Photocatalytic Hydrogen Evolution Activity of Nitrogen/Fluorine-Codoped Rutile TiO2

The development of a photocatalyst capable of evolving H2 from water under visible light is important. Here, the photocatalytic activity of N/F-codoped rutile TiO2 (TiO2:N,F) for H2 evolution was examined with respect to metal cocatalyst loading and irradiation conditions. Among the metal species examined, Pd was the best-performing cocatalyst for TiO2:N,F under UV–vis irradiation (λ > 350 nm), producing H2 from an aqueous methanol solution. The H2 evolution activity was also dependent on the state of the loaded Pd species on the TiO2:N,F, which varied depending on the preparation conditions. Pd/TiO2:N,F prepared by an impregnation–H2 reduction method, showed the highest performance. However, the activity of the optimized Pd/TiO2:N,F toward H2 evolution from an aqueous methanol solution was negligibly small under visible-light irradiation (λ > 400 nm), although the use of an ethylenediaminetetraacetic acid disodium salt as an electron donor resulted in observable H2 evolution. Transient absorption spectroscopy revealed that although a relatively large population of reactive electrons was generated in the TiO2:N,F under 355 nm UV-pulse photoexcitation, the density of reactive electrons generated under 480 nm visible light was lower. This wavelength-dependent behavior in photogenerated charge carrier dynamics could explain the different photocatalytic activities of the TiO2:N,F catalysts under different irradiation conditions.


■ INTRODUCTION
−7 Although metal oxide photocatalysts with high quantum efficiencies have been developed, most of them can absorb only UV light because of their wide band gap. 2,8,9UV light accounts for only ∼5% of the solar spectrum; therefore, utilization of visible light is necessary to achieve greater solar energy conversion efficiency.
−16 The valence-band maxima of N-doped oxides are mainly composed of N 2p states, which are located just above the O 2p states, enabling the realization of band gaps narrower than those of the corresponding metal oxides.
Given the high Earth abundance of TiO 2 , development of a TiO 2 -based photocatalyst that exhibits a visible-light response might be ideal in terms of its potential for use in large-scale applications. 4,17Recently, we developed N/F-codoped rutile TiO 2 (TiO 2 :N,F), which functions as an O 2 -evolution photocatalyst in visible-light Z-scheme water-splitting systems in combination with H 2 -evolving Ru-loaded, Rh-doped SrTiO 3 (Ru/SrTiO 3 :Rh) and [Co(bpy) 3 ] 3+/2+ (bpy = 2,2′-bipyridyl) as a mediator. 18,19When TiO 2 :N,F nanoparticles were used instead of bulk TiO 2 :N,F, the water splitting rate for the Z-scheme tripled.Although oxide-based mixed-anion compounds that contain less-electronegative anions are inherently unstable toward photooxidation reactions, 1,8,12 such N-doped TiO 2 analogs have demonstrated high stability during water oxidation to O 2 . 18,20,21n addition to the development of such an O 2 -evolution photocatalyst, the development of a new H 2 -evolution photocatalyst for visible-light-driven Z-scheme water splitting systems is strongly demanded. 2,7In particular, the use of Earthabundant elements for photocatalysts is desirable from a sustainability viewpoint.−24 Although the rutile TiO 2 :N,F photocatalyst has the ability to thermodynamically reduce H 2 O (or H + ) into H 2 , 18 its water reduction activity has not been investigated in detail.
In the present study, we optimized the loading of cocatalysts on rutile TiO 2 :N,F.The loading of Pd, Pt, or Ir was found to enhance H 2 evolution under sacrificial conditions.Also, the catalyst loaded with Pd via the impregnation−H 2 reduction method [(Imp−H 2 )Pd/TiO 2 :N,F] showed greater activity than the catalyst loaded with Pd via the photodeposition (PD) method [(PD)Pd/TiO 2 :N,F], demonstrating a H 2 evolution rate of 55 μmol h −1 under irradiation with λ > 350 nm light.However, the photocatalytic activity of Pd-or Pt-loaded TiO 2 :N,F under visible light was very low.The different activities with respect to the photoexcitation conditions were investigated by transient absorption spectroscopy.General Characterization.The materials were characterized by powder X-ray diffraction (XRD; Rigaku MiniFlex 600; Cu Kα), UV−vis diffuse-reflectance spectroscopy (DRS; JASCO, V-670), and scanning electron microscopy (SEM; Hitachi High-Technologies, SU9000).X-ray absorption fine structure (XAFS) measurements were performed at beamline NW10A (PF-AR) of the High Energy Accelerator Research Organization, Tsukuba, Japan.The X-ray energy was varied by using a Si(111) double-crystal monochromator.The data were processed using Athena. 25reparation of TiO 2 :N,F.Rutile TiO 2 :N,F was synthesized according to the previously reported method by mixing rutile TiO 2 (JRC-TIO-16) and (NH 4 ) 2 TiF 6 in a molar ratio of 95:5 using an agate mortar and pestle. 18,19The mixture was loaded onto a Ni plate to prevent contamination from Al 2 O 3 and placed at the center of an alumina tube reactor. 22After the system was purged with dry NH 3 , the reactor was heated to 673 K (ramp: 10 K min −1 ) for 15 h under dry NH 3 flow (flow rate: 100 mL min −1 ).The specific surface area of the material thus obtained was 40 m 2 g −1 , as determined by an N 2adsorption measurement at 77 K.
Impregnation of Metal Species onto TiO 2 :N,F.TiO 2 :N,F was dispersed into water containing metal species (1 wt % metal relative to TiO 2 :N,F).As the metal precursors, Na 2 PdCl 4 , H 2 PtCl 6 , Na 2 IrCl 6 •6H 2 O, HAuCl 4 , Na 2 RhCl 6 • 6H 2 O, or RuCl 3 •nH 2 O was used.The powder suspension was heated on a steam bath while being continuously stirred with a glass rod until dry.The dried powder was lightly ground using an agate mortar and pestle and then heated at 473 K for 1 h under a H 2 flow (20 mL min −1 ) or in air.
Photocatalytic Reactions.Photocatalytic reactions were conducted in a Pyrex top-irradiation-type reaction vessel connected to a closed gas circulation system. 26In a typical H 2 evolution reaction, TiO 2 :N,F was dispersed in 140 mL of 10 vol % methanol aqueous solution or 10 mM EDTA•2Na solution, where methanol or EDTA acted as an electron donor.For in situ PD experiments, an appropriate metal precursor (1 wt % metal relative to TiO 2 :N,F) was dissolved in the solution.
The system was evacuated several times to remove air and then a small amount of Ar gas was introduced prior to irradiation.For the light source, a 300 W Xe lamp (Cermax, PE300BF) operating with an output current of 20 A or a solar simulator (Asahi Spectra HAL-320, irradiation area: 16 cm 2 ) was used.Light from the Xe lamp was passed through a water filter in combination with a CM-1 cold mirror.To change the irradiation wavelength, an L42 cutoff filter was used.Evolved gases were analyzed by an online gas chromatograph (Shimadzu, GC-2014s equipped with a thermal conductivity detector and MS-5A column; Ar carrier gas).The solution was kept at room temperature with a flow of cooling water during the reaction.The material after the reaction was collected by filtration and dried at 343 K for further characterization.
Transient Absorption Spectroscopy.Transient absorption spectroscopy measurements were conducted using a custom-made spectrometer. 27TiO 2 :N,F samples were fixed onto a CaF 2 plate at a density of 1.5 mg cm −2 and placed in an IR cell for measurement.The samples were photoexcited using 355 nm pulses from a Nd:YAG laser (Continuum Surelite I with Surelite OPO; duration, 6 ns; power, 2 mJ; repetition rate, 5−0.1 Hz), and transient absorption in the visible to mid-IR region was measured with the sample under a N 2 atmosphere.The same measurements were also conducted by using 480 nm pulses (power, 5 mJ).The time resolution of this spectrometer was limited to 1−2 μs by the bandwidth of the amplifier (Stanford Research Systems, SR560, 1 MHz).
■ RESULTS AND DISCUSSION PD of Metal Species onto Rutile TiO 2 :N,F.As the first step, the PD method, which does not require a heating procedure, 28 was used as a cocatalyst loading method.TiO 2 :N,F nanoparticles, which showed greater water oxidation activity than bulk TiO 2 :N,F, 19 was used unless otherwise stated.Figure 1 shows the time courses of H 2 evolution from a 10 vol % aqueous methanol solution containing various metal precursors under irradiation with λ > 350 nm light.Irrespective of the metal species, an induction period was observed at the beginning, suggesting in situ PD of the metal species onto the TiO 2 :N,F.The higher H 2 evolution rate for the metal precursor-containing reaction solution than for the solution without a metal (indicated as "none") suggests that the in situ loaded metal species functioned as cocatalysts for H 2 evolution.Among the investigated metals, Pd exhibited the highest activity, with the activity of Pt being slightly lower.Although Ir initially showed activity comparable to that of Pd and Pt, its H 2 evolution rate decreased after 1 h.These results suggest that the Ir state changed during the reaction and/or that the loading of excess Ir following the initial formation of active Ir species resulted in deactivation.
Comparison of Impregnation and PD Methods for H 2 Evolution Reaction under UV−Vis Irradiation.The Pd cocatalyst loading methods were optimized under sacrificial conditions by using methanol as an electron donor.Figure 2A shows the time courses of H 2 evolution from Pd/TiO 2 :N,F prepared by in situ PD [(PD)Pd/TiO 2 :N,F], impregnation followed by H 2 reduction [(Imp−H 2 )Pd/TiO 2 :N,F], and impregnation followed by heating in air [(Imp−air)Pd/ TiO 2 :N,F].The catalyst prepared by Imp−H 2 showed twofold greater activity than that prepared by PD.All of the Pd/ TiO 2 :N,F catalysts showed a ∼10 min induction period, suggesting that the catalyst state changed before the start of H 2 evolution.In the case of (PD)Pd/TiO 2 :N,F, the in situ formation of Pd species on the surface of TiO 2 :N,F might have caused such an induction period.However, both the (Imp− H 2 )Pd/TiO 2 :N,F and (Imp−air)Pd/TiO 2 :N,F catalysts showed a similar induction period.These results suggest that the state of the TiO 2 :N,F and/or loaded Pd also changed during the initial period.A change in the TiO 2 :N,F state during light irradiation is also supported by the observation of a ∼40 min induction period for H 2 evolution from the bare TiO 2 :N,F catalyst (Figure S1).In addition, an increase in absorption at ∼450 nm, which is caused by N doping, was detected in the DRS of the catalyst after the reaction (Figure S2A).Because the XRD pattern of the catalyst (Figure S2B) did not change after the reaction, the change in the DRS likely resulted from changes that occurred in limited regions, such as the surface.Notably, all of the Pd/TiO 2 :N,F catalysts showed a shorter induction period than the bare TiO 2 :N,F.If this induction period represents the time required to accumulate sufficient carriers in the semiconductor, then the shorter induction period for the Pd/TiO 2 :N,F catalysts might indicate a decrease in the overpotential needed to drive H 2 evolution as a consequence of the loaded cocatalysts.The (Imp−H 2 )Pd/ TiO 2 :N,F catalyst showed enhanced absorption at λ > 500 nm, whereas the (Imp−air)Pd/TiO 2 :N,F catalyst did not.The enhanced absorption in this region is likely the result of reduced Ti species generated during the H 2 reduction process. 29he effect of the cocatalyst loading is also apparent in the time course of N 2 evolution (Figure 2B), which results from self-oxidation of TiO 2 :N,F. 22Compared with a bare sample, the metal-photodeposited TiO 2 :N,F (PD) catalyst showed enhanced N 2 evolution.This result suggests that the loaded Pd species enhanced H 2 evolution and that the increase in the number of holes left in the semiconductor drove the selfoxidation reaction.However, the catalysts prepared by Imp− H 2 and Imp−air showed suppressed N 2 evolution, suggesting that the heat treatment resulted in improved stability of the doped nitrides or the removal of weakly bonded nitrogen species from the surface region of the catalysts.
To characterize the loaded Pd species, we conducted Pd Kedge XAFS measurements for the catalysts prepared by Imp− H 2 and PD.The X-ray absorption near edge structure  (XANES) spectra (Figure 3A) suggest that the Pd species deposited on TiO 2 :N,F was a mixture of metallic Pd and Na 2 PdCl 4 .Oscillations in the extended X-ray absorption fine structure (EXAFS) region (Figure 3B) support this finding.The shapes of the XANES and EXAFS oscillations indicate that the catalyst prepared by PD contains more Pd metal than that prepared by Imp−H 2 .Fourier transformed EXAFS spectra are shown in Figure 3C.The disappearance of the peak at 1.9 Å in the spectrum of the catalyst prepared by PD indicates the (partial) decomposition of Na 2 PdCl 4 during the PD procedure.
The scanning electron microscopy (SEM) images in Figure 4 reveal that ∼5 nm Pd nanoparticles were deposited onto the surface of TiO 2 :N,F in the photodeposited catalyst.By contrast, few deposits were observed for the (Imp−H 2 )Pd/ TiO 2 :N,F catalyst, although careful observation in a selected area revealed a few ∼5 nm Pd species (Figure 4 inset).Thus, not only the valence state of Pd but also the morphological character were found to differ among the prepared catalysts.
These results indicate that the enhanced activity of the (Imp−H 2 )Pd/TiO 2 :N,F catalyst compared with that of the (PD)Pd/TiO 2 :N,F or (Imp−air)Pd/TiO 2 :N,F catalyst might be caused by differences in TiO 2 :N,F itself and in the loaded Pd species.The DRS results suggest that the (Imp−H 2 )Pd/ TiO 2 :N,F had a greater donor (electron) concentration (Figure S2A).−32 Thus, partial reduction of Ti in TiO 2 :N,F might be one of the reasons for the enhanced activity.However, the state of the loaded Pd was revealed to differ depending on the loading method.Although both the (Imp−H 2 )Pd/TiO 2 :N,F and (PD)Pd/TiO 2 :N,F samples appear to contain both metal and chloride species, the (Imp−H 2 )Pd/TiO 2 :N,F sample has a higher chloride fraction than the (PD)Pd/TiO 2 :N,F sample.This result suggests that the remaining chloride species might be responsible for the high activity of the (Imp−H 2 )Pd/ TiO 2 :N,F sample.
The chloride species can enhance activity through two possible mechanisms: (1) the chloride itself can act as an electron donor or (2) the chloride can function as a precursor for the "real" catalyst. 33,34Possibility (1) is less likely than possibility (2) because no H 2 evolution was observed in the absence of methanol as an electron donor.SEM observations showed that highly dispersed Pd species, enabled by the Imp− H 2 procedure, on TiO 2 :N,F likely contribute, at least in part, to the higher photocatalytic activity of the (Imp−H 2 )Pd/ TiO 2 :N,F catalyst because smaller nanoparticle catalysts are generally beneficial to photocatalytic H 2 evolution. 5,8,35hotocatalytic H 2 Evolution Activity of M/TiO 2 :N,F under Visible-Light Irradiation.In the previous section, Pd was shown to be the most effective H 2 evolution cocatalyst for TiO 2 :N,F in aqueous methanol under UV−vis irradiation.We subsequently conducted H 2 evolution reactions using Pd/ TiO 2 :N,F under visible-light irradiation (λ > 400 nm).However, no H 2 evolution was observed from aqueous methanol (Table 1).The activity was also negligible when Pt/TiO 2 :N,F was used.
Previous studies involving oxynitride photocatalysts sometimes suggested that methanol was not the optimal electron donor for driving H 2 evolution. 36We therefore investigated other electron donors.H 2 evolution was observed when the reaction was conducted in an aqueous solution containing dissolved ethylenediaminetetraacetic acid disodium salt (EDTA•2Na), which is a stronger electron donor than methanol.As shown in Figure 5, H 2 evolution occurred almost linearly at a maximum rate of 1.7 μmol h −1 .The short induction period observed at the beginning might be a result of photoreduction of the palladium chloride precursors remaining on the powder after H 2 reduction (Figure 3).When the same reaction was conducted under AM 1.5G simulated sunlight, the maximum rate of H 2 evolution increased to 7.0 μmol h −1 (Figure S3).This increase in activity under weak simulated sunlight compared with that under a high-intensity Xe lamp suggests the importance of the UV component of irradiated light for efficient H 2 evolution.
Photogenerated Charge Carrier Dynamics of TiO 2 :N,F.To investigate the reason for the dependence of the H 2 evolution activity of TiO 2 :N,F on the irradiation  wavelength, we investigated the dependence of the carrier dynamics on the excitation wavelength by using transient absorption spectroscopy.Figure 6 shows transient absorption spectra of bare TiO 2 :N,F excited at different laser-pulse wavelengths.The signal at ∼19,500 cm −1 is usually ascribed to trapped photogenerated holes. 19,37Signals in the range 17,000−2000 cm −1 are usually ascribed to deeply trapped electrons, and signals at <2000 cm −1 , which increase in intensity with decreasing wavenumber, are ascribed to free or shallowly trapped electrons. 37,38The negative peak at 3200 cm −1 is due to the noise caused by adsorbed water.Notably, the spectral shape in Figure 6A differs from that recorded under 480 nm irradiation (Figure 6B).This difference indicates that the carrier dynamics in TiO 2 :N,F are excitation-wavelength-dependent, which might be the cause of the difference in the photocatalytic activities of the TiO 2 :N,F catalysts under different irradiation conditions.Notably, under 480 nm excitation, the absorption for deeply trapped electrons shifted to higher wavenumbers compared to that under 355 nm excitation.Although the reason for this difference is not yet clear, the number of free or shallowly trapped electrons appears to be associated with the dependence of the H 2 evolution activity on the excitation wavelength.That is, a high H 2 evolution rate under UV irradiation might be associated with a higher concentration of free or shallowly trapped electrons, whereas the absence of H 2 evolution under visible-light irradiation might be associated with a lower concentration of free or shallowly trapped electrons as a result of the fast deep trapping of electrons.

■ CONCLUSIONS
This work investigated the photocatalytic activity of rutile-type TiO 2 :N,F for H 2 evolution with respect to the metal cocatalyst loading and irradiation wavelength.Among the samples examined, (Imp−H 2 )Pd/TiO 2 :N,F demonstrated the highest rate of H 2 evolution from aqueous methanol under UV-and visible-light irradiation (λ > 350 nm).Under visible-light irradiation (λ > 420 nm), however, the H 2 evolution rate was substantially lower.This result was attributed to the lower population of reactive electrons generated under visible light in    Transmittance and reflectance were measured below and above 6000 cm −1 , respectively.

Figure 2 .
Figure 2. Time courses of (A) H 2 and (B) N 2 evolution from Pd(1 wt %)/TiO 2 :N,F catalysts prepared using different loading methods when the catalysts were dispersed in 10 vol % aqueous methanol and irradiated with UV−Vis (λ > 350 nm) light.Reaction conditions: catalyst, 50 mg; 10 vol % aqueous methanol solution, 140 mL; light source, 300 W Xe lamp fitted with a CM-1 mirror.For in situ photodeposition, Na 2 PdCl 4 (1 wt %) was added to the reaction solution.

Figure 4 .
Figure 4. SEM images of Pd-loaded TiO 2 :N,F samples and a bare TiO 2 :N,F sample.The inset in the left panel is a magnified image of a selected area.
TiO 2 :N,F as a result of the fast deep trapping of excited electrons.■ ASSOCIATED CONTENT * sı Supporting Information The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.3c06492.Time course of H 2 evolution using unmodified TiO 2 :N,F under UV−vis (λ > 350 nm) irradiation; UV−vis DRS and XRD patterns for Pd/TiO 2 :N,F before and after the reaction; and time course of H 2 evolution from Pd (Imp-H 2 )/TiO 2 :N,F under simulated sunlight in the presence of EDTA•2Na (PDF)■ AUTHORINFORMATION

Figure 6 .
Figure 6.Transient absorption spectra of rutile TiO 2 :N,F: (A) 355 and (B) 480 nm excitation.The spectra were recorded under a N 2 atmosphere.Transmittance and reflectance were measured below and above 6000 cm −1 , respectively.

Table 1 .
H 2 Evolution Rates for Cocatalyst-Loaded TiO 2 :N,F (Loading Amount 1 wt % via the Imp−H 2 Method) from Electron-Donor Solutions a