Electronic Structure and Photoelectrochemical Properties of an Ir

Aug 13, 2014 - ABSTRACT: The effect of iridium valence in Ir:SrTiO3 on the electronic structure and the photocatalytic activity in a water splitting r...
0 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/JPCC

Electronic Structure and Photoelectrochemical Properties of an IrDoped SrTiO3 Photocatalyst Seiji Kawasaki,*,† Ryota Takahashi,† Kazuto Akagi,‡ Jun Yoshinobu,† Fumio Komori,† Koji Horiba,§ Hiroshi Kumigashira,§ Katsuya Iwashina,¶ Akihiko Kudo,¶ and Mikk Lippmaa*,† †

Institute for Solid State Physics, University of Tokyo, Chiba 277-8581, Japan WPI-Advanced Institute for Materials Research (WPI-AIMR), Tohoku University, Sendai 980-8577, Japan § Photon Factory, High Energy Accelerator Research Organization, Tsukuba 305-0801, Japan ¶ Faculty of Science, Tokyo University of Science, Tokyo 162-8601, Japan ‡

S Supporting Information *

ABSTRACT: The effect of iridium valence in Ir:SrTiO3 on the electronic structure and the photocatalytic activity in a water splitting reaction was studied. Epitaxial thin film photoelectrodes were grown with controlled Ir valence and used to measure the electrochemical efficiency of Ir:SrTiO3. The positions of the in-gap Ir impurity levels were determined by optical and X-ray photoelectron spectroscopies. Comparison with ab initio calculations was used to assign the observed electronic states to either Ir4+ or Ir3+ dopants in SrTiO3. The measurements show that Ir3+:SrTiO3 forms a single midgap impurity state that is strongly localized, completely quenching the photoelectrochemical response. An anodic photoresponse was seen in Ir4+:SrTiO3 under visible-light illumination up to a wavelength of 600 nm (hν = 2.0 eV). Ir4+:SrTiO3 contains an in-gap state close to the top of the SrTiO3 valence band. The performance of Ir4+:SrTiO3 in electrochemical reactions is compared with cathodic Rh3+:SrTiO3, clearly illustrating the importance of strong dopant d-electron hybridization with the oxygen 2p valence band of SrTiO3 for improving the energy conversion efficiency in SrTiO3-based photocatalysts.

■

INTRODUCTION Photocatalytic and photoelectrochemical routes to water splitting are promising ways for converting solar energy to chemical energy.1−5 Photocatalytic materials have been widely studied since the discovery of the water splitting effect on TiO2 in the beginning of the 1970s.6 To create efficient photocatalysts for solar water splitting, the ability to harness visible light, suitable band alignment with water redox potentials, and long-term stability in water are the most important factors. From the viewpoint of water stability, oxides are generally more stable than silicon, sulfides, nitrides, or other compound semiconductors. However, in most oxide semiconductors, the valence band is formed primarily by the oxygen 2p states that are located at +2.94 V vs SHE7 while the O2/H2O redox potential is at +1.23 V vs SHE. Therefore, the number of oxide semiconductor candidates for solar water splitting is limited, and careful materials design is required to obtain an efficient photocatalyst. SrTiO3 is a good starting point for photocatalyst design; it has a band gap of 3.2 eV8 and shows photocatalytic water splitting activity under ultraviolet (UV) light.9−11 The band alignment of SrTiO3 with water is such that both hydrogen and oxygen evolution reactions are, in principle, possible, but some form of band engineering is required to reduce the band gap to utilize the visible part of the solar spectrum. For example, © 2014 American Chemical Society

photocatalytic activity under visible light has been reported for Rh-, Ir-, and Cr-doped SrTiO3 powders at wavelengths of up to about 500 nm.12−15 Rh-doped SrTiO3 (Rh:SrTiO3) in particular is known for its high activity in the hydrogen evolution reaction,12 and it has been used as a H2 evolution photocatalyst in a Z-scheme photocatalytic system16−20 where the flow of a cathodic photocurrent under visible light irradiation indicates the presence of a downward band bending at the interface with water. Rh:SrTiO3 therefore behaves as a ptype semiconductor in an electrochemical reaction, and the Rh doping affects the band gap primarily by lifting the valence band maximum of SrTiO3, without lowering the conduction band minimum.21−23 Ir is a 5d element in the same group with Rh (a 4d element) in the periodic table. Although the photocatalytic activity of Ir-doped SrTiO3 (Ir:SrTiO3) has been reported to be lower than that of Rh:SrTiO3, it has the advantage that it shows absorption over a wider spectral range of visible light and can be used in sacrificial H2 and O2 evolution reactions.12 Iridium doping has been used to extend the spectral response of other oxide host materials as well, such as NaMO3 (M: Nb and Ta), which shows a photocatalytic Received: June 24, 2014 Revised: August 12, 2014 Published: August 13, 2014 20222

dx.doi.org/10.1021/jp5062573 | J. Phys. Chem. C 2014, 118, 20222−20228

The Journal of Physical Chemistry C

Article

visible light response up to 650 nm.24 Despite the similarities between Rh and Ir doping, there are also important differences in the photocatalytic response. It is therefore instructive to study the differences in the electronic structure and photocatalytic response of Ir:SrTiO3 and Rh:SrTiO3. In the present work, we focus on Ir:SrTiO3 and investigate the electronic structure and the photoelectrochemical properties of epitaxial Ir:SrTiO3 thin films grown by pulsed laser deposition (PLD). We show that the Ir valence in SrTiO3 can be controlled between 4+ and 3+ by tuning thin film fabrication conditions, similar to the case of Rh:SrTiO3.22 The thin film samples with different Ir valence states were used to study the differences in the electronic structure and the impact on photoelectrochemical properties of Ir4+- and Ir3+-doped SrTiO3. The present study suggests that the Ir impurity level positions in the band gap of SrTiO3 and the hybridization between the Ir donor level and the valence band are crucial for obtaining a photocatalytic material with high photoelectrochemical activity.

The density of states (DOS) of nondoped and Ir4+/3+(3.70 atom %)-doped SrTiO3 were obtained by first-principles calculations using the VASP code.25,26 The HSE06 hybrid functional27 was used to reproduce accurately the band gap width and the positions of the in-gap impurity levels. A 3 × 3 × 3 SrTiO3 unit cell was used with one of the Ti4+ sites substituted by an Ir4+ ion, corresponding to a doping level of 3.70 atom %. The Ir3+:SrTiO3 system was modeled by introducing an additional electron with the same amount of positive background charge. The photoelectrochemical characteristics of the film samples were evaluated by cyclic voltammetry using a potentiostat (Hokuto Denko; HZ-5000) under visible light irradiation (300 W Xe lamp with an L42 optical filter). The measurements were performed in a three-electrode cell using Pt (Nilaco, 99.98%) as a counter electrode and a Ag/AgCl sat. KCl (TOA DKK; HS205C) as a reference electrode. The electrolyte solution was 0.1 M K2SO4 aq (pH = 5). The incident photon-to-current efficiency (IPCE) was evaluated with a Xe light source (Asahi Spectra, Torrance, CA; LAX-102), filtered by a set of band-pass filters centered at 400, 420, 440, 460, 480, 500, 520, 540, 560, 580, 600, 620, 640, 660, 680, 700, and 720 nm (Asahi Spectra; fwhm = 10 nm). Further information on sample fabrication, characterization methods, and first-principles calculation can be found in the Supporting Information (SI).

■

MATERIALS AND METHODS Ir:SrTiO3 powders were synthesized following the procedures used in earlier bulk photocatalyst work.12 The starting materials were SrCO3 (Wako Pure Chemical; >95.0%, heated in air at 300 °C for 2 h before use), TiO2 (Wako Pure Chemical; >99.0%, rutile), and IrO2 (Wako Pure Chemical; > 97.0%). The powders were mixed with a small amount of methanol in ratios corresponding to the compositions of the desired SrTi1−xIrxO3 (x = 0.01, 0.03, 0.05) targets and precalcined in air at 900 °C for 1 h, followed by calcining in air at 1100 °C for 10 h in an alumina crucible. For PLD ablation targets, the powders were shaped into pellets by powder molding in a press at 15 MPa, followed by calcining in air at 1200 °C for 24 h. The Ir:SrTiO3 films were grown by PLD on double-side-polished SrTiO3 (001) substrates (Shinkosha) for optical absorption measurements and on metallic Nb(0.05 wt %):SrTiO3 (001) substrates (Shinkosha) for X-ray photoelectron spectroscopy (XPS) and photoelectrochemical measurements. The substrates were annealed just before film growth in situ at 1100 °C for 10 min under a partial oxygen pressure of 10−5 Torr (background pressure ∼5 × 10−9 Torr) to obtain a chemically clean substrate surface with a regular step-and-terrace morphology. A KrF excimer laser was used to ablate the sintered Ir:SrTiO3 targets at a fluence of ∼1 J/cm2. The sample temperature was kept at 700 °C during film deposition. Epitaxial growth of Ir:SrTiO3 was confirmed by in situ reflection high-energy electron diffraction (RHEED) and ex situ X-ray diffraction (XRD). The film thickness was 20 nm for XPS and electrochemistry samples and 400 nm for samples used in optical absorption measurements. The Ir valence was determined by XPS (JPS-9010MC, JEOL) using a Mg Kα X-ray source (hν = 1253.6 eV). The electronic structure of Ir4+/3+:SrTiO3 was evaluated by optical absorption spectroscopy in ultraviolet, visible, and near-infrared (UV−vis−NIR) spectral ranges, first-principles calculations, and high-resolution XPS. The UV−vis−NIR absorption spectra were measured in transmittance mode for film samples or calculated from reflectance by the Kubelka−Munk method for powder samples. The occupied Ir 5d impurity level positions were determined by high-resolution XPS with a Scienta SES2002 electron analyzer using a synchrotron light source at Photon Factory BL-2A (High Energy Accelerator Research Organization, KEK).

■

RESULTS AND DISCUSSION Similar to Rh:SrTiO3, the valence state of Ir can be expected to affect the electrochemical efficiency of Ir:SrTiO3. The main factor affecting the valence of dopants with multiple possible oxidation states in SrTiO3 thin films is the density of oxygen vacancies in the crystal, which can be adjusted by changing the ambient oxygen pressure used during thin film growth. Fully oxidized Ir4+:SrTiO3 films were obtained at a relatively high oxygen pressure of 10−1 Torr, while low oxygen pressure growth at 10−6 Torr produced reduced Ir3+:SrTiO3−δ films. Figure 1 shows two typical RHEED intensity oscillation profiles

Figure 1. (a) RHEED intensity oscillations observed during film deposition and (b) AFM images of sample surfaces (2 × 2 μm2). Start and stop points of film deposition are marked in panel a. Each oscillation corresponds to the growth of a single unit cell (u.c.). The samples were prepared at 700 °C under an oxygen partial pressure of (upper) 10−1 or (lower) 10−6 Torr.

and the corresponding AFM images of samples deposited at 10−1 and 10−6 Torr of oxygen. Each oscillation of the specular RHEED intensity corresponds to the growth of 1 unit cell of SrTiO3 (a = 0.3905 nm). The growth was stopped after 50 oscillations, giving films with a thickness of 20 nm. Although the oxygen pressure differed by 5 orders of magnitude, as shown in Figure 1a, the RHEED intensity oscillations were clearly observed during film deposition in both cases, which 20223

dx.doi.org/10.1021/jp5062573 | J. Phys. Chem. C 2014, 118, 20222−20228

The Journal of Physical Chemistry C

Article

means that the films were grown in the layer-by-layer mode. Figure 1b shows that step-and-terrace morphologies were indeed observed by atomic force microscopy (AFM) for films grown at 10−1 and 10−6 Torr. Optical absorption reference spectra of nondoped SrTiO3 and Ir(5%):SrTiO3 powders are compared in Figure 2 with the

Figure 3. X-ray photoelectron spectra of Ir 4f core levels for Ir(5%):SrTiO3 films deposited at (a) 700 °C, 10−1 Torr, (b) 700 °C, 10−3 Torr, and (c) 700 °C, 10−6 Torr. Reference spectra of (d) Ir(5%):SrTiO3 and (e) IrO2 powders. The deconvoluted spectral components are shown for Ir4+ (orange) and Ir3+ (blue) core levels. Solid lines are the fitting curves. Spectra were measured with a Mg Kα X-ray source (hν = 1253.6 eV).

Figure 2. (a) Absorption spectra of nondoped SrTiO3 and Ir(5%):SrTiO3 powders (upper) and Ir(5%):SrTiO3 films (lower) together with representative sample images. The film color changed systematically from yellow for growth at 10−1 Torr to brown for growth at 10−6 Torr. Absorption spectra of powder samples were obtained from diffuse reflectance data. The proposed band structure for Ir4+/3+:SrTiO3 is shown in panel b.

4f7/2 at 62.7 eV and Ir3+ 4f7/2 at 61.4 eV. For the sample grown at 10−1 Torr, the Ir core level spectrum showed only a single component corresponding to Ir4+ (Figure 3a). The deconvoluted spectra of films grown at lower oxygen pressures always had two components, which means that the reduction of Ir4+ to Ir3+ state was not complete. Even at a growth pressure of 10−6 Torr, the Ir4+ component area was still 28% (Figure 3c). The core-level spectra show that the Ir4+/Ir3+ ratio can be systematically tuned by selecting a suitable oxygen pressure during film growth. The positions of the Ir 5d impurity levels in the band gap region of SrTiO3 were analyzed by high-resolution XPS and by first-principles calculations. The calculated total and partial density of states (PDOS) of Ir for Ir4+/3+(3.7 at%):SrTiO3 and nondoped SrTiO3 are shown in Figure 4. The vertical scale for

spectra of two Ir(5%):SrTiO3 films grown at 10−1 and 10−6 Torr. The film thickness was 400 nm for both samples. The different Ir valence states in the films can be easily distinguished visually, as the films deposited at 700 °C and 10−1 Torr were yellow, while those grown at 10−6 Torr were brown. The absorption edge at 380 nm corresponds to the band gap excitation of SrTiO3 (Eg = 3.2 eV).8 For fully oxidized films grown at 10−1 Torr, the increase in absorption at wavelengths shorter than 600 nm includes two excitations; from the valence band (VB) to an unoccupied Ir4+ acceptor level, marked B in Figure 2b, and from the occupied Ir4+ donor level (A) to the conduction band (CB). The broad peak at 800 nm (1.55 eV) corresponds to a d−d transition (state A to B). On the other hand, the film deposited at 10−6 Torr only showed a broad absorption slope between 400 and 1000 nm. This absorption contains excitations from the Ir3+ donor level (C) to the conduction band in addition to the absorption features observed for the film deposited at 10−1 Torr because the films grown at low pressures contained a mixture of Ir 4+ and 3+ valence states, as shown by XPS. The X-ray photoelectron spectra of Ir 4f core levels for Ir(5%):SrTiO3 films deposited at 700 °C at various oxygen pressures from 10−1 to 10−6 Torr are shown in Figure 3. Ir(5%):SrTiO3 and IrO2 powder spectra are also shown for reference. The binding energies were referenced to the Ti 2p3/2 core level peak at 459.1 eV.28 All spectra were measured with a Mg Kα X-ray source (hν = 1253.6 eV). Deconvoluted Ir4+ and Ir3+ components are shown with orange and blue hatching. The binding energy of the Ir 4f7/2 core level has been reported to be between 62.3 and 62.8 eV for Ir4+ and between 61.6 and 62.0 eV for Ir3+.29,30 The tail of the peak observed on the higher binding energy side for the IrO2 powder sample has been attributed to final-state screening effects observed for conductive rutile-structured oxides.31,32 For the film samples, the best fits were obtained with binding energy positions of Ir4+

Figure 4. Density of states calculated for nondoped SrTiO3 and Ir4+/3+(3.70%):SrTiO3. The total DOS (red line) and PDOS of Ir (blue line) are shown for each sample. The vertical scale for Ir4+/3+(3.70%):SrTiO3 is multiplied by 10. The Ir4+ donor and acceptor levels are marked with A and B, respectively. The Ir3+ donor level is marked with C. D and E are Ir 5d states hybridizing with the O 2p valence band and with the conduction band, respectively. EF marks the calculated Fermi level positions. 20224

dx.doi.org/10.1021/jp5062573 | J. Phys. Chem. C 2014, 118, 20222−20228

The Journal of Physical Chemistry C

Article

Ir4+/3+:SrTiO3 has been multiplied by 10. The Fermi level positions are marked by EF in the plots. The calculation shows that two in-gap features related to the Ir 5d orbitals exist for Ir4+ substituting at the Ti4+ site in SrTiO3, with peaks appearing ∼0.5 eV higher than the top of the VB and ∼0.8 eV below the bottom of the CB, labeled A and B, respectively. For Ir3+, only a single in-gap level was found in the midgap region, ∼1.2 eV higher than the top of VB, labeled C. The A, B, and C states in Figure 4 correspond to the A, B, and C peaks in Figure 2b. It should be noted that all of these impurity levels are derived from Ir 5d t2g states and that state B is induced by the shortrange Hartree−Fock exchange interaction in the HSE06 functional due to the existence of an unpaired electron at the Ir4+ (d5) site. The peaks marked by D and E are Ir 5d states overlapping with the lower-energy O 2p valence band states and with the conduction band, respectively. Simulations using the GGA functional, which does not consider the exchange interaction, produced only a single impurity level in the gap region for Ir- and Rh-doped SrTiO3.33 More detailed information is available in Figures S6−S8, Supporting Information. Figure 5 shows high-resolution X-ray photoelectron valence band spectra of Ir(5%):SrTiO3 films deposited at 700 °C and

valence band spectra taken at photon energies of 600, 1000, and 1400 eV can thus be used to separate the Ir contribution from the O 2p states. The difference spectra, calculated by subtracting the hν = 600 eV spectrum from the hν = 1000 eV and hν = 1400 eV spectra show two peaks at binding energies of ∼8 eV and ∼2 eV. The peak observed at the top of the valence band at ∼2 eV is assigned to the Ir 5d orbitals (states A and C in Figure 4), while the peak at the bottom of the valence band at ∼8 eV corresponds to O 2p-σ orbitals hybridizing with Ir 5d states (state D in Figure 4).32 The Ir 5d spectral components close to the top of the valence band, within the gap of the SrTiO3 host material, were deconvoluted to show the Ir4+ 5d, Ir3+ 5d, Ir metal 5d, and O 2p-π components, as shown in Figure 6. Ir3+ creates a shallower

Figure 6. Synchrotron X-ray photoelectron spectra of the valence band of Ir(5%):SrTiO3 films deposited at (a) 700 °C, 10−1 Torr and (b) 700 °C, 10−6 Torr. The photon energy was 1000 eV. Each raw spectrum (dotted line) is deconvoluted to Ir4+ 5d (red), Ir3+ 5d (blue), Ir metal 5d (orange), and O 2p-π (green) components. The fitting curve is shown in black.

donor level than Ir4+, as shown in Figure 2b. The binding energy positions of Ir4+ 5d and Ir3+ 5d peaks are at 2.24 and 1.74 eV, respectively. The Ir metal contribution was calculated by including a thermally broadened Fermi−Dirac function, because the valence band spectrum of Ir-metal is nearly flat in the 0 to 6 eV binding energy range.35 Exposure of the film surface to an intense synchrotron X-ray beam in ultrahigh vacuum resulted in moderate radiation damage, partially reducing Ir4+ to Ir3+, and Ir3+ to metallic Ir (Figure S4, Supporting Information). Thus, the spectrum of the film deposited at 10−1 Torr has both Ir4+ and Ir3+ components in Figure 6, while only Ir4+ was observed with a laboratory X-ray source (Figure 3). Furthermore, the film deposited at 10−6 Torr has three components, corresponding to Ir4+, Ir3+, and metallic Ir. It is interesting to compare the electronic structure of Ir:SrTiO3 with that of Rh:SrTiO3. The impurity level positions of Ir:SrTiO3 are similar to those of Rh4+/3+:SrTiO3, and the difference between Rh and Ir doping is only seen as a shift of Ir impurity levels by ∼0.5 eV higher than those of Rh in SrTiO3.23 Rh and Ir are in the same group in the periodic table, but being

Figure 5. Synchrotron X-ray photoelectron spectra of the valence band of Ir(5%):SrTiO3 films deposited at (a) 700 °C, 10−1 Torr (Ir4+:SrTiO3) and (b) 700 °C, 10−6 Torr (Ir4+/3+:SrTiO3). The photon energy was set at either 600, 1000, or 1400 eV. An increase of the Irrelated spectral components can be seen in the difference spectra measured at higher photon energies (lower panels); the intensity increase is visualized by subtracting the hν = 600 eV data from the spectra taken at hν = 1000 and 1400 eV.

either 10−1 or 10−6 Torr of oxygen, measured at photon energies hν = 600, 1000, and 1400 eV using a synchrotron light source. The binding energy was referenced to the Au Fermi edge. The main components observed in the valence band region are attributed to bonding (O 2p-σ) and nonbonding O 2p (O 2p-π) orbitals at ∼6 and ∼4 eV binding energies, respectively.23,34 The contribution of Ir states to the valence band spectra cannot be easily seen due to an overlap with the dominant oxygen 2p states. However, the photoexcitation cross section ratio of Ir and O increases with the incident photon energy (Figure S3, Supporting Information). A comparison of 20225

dx.doi.org/10.1021/jp5062573 | J. Phys. Chem. C 2014, 118, 20222−20228

The Journal of Physical Chemistry C

Article

a 5d element, the Ir impurity levels are systematically ∼0.5 eV higher than those of Rh 4d in the SrTiO3 parent compound. At the bottom of the VB, at relative energies of −2 to −4 eV in Figure 4 (state D), Ir 5d and O 2p PDOS components overlap for both Ir4+- and Ir3+-doped samples, which corresponds to the increase of spectral intensity in the 7 to 9 eV binding energy range in Figure 5. The Ir 5d in-gap impurity levels and the O 2p-σ (O−Ir) lower valence band states simulated by firstprinciples calculations are consistent with the experimental results obtained from UV−vis−NIR absorption spectra and XPS measurements. The photoelectrochemical behavior of Ir4+(5%):SrTiO3 and Ir3+(5%):SrTiO3 film samples in the water splitting reaction was characterized by cyclic voltammetry under visible light irradiation (Figure 7). Ir4+(5%):SrTiO3 and Ir3+(5%):SrTiO3

magnitude of the band bending at the water interface, the results of this work suggest that the impurity level position and the Fermi level depth are the most important factors. The hybridization between the impurity level and the valence band has been recognized as being crucial for high photoelectrochemical activity because it strongly affects the photocarrier lifetime and mobility.23,35 The reason why Ir3+:SrTiO3 did not show any photoresponse is that the occupied Ir3+ impurity level is close to the midgap position, ∼1.2 eV above the VB maximum. Therefore, the photogenerated holes in the Ir3+ states cannot be thermally excited to the oxygen 2p valence band and suffer from very low mobility, recombining with electrons before reaching the water interface. The calculated density of states for Ir4+:SrTiO3 (Figure 4) shows that the two Ir-related in-gap states, A and B, are positioned nearly symmetrically relative to the band edges. Although the calculated Fermi level position is close to mid gap, it may shift higher in the presence of surface oxygen defects. The anodic n-type electrochemical response suggests that the actual EF position in a working Ir4+:SrTiO3 photocatalyst is significantly higher than that for Rh4+:SrTiO3, which is a ptype cathodic photocatalyst (Figure S5, Supporting Information). It appears that in the n-type Ir4+:SrTiO3, the presence of the nominally unoccupied feature B in the electronic spectrum only reduces the photogenerated charge collection efficiency but does not render the photocatalyst completely inactive, as happens in the p-type Rh4+:SrTiO3. The IPCE of Ir4+(5%):SrTiO3 was measured at various wavelengths at an applied bias of 1.0 V vs Ag/AgCl, forming the action spectrum shown in Figure 8. The absorption coefficient

Figure 7. Cyclic voltammetry curves for 20 nm thick Ir4+(5%):SrTiO3 (upper) and Ir3+(5%):SrTiO3 (lower) film photoelectrodes measured in 0.1 M K2SO4 aq under visible light illumination (300 W Xe-lamp with L42 filter; λ > 420 nm). The photocurrent is plotted with solid lines, while the dark current is shown with a dashed line. The sweep rate was 20 mV/s.

films were fabricated at 700 °C and oxygen pressures of either 10−1 or 10−6 Torr. Despite the similarities in electronic structure with Rh:SrTiO3, which is known for its cathodic photoresponse under visible light irradiation,21,22 Ir4+:SrTiO3 showed an anodic photocurrent under visible light with an onset potential of −0.1 V vs Ag/AgCl and Ir3+:SrTiO3 showed no photoresponse at all. The photocatalytic activities of Rh4+and Rh3+-doped SrTiO3 are also known to be quite different; Rh4+:SrTiO3 is photocatalytically inert while Rh3+:SrTiO3 is highly active under light irradiation at wavelengths up to 540 nm.12,22,23 However, despite the superficial similarity of the electronic structures with Ir:SrTiO3, the roles of the 4+ and 3+ dopants are reversed. The difference between Rh- and Ir-doped SrTiO3 is related to the small but significant differences in the electronic structure of the two materials. Because the Ir impurity levels are ∼0.5 eV higher than the corresponding impurity levels of a Rh dopant in SrTiO3, the Fermi level of Ir:SrTiO3 is closer to the conduction band, which is why Ir:SrTiO3 delivers an anodic photocurrent and Ir4+:SrTiO3 thus behaves as an n-type semiconductor, while Rh:SrTiO3 has a deeper EF and shows p-type behavior.21−23 Even though Ir:SrTiO3 has an optical absorption coefficient larger than that of Rh:SrTiO3 over a wide spectral range, the saturated photocurrent density was still over 4 times lower than that observed for Rh:SrTiO3 photoelectrodes having ideally the same film thickness, crystal quality, surface morphology (reaction area), and doping level.22 Although many factors can affect the observed electrode current, such as differences in surface reaction sites or in the

Figure 8. APCE (blue) and IPCE (red) spectra of an Ir4+:SrTiO3 thin film (20 nm) electrode and the absorption coefficients of Ir4+:SrTiO3 (black) and SrTiO3 (gray). A bias of 1.0 V vs Ag/AgCl was applied to the sample in 0.1 M K2SO4 aq.

of Ir4+(5%):SrTiO3 is also plotted in Figure 8, showing that the action spectrum is proportional to the absorption coefficient and that the d−d transition (∼800 nm) does not contribute to the photoelectrochemical reaction. A measurable photoresponse was observed in the visible light region up to 600 nm. The absorbed photon-to-current efficiency (APCE) achieves 30% if the rather low absorption coefficient of