Z-scheme photocatalyst systems employing Rh- and Ir-doped metal oxide materials for water splitting under visible light irradiation †

Various types of Z-scheme systems for water splitting under visible light irradiation were successfully developed by employing Rh- and Ir-doped metal oxide powdered materials with relatively narrow energy gaps (EG): BaTa 2 O 6 :Ir,La (EG: 1.9 – 2.0 eV), NaTaO 3 :Ir,La (EG: 2.1 – 2.3 eV), SrTiO 3 :Ir (EG: 1.6 – 1.8 eV), NaNbO 3 :Rh,Ba (EG: 2.5 eV) and TiO 2 :Rh,Sb (EG: 2.1 eV), with conventional SrTiO 3 :Rh (an H 2 -evolving photocatalyst) or BiVO 4 (an O 2 -evolving photocatalyst), and suitable electron mediators. The Z-scheme systems were classi ﬁ ed into three groups depending on the combination of H 2 - and O 2 evolving photocatalysts and electron mediator. The Z-scheme systems combining BaTa 2 O 6 :Ir,La with BiVO 4 , and NaTaO 3 :Ir,La with BiVO 4 were active when a [Co(bpy) 3 ] 3+/ 2+ redox couple was used rather than an Fe 3+/2+ one. The combination of SrTiO 3 :Ir with SrTiO 3 :Rh gave an activity when the [Co(bpy) 3 ] 3+/2+ and Fe 3+/2+ redox couple ionic mediators were used. The Z-scheme systems combining NaNbO 3 :Rh,Ba and TiO 2 :Rh,Sb with SrTiO 3 :Rh showed activities by using the [Co(bpy) 3 ] 3+/2+ and Fe 3+/2+ redox couples and also via interparticle electron transfer by just contact with/without reduced graphene oxide (RGO). These suitable combinations can be explained based on the impurity levels of doped Rh 3+ and Ir 3+ toward the redox potentials of the ionic mediators for the Z-scheme systems employing ionic mediators, and p-/n-type and onset potentials of the photocurrent in the photoelectrochemical properties of those photocatalyst materials for the Z-scheme systems working via interparticle electron transfer.


Introduction
Articial photosynthesis has attracted attention from the view point of solar energy conversion to storable chemical energy. Because solar water splitting is one of the representative reactions, photocatalytic water splitting has extensively been studied. [1][2][3][4][5][6][7] Powder-based metal oxide materials are attractive for photocatalytic water splitting because the cost will be low [8][9][10] and stability will be high compared with other materials such as chalcogenides. Although the efficiency of powder-based photocatalyst systems is behind that of systems of photovoltaic + electrolysis at the present stage, there are some advantages to powder-based photocatalyst systems.
It is crucial to demonstrate a solar water splitting system employing powderbased oxide photocatalysts. The present stage of research of articial synthesis is along this topic. A reactor system including a gas separation system of evolved hydrogen from oxygen has been studied in addition to the development of photocatalyst materials aimed towards the practical use of solar water splitting. 10 The separation system can be achieved by the use of a suitable separation membrane. Moreover, the safety issue of the co-evolved H 2 and O 2 has also been examined via the use of a suitable gas transportation tube. However, even if an excellent reactor system with a gas separation membrane system is established, which photocatalyst is employed for it is still a key issue as high efficient photocatalysts for real solar water splitting into H 2 and O 2 without any sacricial electron donors and acceptors have not yet been developed. It is essential to develop an efficient photocatalyst for demonstrating a solar water splitting system taking gas separation and safety issues into account.
High efficiency of a photocatalyst can be brought about by a high quantum yield and a response to light with a long wavelength. From this viewpoint, it is vital to develop photocatalysts that can utilize up to 600-700 nm of the solar spectrum. Although the use of noble metals might prevent the photocatalyst materials from practical use, the usage would be allowed if the amount is small and the materials can be recycled.
There are single particulate and Z-scheme systems in powder-based photocatalysts. 2,11 Z-scheme photocatalyst systems have the advantage that photocatalysts active for either H 2 or O 2 evolution can be employed. This means that a Z-scheme system can be constructed with various combinations of H 2 -and O 2 -evolving photocatalysts. 12-15 SrTiO 3 :Rh (an H 2 -evolving photocatalyst) and BiVO 4 (an O 2evolving photocatalyst) are representative metal oxide materials for the Z-scheme system. 1,2,16 It should be stressed that photocatalyst sheets prepared by a particle transfer method using SrTiO 3 :Rh,La and BiVO 4 :Mo powders demonstrate a quantum efficiency of 30% at 420 nm and a solar to hydrogen energy conversion efficiency of 1%. [17][18][19] This result suggests that the Z-scheme system is a promising photocatalyst system for practical solar water splitting. H 2 evolution separated from O 2 evolution is possible in the Z-scheme system if a suitable reactor is designed. 8 However, this photocatalyst sheet consisting of the SrTiO 3 :Rh,La and BiVO 4 :Mo powders responds up to 520 nm because the energy gap (EG) of SrTiO 3 :Rh,La and the band gap (BG) of BiVO 4 :Mo are 2.3 eV and 2.4 eV, respectively. So, it is a key issue to develop Z-scheme photocatalyst systems consisting of metal oxide photocatalysts with a response at longer wavelengths than 520 nm.
Transition metal doping is one strategy to make wide band gap photocatalysts responsive to visible light.  25 are active for sacricial O 2 evolution in the presence of electron acceptors such as Ag + . In these photocatalyst materials, the visible light responses are due to electronic transition from the impurity levels formed by the dopants to the conduction bands of the host materials. These energy and band gaps are close to, or narrower than, those of SrTiO 3 :Rh (EG: 2.3 eV) 20 and BiVO 4 (BG: 2.4 eV). 26,27 So, it is attractive to utilize these Rh-and Ir-doped metal oxide photocatalysts for the construction of Z-scheme systems.
There are several types of Z-scheme system employing different electron mediators such as Fe 3+/2+ and [Co(bpy) 3 ] 3+/2+ redox couples, 16,28,29 and reduced graphene oxide (RGO) 30-32 as shown in Fig. 1(a) and (b). Moreover, some Z-scheme systems work even without electron mediators, as shown in Fig. 1(c). 33,34 In this case, electron transfer proceeds through an interface contacted between the particles of the H 2 -and O 2 -evolving photocatalysts. So, it is important to examine the electron mediator in the Z-scheme system.
In the present paper, sacricial H 2 and O 2 evolutions over Rh-and Ir-doped metal oxide photocatalysts under visible light irradiation were examined rst to see the relationship between the photocatalytic properties and their band structures. Then, these H 2 -and O 2 -evolving photocatalysts were employed for various types of Z-scheme system for water splitting into H 2 and O 2 in stoichiometric amounts under visible light irradiation, without any sacricial reagents. The photocatalytic performances for the sacricial H 2 and O 2 evolutions and the Zschematic water splitting were discussed based on the band structures and   23 and BiVO 4 26,27 were prepared by a solid-state reaction, a borate-ux method, and a liquid-solid reaction according to previous reports. In addition to them, NaNbO 3 :Rh(x%),Ba(y%) (x, y) ¼ (1.2, 1.44) or (1.0, 2.0) was newly prepared by a solid-state reaction. The starting materials, Na 2 CO 3 (Kanto Chemical; 99.5 or 99.8%), Nb 2 O 5 (Kanto Chemical; 99.99% or Kojundo Chemical; 99.99%), Rh 2 O 3 (Wako Chemical; 98%), and BaCO 3 (Kanto Chemical; 99%), were mixed at a molar ratio of Na/Nb/Rh/Ba ¼ 1.05-1.05y : 1 À x : x : y. An excess of sodium was added in the starting materials to compensate for volatilization. The starting materials mixture was calcined at 1173 K for 1 h, and then 1423-1473 K for 10 h once or twice. The excess sodium was washed out with water aer the calcination. The obtained powders had nonspecic shapes with aggregations, judging from the SEM images (Jeol; JSM-6700F) (Fig. S1 †). The obtained samples were identied using X-ray diffraction (Rigaku; MiniFlex, Cu Ka). Diffuse reectance spectra were obtained by a UV-vis-NIR spectrometer (JASCO, V-570) equipped with an integrator sphere and were converted to absorbance measurements via the Kubelka-Munk method.

Preparation of an RGO-metal oxide composite
An RGO-incorporated O 2 -evolving photocatalyst was prepared by photocatalytic reduction of graphene oxide (GO) on the photocatalyst according to previous reports. [30][31][32] GO prepared by the Hummers' method 35 and the O 2 -evolving photocatalyst were dispersed in an aqueous methanol (Kanto Chemical; 99.8%) solution (50 vol%). The suspension was irradiated with visible light from a 300 W Xe lamp (PerkinElmer; CERMAX PE300BF) with a long pass lter (HOYA; L42) under a N 2 atmosphere with a pressure of 1 atm to obtain the RGO-photocatalyst composite. The methanol was carefully removed by washing with water. The RGO-photocatalyst composite was collected by ltration and was dried at room temperature in air.

Sacricial H 2 and O 2 evolutions (half reactions of water splitting)
H 2 and O 2 evolutions from aqueous solutions containing the sacricial reagents CH 3 OH (Kanto Chemical; 99.8%) and AgNO 3 (Kojima Chemical; 99.9% or Toyo Chemical; 99.9%) that were half reactions of water splitting were examined using a top-irradiation reaction cell with a Pyrex window and a 300 W Xe lamp (Perki-nElmer; CERMAX PE300BF). NaTaO 3 :Ir,La, BaTa 2 O 6 :Ir,La, NaNbO 3 :Rh,Ba, and TiO 2 :Rh,Sb were used as prepared, whereas SrTiO 3 :Ir without a cocatalyst was reduced under 1 atm of H 2 at 473 K for 2 h as a pretreatment for sacricial O 2 evolution. The photocatalyst powders (0.1-0.3 g) were suspended in aqueous solutions (120-150 mL) and irradiated with visible light. For the H 2 evolution, Pt (0.3 wt%) cocatalyst, working as an H 2 evolution site, was loaded on photocatalysts by photodeposition from an aqueous methanol solution containing H 2 PtCl 6 (Tanaka Kikinzoku; 37.55% as Pt). The wavelength of the irradiation light was controlled to visible light using long-pass lters (HOYA; L42 and Y44). The amounts of evolved H 2 and O 2 were determined using an online gas chromatograph (Shimadzu; GC-8A, MS-5A column, TCD, Ar carrier).

Z-schematic water splitting
Z-schematic water splitting was conducted using a gas-closed system with a topirradiation cell with a Pyrex window. H 2 -evolving photocatalyst and O 2 -evolving photocatalyst powders (0.05 or 0.1 g, respectively) were suspended in 120 mL of water. For the interparticle Z-scheme systems without an electron mediator and with an RGO solid-state electron mediator, water not containing any ionic mediators was used. For the Z-scheme system with ionic mediator, an aqueous solution containing [Co(bpy) 3 ]SO 4 or FeCl 3 as a mediator was used. The pH was adjusted with H 2 SO 4 in each of the Z-scheme systems with and without electron mediator, if necessary. Ru (0.7 wt%) cocatalyst, functioning as an H 2 evolution site, was loaded on SrTiO 3 :Rh (an H 2 -evolving photocatalyst) by photodeposition from an aqueous methanol solution containing RuCl 3 $nH 2 O (Tanaka Kikinzoku; 36% as Ru in RuCl 3 $nH 2 O). Pt (0.3-1 wt%) was loaded on the NaTaO 3 :Ir,La and BaTa 2 O 6 :Ir,La (H 2 -evolving photocatalysts), and SrTiO 3 :Ir (an O 2 -evolving photocatalyst) by an impregnation method. The photocatalyst powders and an aqueous H 2 PtCl 6 solution were placed in a porcelain crucible and dried on a hot plate. The H 2 PtCl 6 -impregnated NaTaO 3 :Ir,La and BaTa 2 O 6 :Ir,La powders were calcined at 673 K for 2 h in air, whereas the SrTiO 3 :Ir was not. The Pt-loaded NaTaO 3 :Ir,La and Pt-loaded BaTa 2 O 6 :Ir,La were subsequently reduced at 673 K for 1 h under 1 atm of H 2 as a pretreatment, while Pt-loaded SrTiO 3 :Ir was reduced at 573 K for 1 h. The light source and GC setup were the same as those for the sacricial H 2 and O 2 evolutions.

Photocatalytic activities for sacricial H 2 and O 2 evolutions over Rh-and Irdoped metal oxide materials and their band structures
Sacricial H 2 and O 2 evolutions of half reactions were carried out as test reactions of water splitting over Rh-and Ir-doped metal oxide photocatalysts using a sacricial electron donor and acceptor to see the ability for photocatalytic H 2 or O 2 evolution, as shown in Table 1, prior to conducting water splitting. The energy gaps were determined from the diffuse reectance spectra and wavelength dependence of the photocatalytic activities as shown in Fig. 2    photocatalysts for the construction of Z-scheme systems, while SrTiO 3 :Ir, NaNbO 3 :Rh,Ba and TiO 2 :Rh,Sb are expected to be employed as O 2 -evolving photocatalysts. Fig. 2 and 3 show the diffuse reectance spectra and wavelength dependence of the photocatalytic H 2 and O 2 evolutions of the Rh-and Ir-doped metal oxide photocatalysts in the presence of sacricial reagents. The wavelengths were controlled with long-pass lters. It is vital to see the wavelength dependency of the photocatalytic activity because it is not guaranteed that photocatalysts with visible light absorption bands always give activities under visible light irradiation. The onsets of the wavelength dependence agreed with those of the diffuse reection spectra. The onset wavelengths for the H 2 evolutions were 640 and 600 nm for BaTa 2 O 6 :Ir,La and NaTaO 3 :Ir,La, respectively. These onset wavelengths were longer than the 540 nm of SrTiO 3 :Rh (a conventional H 2 -evolving photocatalyst). The onset wavelengths for O 2 evolution were 600, 500 and 700 nm for TiO 2 :Rh,Sb, NaNbO 3 :Rh,Ba and SrTiO 3 :Ir, respectively. It is noteworthy that TiO 2 :Rh,Sb and SrTiO 3 :Ir responded at longer wavelengths than the BiVO 4 (BG: 2.4 eV) (a conventional O 2 -evolving photocatalyst). Fig. 4 shows the band structures of the Rh-and Ir-doped metal oxide photocatalysts. The impurity levels of Rh 3+ and Ir 3+ were estimated from the energy gaps determined by diffuse reection spectra supposing that the valence band consisting of O 2p located at +3 V vs. NHE at a pH of 0. 39 The absorption bands in the visible light region shown in Fig. 2 and 3 are due to electronic transition from the impurity levels consisting of Rh 3+ and Ir 3+ to the conduction bands of the host materials. The impurity levels formed with electron-lled orbitals of Ir 3+ were around 1.0-1.2 V for NaTaO 3 and BaTa 2 O 6 , while Ir 3+ in SrTiO 3 formed an impurity level around 1.4-1.6 V that was deeper than those in the cases of the tantalates. The energy levels formed with electron-lled orbitals of Rh 3+ located around 2.0-2.1 V that were similar to those of SrTiO 3 :Rh and SrTiO 3 :Rh,Sb. 20,21 The reason Rh 3+ forms a deeper impurity level than Ir 3+ is due to Ir 4+ being more stable than Rh 4+ in metal oxides. Therefore, electronic transition from the Ir 3+ impurity level to a conduction band is easier than that from Rh 3+ , resulting in the formation of the shallow impurity level by Ir 3+ .  Z-schematic systems for photocatalytic water splitting employing Rh-and Irdoped metal oxide materials The combination of SrTiO 3 :Rh and BiVO 4 photocatalysts can be a benchmark of a Z-scheme photocatalyst system. The Z-schematic water splitting proceeds using Fe 3+/2+ and [Co(bpy) 3 ] 3+/2+ redox couples as ionic mediators ( Fig. 1(a)) 16,28,29 and also via interparticle electron transfer between SrTiO 3 :Rh and BiVO 4 particles with and without RGO ( Fig. 1(b) and (c)). 30,33,34 In general, water splitting via Zschematic interparticle electron transfer by contact between the particles of the H 2 -and O 2 -evolving photocatalysts with and without RGO can be achieved ( Fig. 1(b) and (c)) when the H 2 -and O 2 -evolving photocatalysts satisfy the following two requirements; (i) H 2 -and O 2 -evolving photocatalysts possess p-and n-type semiconductor properties, respectively, (ii) there is a certain electrode potential at which the cathodic photocurrent of the p-type semiconductor overlaps with the anodic photocurrent of the n-type semiconductor, being similar to water splitting using a photoelectrochemical cell working with no applied external bias as shown in Fig. 1(d). 31,32 Moreover, H 2 -and O 2 -evolving photocatalysts must have contact with each other for interparticle electron transfer. In contrast to this, a Z-scheme system employing an ionic electron mediator could work regardless of the p-or n-type properties of the H 2 -or O 2 -evolving photocatalysts, if the photocatalysts have potentials for the reduction or oxidation of redox couple ionic mediators and the adsorption abilities for the redox couples. The SrTiO 3 :Rh and BiVO 4 photocatalysts satisfy these factors resulting in all of the Z-scheme systems showing activities for water splitting into H 2 and O 2 in stoichiometric amounts without any sacricial reagents ( Fig. 1(a)-(d)). Various types of Z-scheme photocatalyst systems employing Ir-and Rh-doped photocatalysts for water splitting into H 2 and O 2 under visible light irradiation are shown in Table 2. Here, the absolute evaluation of the performance of photocatalyst systems for water splitting is how much H 2 and O 2 is obtained under certain experimental conditions. In this sense, the activities of the different Zscheme systems in Table 2 are comparable with each other, but not with those of Table 1, because of the almost identical experimental conditions for the water splitting, even if the kinetics would be different among the systems.
For the construction of Z-scheme systems, BiVO 4 (an O 2 -evolving photocatalyst) was combined with BaTa 2 O 6 :Ir,La and NaTaO 3 :Ir,La (H 2 -evolving photocatalysts), while SrTiO 3 :Rh (an H 2 -evolving photocatalyst) was combined with TiO 2 :Rh,Sb, NaNbO 3 :Rh,Ba and SrTiO 3 :Ir (O 2 -evolving photocatalysts) as suggested by their H 2 and O 2 evolution abilities. In addition to the suspension system, the photoelectrochemical properties of photocatalyst powders immobilized on a conducting substrate as shown in Fig. 1(d) were examined using a Pt counter electrode without any sacricial reagents to see the cathodic or anodic photocurrent and the onset potentials of semiconductor properties, not as a water splitting device, as shown in Fig. 5, in order to consider the potential overlap for Zschematic water splitting via interparticle electron transfer with and without RGO. It has been reported that SrTiO 3     those photocurrents might not be due to H 2 and O 2 evolutions, but possibly due to redox reactions of the doped Ir species. In contrast, TiO 2 :Rh,Sb, 45 and NaNbO 3 :-Rh,Ba gave clear anodic photocurrents indicating an n-type semiconductor character. The onset potential of NaNbO 3 :Rh,Ba was more negative than that of TiO 2 :Rh,Sb, whereas the anodic photocurrent of NaNbO 3 :Rh,Ba was much smaller than that of TiO 2 :Rh,Sb. These anodic photocurrents overlapped with the cathodic photocurrent of SrTiO 3 :Rh at a certain electrode potential. The Z-scheme systems were classied into three groups depending on the combination of H 2 -and O 2 -evolving photocatalysts and an electron mediator;   Table 2 are classied into three groups, based on the band structure and photoelectrochemical properties. Fig. 6 Fig. 6(a), whereas they possess enough potential for oxidation of [Co(bpy) 3 ] 2+ at pH 4.2 as shown in Fig. 6(b). The activities from the interparticle electron transfer were very small or negligible because of poor photoresponse and poor overlaps of the potentials giving cathodic photocurrents of NaTaO 3 :Ir,La and BaTa 2 O 6 :Ir,La and an anodic photocurrent of BiVO 4 as shown in Fig. 5, and also may be due to poor contact between  3 ] 3+/2+ and Fe 3+/2+ redox couples but also via interparticle electron transfer with and without RGO. The activity of SrTiO 3 :Rh + TiO 2 :Rh,Sb was higher than that of SrTiO 3 :Rh + NaNbO 3 :Rh,Ba for all of the types of Z-scheme system. In the cases with the use of ionic electron mediators, it is probably due to the higher activity for O 2 evolution and the narrower energy gaps of TiO 2 :Rh,Sb than those of NaNbO 3 :Rh,Ba, as shown in Table 1. The reason why SrTiO 3 :Rh + TiO 2 :Rh,Sb showed a higher activity than SrTiO 3 :Rh + NaNbO 3 :Rh,Ba via interparticle electron transfer is that TiO 2 :Rh,Sb gave much larger anodic photocurrents than NaNbO 3 :Rh,Ba, as shown in Fig. 5.

Conclusions
We have successfully developed Z-scheme photocatalyst systems for water splitting under visible light irradiation employing Rh-and Ir-doped metal oxide photocatalysts with longer wavelength responses than conventional  3+ . This property means that SrTiO 3 :Ir, TiO 2 :Rh,Sb and NaNbO 3 :Rh,Ba could be employed as O 2 -evolving photocatalysts for the construction of Z-scheme systems employing ionic electron mediators. Moreover, photoelectrochemical measurements using photocatalyst powders immobilized on conducting substrate revealed that the n-type characters and relatively negative onset potentials of the TiO 2 :Rh,Sb and NaNbO 3 :Rh,Ba photoanodes enabled the Z-scheme systems to work by interparticle electron transfer. Although the efficiencies of the present Z-scheme systems are low at the present stage, these will be improved by interfacial controls from a kinetic point of view in basic research. These results and discussion will contribute to the design of a highly active photocatalyst system for water splitting into H 2 and O 2 , aiming for the demonstration of actual solar water splitting using a suitable reactor.

Conflicts of interest
There are no conicts to declare.