentials.The near-vertical imaginary parts for OCP, -0.3 V and -0.4 V in the low frequency region indicate a capacitive-like charge storage mechanism, while the Nyquist plots for -0.5 V, -0.6 V and -0.7 V exhibit a clear 45° Warburgtype impedance in the mid-frequency range, indicating a diffusion-limited process.A Bodetype plot is further used to understand the energy storage mechanism of NbWO6 (Fig. S28f).
The real capacitance C' exhibits almost constant values in the low frequency range from OCP to -0.4 V, which is characteristic of a capacitive process.In contrast, C' increases significantly with increasing negative bias potential, indicating a diffusion-limited process. 14 As shown in Fig. S29, the photoanode shows a change towards blue color upon 365 nm UV illumination.The UV-vis transmittance/absorbance spectra (Fig. S30a-b) further confirm the change in blue color, while there is no obvious absorption intensity change at 575 nm under 1 sun illumination (Fig. S30d).
Figure S1.Schematic illustration of the liquid phase exfoliation process for 2D NbWO6 nanosheets.
7) was used to refine the recorded XRPD data.The peak profile was described by the fundamental parameter approach implemented into TOPAS (ref.8) and the background modeled by Chebychev polynomials of 6 th order.The pattern of LiNbWO6 was subjected to a fully weighted Rietveld refinement 9 using the dataset of α-LiNbWO6 published by Fourquet et al. 1 as starting model.In the crystal structure of α-LiNbWO6 there are three metal position situated all on 2c sites.In their crystal structure refinement Fourquet et al. used an ordered distribution of the cations within the cation substructure, i.e. lithium, tungsten and niobium occupying separate sites (Table

Figure S3 .
Figure S3.The powder XRD patterns of LiNbWO6 and protonated HNbWO6•xH2O powder including selected reflections indices and corresponding d-spacings for α-LiNbWO6 in the LiNbWO6 sample (red font color), selected d-spacings of diffraction lines in the LiNbWO6 sample (black font color) and peak positions of β-LiNbWO6 (green) and LiNb3O8 (magenta) impurities.

Figure
Figure S4.(a) Temperature dependent in situ PXRD patterns of the HNbWO6 sample (HNbWO6•xH2O type compound).White lines indicate phase transitions (from i to ii and subsequently to iii).(b) excerpt of the temperature dependent in situ PXRD patterns of the HNbWO6 sample including d-spacings of selected reflections and peak positions of β-LiNbWO6 (green) and LiNb3O8 (magenta) impurities.

Figure S5 .
Figure S5.SEM images of (a) LiNbWO6 and (b) HNbWO6•xH2O powders, which exhibit welldefined layered structures.(c) AFM image of exfoliated 2D NbWO6 nanosheets which were exfoliated from HNbWO6•xH2O with the presence of TBAOH as shown in Fig. S1.The possible counter ion is TBA + to compensate the negative charge of the nanosheets.

Figure
Figure S6.(a) The (hk0) selected-area electron diffraction (SAED) pattern from a multilayer array of NbWO6 nanosheets, indicating the presence of crystalline yet turbostratically disordered few-layer nanosheets, which is consistent with the facile exfoliation and restacking of the layered solid into single-layer nanosheets with rotational layer offsets.(b) TEM image of 2D NbWO6 restacked on a TEM grid.(c) The experimental SAED pattern of two layer NbWO6 sheets.The simulated patterns of (d) single layer and (e) two layers sheets with rotation from bulk LiNbWO6.

Figure
Figure S7.(a) The analytical results from ICP-OES on the weight percent of Nb and W in HNbWO6•xH2O and NbWO6 powder.(b) The calculated molar ratios between Nb and W are 1.00 and 1.06 for HNbWO6•xH2O and NbWO6, respectively.Values are means with standard deviation from two different measurements.

Figure
Figure S8.(a) XPS survey scan spectrum of NbWO6.High resolution XPS of (b) O 1s, (c) C 1s, (d) W 4f spectrum under different X-ray powers, and (e) W 4f spectra development with scanning time under the power of 140 W in NbWO 6 .The increase in the red line shows the increasing fraction of W(V).The XPS survey (Fig. S8a) indicates Nb, W and O as well as some C impurities.In Fig. S8b the O 1s spectrum is displayed, which shows a second component around 531.4 eV besides that attributed to the O-W and O-Nb bonds around 530.5 eV, which we relate to the OH components.XPS data of W 4f taken under different power and charge neutralizer settings

Figure
FigureS10.The UV-Vis-NIR spectra of 1 mg ml -1 NbWO6 suspension in the presence of 10 vol% MeOH before and after 365 nm UV irradiation for 30 min.

Figure S11 .
Figure S11.The Tauc plot of NbWO6 nanosheets obtained from UV-vis indicates an indirect band gap of 3.43 eV.

Figure
Figure S12.pH-dependent photocurrent in oxygen-free 1M LiCl in the presence of MeOH donor under 365 nm UV illumination.

Figure S13 .
Figure S13.Photographs of (a) the front and (b) back side of home-made reactor for threeelectrode measurements.The light impinges from the front side of the quartz window.

Figure S14 .
Figure S14.Photocurrent of NbWO6 photoanodes under (a) 1 sun and (b) 365 nm UV illumination in oxygen-free 1M LiCl in the presence of MeOH.A pronounced decrease of the maximum photocurrent over time is observed under 1 sun illumination, while very little decrease is seen under 365 nm UV illumination (see also Fig.S33).Chronoamperometry (CA) experiments were performed under an applied potential of -0.1 V.

Figure S15 .
Figure S15.Photocurrent of NbWO6 photoanodes in oxygen-free (a) 1M LiCl in the presence of 10 mM 4-MBA and (c) 1M LiCl electrolyte in pure H2O under 1 sun illumination.The first off/on/off cycles of (a, c) are shown in (b, d), respectively.CA experiments were performed under an applied potential of -0.1 V.

Figure S16 .
Figure S16.Photocurrents of NbWO6 electrodes in oxygen-free (a) 1M LiCl in the presence of 10 mM 4-MBA, and (c) 1M LiCl electrolyte in H2O under 365 nm UV illumination.The first off/on/off cycles of (a, c) are shown in (b, d), respectively.CA experiments were performed with an applied potential of -0.1 V.The photocurrent transient spikes may be caused by the discrepancy between the fast carrier generation, recombination, and slow surface reaction dynamics. 11They confirm the lower efficiency of 4-MBA and H2O donors compared to MeOH.

Figure S17 .
Figure S17.Photocurrents at different potentials in different donors under 1 sun illumination.A negligible current is observed in the dark under an applied potential of 0.2 V for both cases.Like in Fig.4e, a positive current flow is observed between 0.2 V to -0.3 V upon illumination, indicating the photo-generated electrons in the CB were extracted.Further increasing the potential to negative applied potentials decreases the current due to the decrease in the driving force for electron extraction.

Figure S18 .
Figure S18.Photocurrents at different potentials in different donors under 365 nm UV illumination.A negligible current is observed in the dark under an applied potential of 0.2 V for both cases.Upon illumination, a positive current flow is observed between 0.2 V to -0.4 V, indicating the photo-generated electrons in the CB were extracted.A further decrease in potential decreases the current due to the decrease in the driven force.The positive photocurrent generated in 1M LiCl electrolyte suggest that water can be used as donor, but less efficiently than 4-MBA and MeOH.

Figure
Figure S19.UV-Vis absorbance spectra of oxygen-free NbWO6 nanosheet suspension in the presence of (a) H2O, (b) 4-MBA, and (c) MeOH donors under light illumination for 10 min.The insert in (c) shows the magnified region between 570 nm to 580 nm.(d) Summary of the dependence of absorption intensity at a wavelength of 575 nm as a function of the different donors.

Figure S20 .
Figure S20.Color change of oxygen-free NbWO6 nanosheet suspensions in the presence of different donors before and after light illumination for 10 min.The suspension shows light milky color before illumination.After 1 sun illumination, no obvious color change was observed.However, the suspension exhibits significant color change under 365 nm UV illumination.The suspension with MeOH as donor shows dark blue color, while the one in H 2 O and 4-MBA donors also show blue color but less intense than for MeOH.The color change indicates that NbWO 6 is able to store photogenerated electrons in the conduction band under 365 nm UV illumination, while the storage ability under 1 sun illumination is significantly lower.

Figure S21 .
Figure S21.Photocurrents of NbWO6 electrodes in oxygen-rich 1M LiCl in the presence of MeOH (10 vol%) donor under (a) 1 sun and (c) 365 nm UV illumination.The first off/on/off cycles of (a, c) are shown in (b, d), respectively.CA experiments were performed with an applied potential of -0.1 V.

Figure S22 .
Figure S22.CV sweep in 1M LiCl and MeOH electrolyte under 365 nm UV illumination and continuous Ar and oxygen purging, respectively.The scan rate is 10 mV s -1 .

Figure
Figure S23.Dark CV sweep measurement shows reversible electron storage and release in continuous Ar purging, and the oxygen reduction reaction (ORR) in oxygen rich environment.The results highlight the importance of oxygen-free electrolyte.

Figure
Figure S24.OCP of NbWO6 electrodes in oxygen-rich 1 M LiCl in the presence of MeOH donor under 1 sun and 365nm UV illumination.The electrode was illuminated for 10 min and left in the dark for 1 h.The presence of oxygen would scavenge photo-generated electrons from NbWO6, which decreases the OCP of the electrodes during light charge.The OCP drops dramatically when light charging is stopped, indicating the electron scavenging side reaction by oxygen, which quickly consumes the photo-generated electrons.

Figure S25 .
Figure S25.Chronoamperometry photocurrent stability measurements under (a) 1 sun and (b) 365 nm UV illumination in oxygen-free 1M LiCl and MeOH electrolyte.The applied potential is -0.1 V.A decrease in photocurrent is observed under 1 sun illumination, while almost no decrease is observed under 365 nm UV illumination, indicating more stable charge storage under UV illumination.

Figure S26 .
Figure S26.Color change of the NbWO6 electrode during electric charge and discharge in oxygen-free 1M LiCl electrolyte.

Figure
Figure S27.(a) Operando UV-vis transmittance spectra and corresponding (b) absorbance spectra of a NbWO6 photoanode under different applied bias in oxygen-free 1M LiCl.The inserts in (a) and (b) show the magnified region between 570 nm and 580 nm.(c) The absorbance intensity at 575 nm under different applied potentials.The left plot shows the intensity trend during electric charging from OCP (0.18 V) to -0.7 V.The right plot shows the intensity trend during electric discharging from -0.65 V to 0 V.The arrows indicate the potential scanning direction.

Figure S28 .
Figure S28.Kinetic study of the electrochemical behavior of a NbWO6 electrode.CV scanning from 2 to 100 mV s -1 in the potential window between (a) 0.2 to -0.7 V and (c) 0.2 to -0.5 V.The log(i) vs log(v) plots for anodic and cathodic currents at (b) -0.65 V and (d) -0.45 V. bvalues are 0.83 and 0.71 for cathodic and anodic currents in (b), and 0.80 and 0.88 for cathodic and anodic currents in (d), respectively.(e) Nyquist plots under different applied bias potentials.The insert shows the magnified high-frequency region.(f) 3D Bode plot of the capacitance vs frequency and potential in (e).The dashed lines connect the C' vs potential at a specific frequency.The electrolyte used in (a, c, e) is oxygen-free 1 M LiCl.

Figure S29 .
Figure S29.Color change of a NbWO6 photoanode under (a) 1 sun and (b) 365 nm UV illumination for 10 min in oxygen-free 1M LiCl in the presence of MeOH (10 vol%).There is no obvious color change under 1 sun illumination, while the one under 365 nm UV illumination exhibits blue color.

Figure
Figure S30.UV-vis transmittance spectra and its corresponding absorbance spectra for a NbWO6 photoanode under (a, b) 365 nm UV illumination and (c, d) 1 sun illumination for 10 min.The electrolyte is oxygen-free 1M LiCl in the presence of 10 vol% MeOH.The inserts in (c) and (d) show the magnified region between 570 nm and 580 nm.

Figure S31 .
Figure S31.Electrical discharge profiles at different discharge current densities after illumination under (a) 1 sun and (b) 365 nm UV light for 10 min in 1M LiCl and MeOH electrolyte under continuous Ar purging.

Figure S32 .
Figure S32.The light charging profiles under different illumination times for (a) 1 sun and (c) 365 nm UV illumination and the corresponding electric discharge profiles.The discharge current densities are (b) 0.48 mA g -1 and (d) 4.8 mA g -1 .Inserts show the OCP under illumination at the beginning, i.e. up to one minute.The electrolyte is oxygen-free 1M LiCl with the presence of MeOH.

Figure S33 .
Figure S33.Reproducibility measurements of OCP stability of different batches of NbWO6 photoanodes.The photoanodes were illuminated under light for 10 min in 1M LiCl and MeOH electrolyte under continuous Ar purging.Then the OCP of photoanodes were left in the dark.The yellow lines in the left image indicate 1 sun illumination, the blue lines indicate 365 nm UV illumination.

Figure
Figure S34.(a) Nyquist plot of NbWO6 photoanode under 1 sun illumination.The fitted equivalent circuit models used for the NbWO6 photoanode (b) in the dark and (c) under 1 sun and 365 nm UV illumination in Fig. 5d.

Figure S35 .
Figure S35.Electric discharge profiles after (a) 1 sun illumination for 10 min and (b) 365 nm UV illumination for 5 min for different cycles.The discharge current densities are (a) 4.8 mA g -1 and (b) 24 mA g -1 .

Figure
Figure S36.(a) The hydrogen amount as a function of illumination time.(b) Long-term measurement for dark hydrogen generation.The yellow regions correspond the light illumination.(c) Histogram plot shows the turnover frequency (TOF) after 365 UV illumination for 30 min without delay time and with 1 h delay time.