Improved overall water splitting with barium tantalate mixed oxide composites

The combination of effective charge carrier separation and improved electron transfer in highly crystalline barium tantalate composites modified with Rh–Cr2O3 core–shell co-catalyst systems induces enhanced activity for overall water splitting (OWS) with stoichiometric amounts of H2 and O2 (2 : 1). A sol–gel route employing complexing reagents was investigated to prepare selectively defined mixed oxide materials with improved surface areas and smaller particle sizes compared to the conventional solid state reaction (SSR). The catalytic activities of the materials are investigated in photocatalytic test reactions for hydrogen production and overall water splitting. The formation of Rh–Cr2O3 core–shell co-catalyst systems for water splitting is evidenced by transmission electron microscopy (TEM) and X-ray Photoelectron Spectroscopy (XPS). Moreover, we developed new and highly active barium tantalate composites for hydrogen generation from aqueous methanol solutions under UV-light, which show the highest hydrogen evolution rate for a three-component composite consisting of Ba5Ta4O15/Ba3Ta5O15/ BaTa2O6. Hydrogen rates of more than 6 mmol h 1 can be achieved without any co-catalyst. Using Rh–Cr2O3 core–shell co-catalysts on these three-component composites simultaneous generation of H2 and O2 from pure water splitting reaches rates up to 70% higher than for the pure Ba5Ta4O15.


Introduction
Photocatalysis with semiconductor materials exhibits a very important area of research. In the last decades, especially mixed-oxide photocatalysts with layered perovskite structure were discovered being very active materials for clean hydrogen production and direct water splitting. [1][2][3][4][5] In particular (111)layered materials with the composition A 5 M 4 O 15 (A ¼ Sr, Ba; M ¼ Ta, Nb) have made remarkable progress under ultra-violet (UV) light. [6][7][8][9] On the other hand enhanced photocatalytic hydrogen generation has been obtained with composites, 10 which include different compounds or phases of classical semiconductor systems, like CdS-TiO 2 (ref. 3) or a-/b-Ga 2 O 3 . 11 The increase in photocatalytic activity is based on an improved charge carrier separation in the composite photocatalyst and reduced electron-hole recombination since the charge carriers are spatially separated on different crystal compounds or phases.
However, there are several more parameters which affect the photocatalytic activity and efficiency. The electronic structure, morphology, high crystallinity, but concurrently small particle sizes are important challenges. 12 The solid state reaction (SSR) presents still an easy and conventional way to synthesize photocatalyst materials like Ba 5 Ta 4 O 15 . 13 However, this method applies high reaction temperatures to the oxide precursors, the particle sizes of the resulting materials are usually in the range of several micrometers. Therefore, the surface areas of the SSR products are very low in the range of only a few m 2 g À1 . To overcome this problem we applied a sol-gel synthesis using the complexing reagents EDTA (ethylene diamine tetraacetic acid) and citric acid (citrate route) 14 for the synthesis of barium tantalate. 15 Also several other complex semiconductor materials like CsTaWO 6 have been successfully prepared by this method recently. 16 One of the key challenges in photocatalysis is overall water splitting (OWS).
This reaction with semiconductor photocatalysts has been studied extensively as a potential method to supply clean and renewable hydrogen without the need of sacricial electron donors. [17][18][19][20][21][22] Thus, an effective formation of a Rh-Cr 2 O 3 coreshell co-catalyst structure has been described by Domen and his co-workers for efficient OWS, using it onto GaN-ZnO solid solution catalysts to optimize photocatalytic OWS in visible light. 23 This core-shell structure contains an amorphous Cr 2 O 3 shell which is reported to be impermeable for O 2 molecules but very well permeable for protons and evolved H 2 , thus inhibiting the unwanted back reaction of H 2 and O 2 during light irradiation. 23 In another report the use of CuO x species in combination with Cr 2 O 3 as co-catalyst for enabling OWS has been demonstrated on Ga 2 O 3 recently, although no core-shell structure was formed, 24 raising questions about the inuence of Cr 2 O 3 . In the present manuscript, we report for the rst time of a Rh-Cr 2 O 3 core-shell co-catalyst system on barium tantalate Ba 5 Ta 4 O 15 to achieve overall water splitting. More importantly, we describe the formation of multicomponent heterojunction photocatalysts consisting of Ba 5 Ta 4 O 15 , Ba 3 Ta 5 O 15 and BaTa 2 O 6 , leading to highly efficient charge carrier separation. Optimum ratios of those three components result in very high H 2 evolution rates from methanol containing solutions even without cocatalysts, and the optimized OWS on those photocatalyst composites using very small amounts of Rh-Cr 2 O 3 core-shell co-catalysts. These investigations lead to optimized materials which are excellent starting materials for future doping and OWS in visible light.

Synthesis of materials
For the synthesis of Ba 5 Ta 4 O 15 by the citrate route, 7 g of EDTA (99%, alfa-aesar) and 7.25 g of citric acid (98%, alfa aesar) were dissolved in 570 mL water, the pH was adjusted to 8.3 using conc. ammonia water (33%, J. T. Baker). Aer dissolution, the pH was set to 5 using conc. nitric acid (J. T. Baker), and 15 mL of hydrogen peroxide were added to stabilize the highest oxidation state of Ta. 10.224 mmol of Ta(OEt) 5 (99+%, alfa-aesar) was dissolved in abs. ethanol, and added in small portions to the solution while heating to 90 C. Finally, a solution of 12.78 mmol Ba(NO 3 ) 2 (99.95%, alfa-aesar) was added in small portions to avoid any precipitation and to get the nal clear solution. For preparation of barium tantalate composites the barium precursor amount was stepwise decreased from 12.78 mmol to 5.6 mmol Ba(NO 3 ) 2 . The samples were named according to the amount of Ba(NO 3 ) 2 added (e.g. BaTa 12.78 equates to 12.78 mmol Ba precursor); the Ta(OEt) 5 amount was always kept constant. In all the syntheses aer solvent evaporation a black powder precursor remained which was calcined at 600 C for 4 hours to evaporate any carbon leaving a white powder as raw product. A second calcination step was performed in a furnace for 10 hours at 1000 C to induce high crystallinity of the nal product.

Preparation of Rh-Cr 2 O 3 core-shell co-catalyst
Rh nanoparticles were deposited onto the photocatalyst materials via reductive photodeposition with a specic amount of Na 3 RhCl 6 (99.999%, Aldrich) precursor solution using the setup described below. Further photodeposition of Cr 2 O 3 was performed from sequential amounts of K 2 CrO 4 (99.9%, Wako Pure Chemicals) precursor solution by a stepwise in situ photodeposition 25 to nd the optimum loading of Cr 2 O 3 .
Characterization X-ray powder diffraction measurements were carried out to characterize the phase compositions of the precursors and the calcined samples. Diffraction patterns were recorded in reection geometry with an Empyrean Theta-Theta diffractometer (Panalytical, Almelo) equipped with a copper tube, 0.25 divergent slit, 0.5 antiscatter slit (incident beam), 7.5 mm high antiscatter slit (diffracted beam), incident and diffracted beam 0.04 rad soller slits, and a position sensitive PIXcel-1d detector. The Cu K-beta emission line is suppressed by a Ni lter. For qualitative phase analysis the specimens were scanned in the 10-75 2q range with a step width of 0.0131 and 250 s collection time at an ambient temperature of 300 K. The ICDD powder diffraction le (PDF2) in conjunction with the HighScore Plus soware (Panalytical, Almelo) was used for qualitative phase analysis. Transmission electron microscopy (TEM) images of selected samples were measured with an H-7100 electron microscope (100 kV) from Hitachi. BET surface areas were obtained using Kr gas with a Quantachrome Autosorb-1-MP. UV-Vis diffuse reectance spectra were measured using a Perkin Elmer Lambda 650 UV-Vis spectrometer equipped with a Praying-Mantis mirror construction. The obtained spectra were converted by the Kubelka-Munk function F(R) into absorption spectra, using MgO nanopowder as a white standard. Optical band gaps (E g ) were obtained via Tauc-plots, a method invented by the physicist Jan Tauc, 26 using the calculation a ¼ A(hn À E g ) n /hn, where a is the absorption coefficient, A is a constant, hn is the energy of light, and n ¼ 2 stands for materials with indirect transition, respectively. [27][28][29][30][31] Assuming the absorption coefficient a being proportional to the Kubelka-Munk function F(R), the E g can be obtained from the plot of [F(R)hn] 1/n versus hn, by extrapolation of the linear part near the onset of the absorption edge to intersect the energy axis.

Photocatalytic reactions
Photocatalytic hydrogen generation was measured in a homemade air free closed gas system using an inner irradiation-type quartz reactor with water cooling jacket. As a light source, a 700 W Hg mid-pressure immersion lamp (Peschl UV-Consulting, set to a power of 500 W) was used for irradiation and cooled with a double-walled quartz mantle using a thermostat (LAUDA). Gas evolution was detected online using a multichannel analyzer (Emerson) equipped with a detector for the determination of the concentration of hydrogen (thermal conductivity detector), oxygen (paramagnetism) and carbon dioxide (IR). Argon was used as carrier gas; the continuous gas ow was controlled by a Brockhorst mass ow controller. The gas ow was set to 50 NmL min À1 (normal ow throughput: gas ow at normal temperature and pressure per minute). All reactions were performed at 13 C.
For reactions in presence of methanol as electron donor, 500 mg of photocatalyst were suspended in 550 mL water and 50 mL methanol (electron donor). To get a homogeneous suspension a treatment in the ultrasonication bath for 10 minutes at 30 C was performed. Before any photocatalytic reaction was initiated, the whole system including the photocatalyst was ushed with argon at 100 NmL min À1 for 30 minutes to remove any trace of air.
Before adding the Na 3 RhCl 6 or K 2 CrO 4 solution for photodeposition, the light irradiation was stopped, and the precursor solutions were added with a syringe through a rubber seal without opening the reactor. Traces of air were subsequently removed by ushing the reactor again with Argon before starting light irradiation. Upon light irradiation, metallic Rh is deposited onto the photocatalyst surface sites preferentially accessible for electrons, while CO 2 is formed from methanol as also detected with our gas analyzer. Cr 2 O 3 is only deposited onto Rh nanoparticles. Gas evolution measurements were continued to investigate the hydrogen generation properties with Rh-Cr 2 O 3 providing active sites for hydrogen evolution while using the residual methanol as sacricial agent. Measurements were repeated several times, the error is below 30 mmol h À1 . For OWS, the samples aer sequential photodeposition were ltered, washed extensively with distilled water and dried overnight, before using them in pure water for stoichiometric H 2 and O 2 evolution.  32 The XRD pattern of the barium tantalate material synthesized via citrate route according to literature 15 shows the pure Ba 5 Ta 4 O 15 crystal structure, in very good agreement with the standard diffraction pattern of Ba 5 Ta 4 O 15 (JCPDS 18-0193/ JCPDS 72-0631); all reections can be indexed accordingly. The BET surface of the pure phase SSR product exhibits more than three times lower surface area compared to the citrate product (1.2 m 2 g À1 vs. 4.0 m 2 g À1 ). 15 When the total amount of barium nitrate precursor in the synthesis was slightly decreased to 11.73 mmol, a small amount of a second material can be observed which was identied as Ba 3 Ta 5 O 15 (indicated by stars in Fig. 2) in agreement with its standard diffraction pattern (JCPDS 83-0713). Ba 3 Ta 5 O 15 exhibits a tetragonal structure with space group P4/mbm. 15,33 Aer a stepwise decrease of the Ba 2+ precursor concentration down to 8.45 mmol a third compound is observed which can be identied as BaTa 2 O 6 (reections indicated by rhombs). It is well known that BaTa 2 O 6 crystallizes in three different temperature dependent modications: below 1150 C a modi-cation isostructural with orthorhombic BaNb 2 O 6 is formed, between 1150 C and 1300 C a tetragonal tungsten bronze type structure is apparent, and above this temperature the hexagonal structure occurs. 34 Due to the calcination temperature of only 1000 C and the analysis by XRD technique the third component can be indexed as the orthorhombic form of BaTa 2 O 6 (JCPDS 20-0146). All determined reections are sharp which indicates a good crystallinity of the materials aer citrate route synthesis and calcination. A further decrease in the barium precursor concentration to less than 6 mmol leads to additional reections in the XRD (BaTa 5.6), which could not be identied yet.

Results and discussion
The different barium tantalate multicomponent composite materials were further investigated by UV-Vis absorption spectroscopy. Fig. 2 shows the Tauc plots of the pure phase material     O 15 in pure yet, the rough estimation from the current absorption spectra gives an indication about its band gap. 15 The different compositions, estimated band gaps and measured surface areas are summarized in Table 1. All the presented composite materials exhibit surface areas in the same range, thus indicating that the surface area might be a negligible factor when discussing the photocatalytic activities. However, three-component composites seem to have lower surface areas (2.44-3.2 m 2 g À1 ) than two-component composites.

Photocatalytic results
Hydrogen generation from water with and without methanol was chosen as the main photocatalytic reaction for the reactivity evaluation since (111)-layered barium tantalates are known as very active photocatalysts for water splitting. 6 Previous measurements have shown that the optimum loading of cocatalyst for barium tantalate is only 0.0125 wt% of Rh, 15 which is much smaller than any reported co-catalyst loadings for any (111) layered perovskites so far ($0.1 wt%). 9 The materials are initially tested without any co-catalyst in water containing $10 vol% MeOH. Pure Ba 5 Ta 4 O 15 (BaTa 12.78) is active without any deposited co-catalyst as shown in Fig. 3 The resulting core-shell structures aer stepwise Rh deposition and Cr 2 O 3 loading are characterized by TEM and X-ray Photoelectron Spectroscopy (XPS). In Fig. 4 the XPS spectrum conrms the successful formation of metallic Rh and Cr 2 O 3 in the oxidation state (III). Fig. 5 presents TEM images of Ba 5 Ta 4 O 15 loaded with 1.125 wt% Rh and 1.187 wt%; Cr 2 O 3 , which were prepared to better visualize the core-shell nature of  our co-catalysts. Moreover, these amounts are in the range of typical loadings reported earlier. 36 The primary particle size of the Rh nanoparticles is in a range of 3-5 nm, although some of them aggregate to form larger secondary particles. These single Rh particles have been coated with a shell layer to form the whole core-shell structure on the catalyst surface. Interestingly the Cr 2 O 3 shell shows always constant thickness of about 2-3 nm which was also already reported by Domen and his coworkers. 17 In Fig. 6 the OWS results are shown for 0.0125 wt% Rh-loaded Ba 5 Ta 4 O 15 with different amounts of deposited Cr 2 O 3 . A content of 0.00625 wt% Cr 2 O 3 leads to no O 2 generation but shows an H 2 rate of 441 mmol h À1 in pure water, suggesting that the back reaction to H 2 O is not successfully suppressed due to the low amount of Cr 2 O 3 .
The most active catalyst can be achieved with only 0.0125 wt% of Cr 2 O 3 . Pure water is split into H 2 (446 mmol h À1 ) and O 2 (229 mmol h À1 ) in the nearly optimum stoichiometric ratio. With further Cr 2 O 3 addition the H 2 and O 2 rates decrease signicantly to 358 mmol h À1 and 163 mmol h À1 , suggesting that Cr 2 O 3 added in excess acts as inhibitor. Taking the above results of Ba 5 Ta 4 O 15 into account, we used 0.0125 wt% Cr 2 O 3 for the deposition of Rh-Cr 2 O 3 core-shell co-catalysts onto the composite materials as well as shown below.

Multicomponent photocatalysts
An overview for the H 2 evolution rates including all barium tantalate multicomponent composites is shown in Fig. 7. All materials are active for H 2 generation from H 2 O/MeOH solutions without any co-catalyst on the surface (squares), however show different activities depending on the composition, as will be discussed later. The deposition of 0.0125 wt% Rh leads to different behaviors in the composites. On pure Ba 5 Ta 4 O 15 (BaTa 12.78), the effect of increased activity in presence of Rh is far more pronounced than on the multicomponent composite catalysts BaTa 11.73 to BaTa 5.6. For BaTa 12.78 upon Rh deposition the H 2 evolution increases from 1.8 up to 4 mmol h À1 . Aer a strong decrease for BaTa 11.73, an increase in H 2 evolution for BaTa 10.86 of nearly 150% was detected without co-catalyst, suggesting an optimum ratio of phase coexistence and surface junctions for this two-component heterojunction. 15 In the BaTa 9.94 to BaTa 9.68 region the rate rapidly decreases again, suggesting a loss of sufficient surface junctions. The ratio of components leads apparently to a reduction of possible interfaces and to a decreased electron transfer, as was observed recently for calcium tantalate composite photocatalysts. 37 Below BaTa 8.83 BaTa 2 O 6 arises as third material as seen from the XRD patterns in Fig. 1. Down to BaTa 6.13 an extraordinary increase in H 2 evolution (up to 6.2 mmol h À1 ) by over 300% could be obtained. The in situ formation of a Ba 5 Ta 4 O 15 , Ba 3 Ta 5 O 15 and BaTa 2 O 6 three-component heterojunction obviously leads to reduced charge carrier recombination by spatial separation, with possibly also optimized interface junctions.    A further reduction of barium concentration (BaTa 5.6) in the synthesis shows a dramatic decrease in activity, due to a lower content of Ba 5 Ta 4 O 15 and Ba 3 Ta 5 O 15 and the transition to the single BaTa 2 O 6 material, which was also reported before as a less active material in water splitting compared to Ba 5 Ta 4 O 15 . 1 Furthermore, additional reexes in the XRD pattern of BaTa 5.6 were observed which could not be identied yet.
From all these results, the role of composite materials becomes more and more evident. We have roughly calculated band positions of Ba 5 Ta 4 O 15 , Ba 3 Ta 5 O 15 and BaTa 2 O 6 according to a simple procedure proposed by Butler and Ginley, 38 where the estimated band gaps from UV-Vis spectroscopy are used. A possible mechanism for charge transfer is shown in Fig. 8 Moreover, in the presence of a multicomponent heterojunction with three barium tantalate crystal structures with optimum ratio (BaTa 6.13), charge carrier separation on different components seems to be more effective than charge carrier extraction by co-catalyst photodeposition onto single-component Ba 5 Ta 4 O 15 for H 2 generation. BaTa 6.13 generates 6.2 mmol h À1 H 2 without any co-catalyst, while pure Ba 5 Ta 4 O 15 evolves only 4 mmol h À1 of H 2 with 0.0125 wt% Rh, which is more than 35% less. This is a clear indication that the formation of intrinsic multicomponent heterojunction photocatalysts is very important as a novel strategy for charge carrier separation without noble metal or other co-catalysts, strongly enhancing photocatalytic activities. 10,15,36 To reveal the OWS in the next step we deposited 0.0125 wt% of Cr 2 O 3 on the composite materials (Fig. 7, triangles). Thereby deposition of Cr 2 O 3 species leads to a reduced H 2 evolution as shown for Ba 5 Ta 4 O 15 , indicating a successful Cr 2 O 3 deposition on Rh taking place. Fig. 9 demonstrates the H 2 and O 2 rates for all barium tantalate composites in pure water with 0.0125 wt% Rh and 0.0125 wt% Cr 2 O 3 . The diagram indicates that all composite photocatalysts are able to generate both gases simultaneously. As mentioned before, the Rh deposition in water containing $10% MeOH solution shows a higher efficiency for BaTa12.78 than for the composites. Thus, we propose the preferred co-catalyst deposition in case of Rh on the main component Ba 5 Ta 4 O 15 . Nevertheless, Fig. 9 exhibits an enhanced OWS for samples synthesized with less than 8.45 mmol Ba 2+ precursor compared to the pure Ba 5 Ta 4 O 15 (Fig. 6). Especially the most active composite BaTa 6.13 consisting of three components yields in an increase up to 650 mmol h À1 H 2 and 348 mmol h À1 O 2 . Aer further reducing the barium amount the rates rapidly decrease again which is similar to the trend in H 2 O/MeOH solution. This distinguishes obviously a signicant reduction of interfaces in the system which directly leads to a decrease of electron transfer and charge separation. Among those newly prepared barium tantalate composites BaTa 10.86 and BaTa 8.83 as two-component materials and particularly BaTa 6.13 as a three-component heterojunction photocatalyst show remarkable photocatalytic activity in hydrogen production from H 2 O/MeOH solutions without any co-catalyst as well as with Rh-Cr 2 O 3 in overall water splitting.

Conclusions
New types of barium tantalate composite photocatalysts prepared via sol-gel citrate route showed excellent activities in photocatalytic hydrogen generation and overall water splitting, higher than a phase-pure Ba 5 Ta 4 O 15 material. The improved activity of the composite is assumed to originate from improved charge carrier separation aer light absorption in the composite photocatalysts. All presented materials and composites generate hydrogen without co-catalyst. The best three component heterojunction Ba 5 Ta 4 O 15 /Ba 3 Ta 5 O 15 /BaTa 2 O 6 (BaTa 6.13) evolves hydrogen with rates up to 6.2 mmol h À1 compared to the pure Ba 5 Ta 4 O 15 (1.8 mmol h À1 ) without any cocatalyst, and up to 6.85 mmol h À1 aer photodeposition of Rh. The formation of a core-shell co-catalyst system consisting of very low amounts of Rh-Cr 2 O 3 exhibits stable rates in overall water splitting up to 650 mmol h À1 H 2 and 348 mmol h À1 O 2 from pure water.