- Email: [email protected]
Layered Dion-Jacobson type niobium oxides for photocatalytic hydrogen production prepared via molten salt synthesis
Catalysis Today 287 (2017) 65–69
Contents lists available at ScienceDirect
Catalysis Today journal homepage: www.elsevier.com/locate/cattod
Layered Dion-Jacobson type niobium oxides for photocatalytic hydrogen production prepared via molten salt synthesis Natalia Kulischow a , Calin Ladasiu a,b , Roland Marschall a,∗ a b
Institute of Physical Chemistry, Justus-Liebig-University Giessen, D-35392 Giessen, Germany Vasile Goldis Western University of Arad, Romania
a r t i c l e
i n f o
Article history: Received 15 June 2016 Received in revised form 22 September 2016 Accepted 10 October 2016 Available online 22 October 2016 Keywords: Dion-Jacobson type niobates Molten salt synthesis Photocatalysis Hydrogen evolution
a b s t r a c t A class of Dion-Jacobson type layered perovskite niobium oxides (AB2 Nb3 O10 with A = K, Rb, Cs and B = Ca, Sr, Ba) was prepared via molten salt method for the first time. By dissolving oxide and carbonate precursors in A-cation chloride melts, this type of synthesis needs only two hours of preparation time resulting in highly crystalline layered niobates at a much shorter synthesis time compared to conventional syntheses, like the solid state reaction. Beside detailed materials characterization, we investigated the influence of A-cation and B-cation variation on the band gap and photocatalytic activity for hydrogen production. It was found that the band gap decreases with the increase in size of the B-cation. A strong dependence on interlayer spacing (influenced by the size of the A-cation) and lattice relaxation can be derived from the steady-state hydrogen evolution rates. © 2016 Elsevier B.V. All rights reserved.
1. Introduction The rising global population, together with the increasing demand for clean energy and shortening fossil fuel resources, requires the development of technologies allowing the conversion of solar light into energy and solar fuels. Water splitting using photocatalytic materials active both in UV and visible light is often discussed as a promising way of hydrogen production as a clean fuel [1,2]. To achieve this, stable photocatalysts with high activity, suitable band positions and band gap energy are needed [3]. Specifically, the conduction band minimum of such catalysts has to be more negative than the reduction potential of H+ /H2 to reduce protons, and the valence band edge has to be more positive than the H2 O/O2 redox potential to oxidize water [4]. Layered perovskite materials, like Dion-Jacobson phase niobium oxides AB2 Nb3 O10 (A = K, Rb, Cs and B = Ca, Sr) are known for their high photocatalytic activity under UV light irradiation [5–10]. Other properties of this group of materials include superconductivity, dielectric behavior, good electron conductivity, photoluminescence and the possibility of easy ion exchange [5–10]. Their structure consists of negatively charged niobium oxide sheets, formed by corner-sharing NbO6 octahedra with B-cations in the
∗ Corresponding author. E-mail address: [email protected] (R. Marschall). http://dx.doi.org/10.1016/j.cattod.2016.10.009 0920-5861/© 2016 Elsevier B.V. All rights reserved.
gaps between them. The positively charged A-cations are located in the interlayer space [11]. It is known that the layered structure of this type of materials is promoting the separation of photogenerated electrons and holes and thus is improving the photocatalytic activity [12]. The relatively large interlayer space allows the exchange of the A-cations by H+ , which dramatically increases the photocatalytic activity [11]. Doping and other types of modification are also more effective compared to non-layered materials. This is because the tunnel like interlayer gaps allowing for a better and more homogenous dopant distribution in the material [13–15]. Another advantage of the layered materials is the possibility to be exfoliated, leading to the production of two-dimensional crystals, also known as “nanosheets”. Concerning the photocatalytic activity, the superiority of “nanosheets” compared to the corresponding bulk materials has been already reported [15–17]. Layered Dion-Jacobson niobium oxides are conventionally prepared by the solid state reaction [9]. The synthesis by the polymerizable complex (PC) method is also known [18]. The disadvantages of these methods are the high reaction temperatures, the exceedingly long reaction times, and the necessity of adding an excess of the A-cation precursor, due to its high volatility during the synthesis process [19,20]. Thus, such methods are unsuitable for industrial application. This work presents the molten salt method as a much more attractive synthesis process for this group of materials.
66
N. Kulischow et al. / Catalysis Today 287 (2017) 65–69
The molten salt method has been previously utilized for the preparation of the calcium tantalate composite photocatalysts and tetragonal tungsten bronze-type nanorods [21,22]. A salt (in our case chlorides) with a lower melting temperature than that of the precursors is added in excess, acting as a solvent. In this respect, the molten salt method differs from the flux method, where the salt is added in a low amount (a few percent of the total weight) in order to enhance the reaction rate. The role of the molten salt is to lower the reaction temperature, increase the reaction rate and the homogeneity of the reaction mixture, as well as to control the particle size, shape and agglomeration rate [23]. We herein report for the first time the successful application of this method for the preparation of the family of layered DionJacobson perovskite type materials AB2 Nb3 O10 (A = K, Rb, Cs and B = Ca, Sr, Ba). The physical properties of the compounds synthesized in this manner, as well as their photocatalytic activity for hydrogen production will be presented, the latter being strongly influenced by interlayer spacing and lattice relaxation. 2. Experimental 2.1. Reagents and materials All chemicals were of analytical grade and used as received. BaCO3 (Alfa Aesar, 99.9%), CaCO3 (Gruessing, 99.0%), CsCl (Alfa Aesar, 99.9%), Cs2 CO3 (Alfa Aesar, 99.9%), KCl (Gruessing, 99.0%), K2 CO3 (Grüssing, 99.0%), methanol (J.T. Baker, a. g.), Na3 RhCl6 (Sigma-Aldrich, 99.9%), Nb2 O5 (Fluka, 99.9%), perchloric acid (Sigma-Aldrich, 70%), RbCl (Acros Organics, 99.9%), Rb2 CO3 (SigmaAldrich, 99.8%), SrCO3 (Gruessing, 99.0%), Sr(NO3 )2 (Acros Organics, 99.9%) 2.2. Catalysts preparation As shown schematically in Fig. S1 (Supplementary data), the AB2 Nb3 O10 compounds were prepared via molten salt method by mixing BCO3 or B(NO3 )2 , (B = Ca, Sr, Ba), Nb2 O5 , A2 CO3 with ACl (A = K, Rb, Cs) at a weight ratio 1:2 and heating the mixture in an alumina crucible at a compound specific reaction temperature (Table 1) for 2 h, using a Carbolite CWF 1300 muffle furnace. The obtained product was washed several times with distilled water and dried at 100 ◦ C overnight. About 65% of the A-cation chlorides (ACl) could be recovered out of the washing water. The compound A = Rb and B = Ba was also synthesized. However, since the impurity phases still dominate the product, it was left out of the following discussion. The synthesis of the compound A = K and B = Ba was not successful with this method. 2.3. Characterization X-ray diffraction patterns were measured on a PANalytical MPD diffractometer using Cu-K␣ radiation ( = 0.1541 nm) in the 2 range from 5◦ to 55◦ . The phase purity was confirmed using the HighScore Plus software version 3.0e (3.0.5) and the ICSD database (CsCa2 Nb3 O10 98-020-1425 [24], CsSr2 Nb3 O10 98-009-3675 [7],
CsBa2 Nb3 O10 98-009-3676 [7], RbCa2 Nb3 O10 98-026-0289 [25], RbSr2 Nb3 O10 98-009-3674 [26], KCa2 Nb3 O10 98-009-1098 [27]). The phase purity of KSr2 Nb3 O10 was confirmed by comparing the XRD pattern of the compound with the current literature [28]. Raman spectra were acquired on a SENTERRA dispersive Raman microscope from Bruker Optics equipped with an Olympus (MPlanN 50x) objective and a Nd:YAG laser ( = 532 nm, P = 2.0 mW). Scanning electron microscopy (SEM) images were recorded on a Philips LEO Gemini 982 field emission SEM at an operating voltage of 3 kV. The EDX elemental analysis was conducted on the same instrument at 20 kV accelerating voltage. Transmission electron microscopy (TEM) and selected area electron diffraction (SAED) were performed on a Philips CM30 with 300 kV acceleration voltage and carbon-filmed copper mesh grids for sample preparation. Thermogravimetric analysis (TGA) was performed on a NETZSCH STA 409 PC/PG instrument using an Al2 O3 crucible under the following conditions: 30 mL/min. Ar gas flow, temperature range between 25 ◦ C and 1000 ◦ C. BET measurements were conducted on a Quantachrome Quadrasorb evo instrument at the temperature of liquid nitrogen (−196 ◦ C). UV–vis diffuse reflectance measurements were performed on a Perkin Elmer Lambda 750 spectrometer equipped with a Praying-Mantis accessory. The reflectance spectra were recorded between 200 and 500 nm using BaSO4 as reference. 2.4. Hydrogen production—experimental setup and procedure The photocatalytic hydrogen production using the synthesized materials was performed in a double-walled inner irradiation-type quartz reactor connected to a homemade closed gas evolution system (Fig. S2 Supplementary data). To prevent any thermal catalytic effect, the reactor was cooled down to 10 ◦ C using a double walled quartz jacket, through which cooling water was circulated from a thermostat (Lauda). A 350 W Hg immersion lamp (UV-Consulting, Peschl) was used as light source. High purity argon was used as the carrier gas for the reaction products. The flow rate was set at 25 mL/min. and controlled by a Bronkhorst mass flow controller. The evolved hydrogen was detected using a Shimadzu GC-2014 gas chromatograph equipped with a thermal conductivity (TCD) detector and RESTEK ShinCarbon ST 100/120 column. The column was kept at 35 ◦ C throughout the measurement. The elution time for H2 was 1 min. In a typical experiment 0.3 g photocatalyst was suspended (under sonication) in 600 mL aqueous methanol (10 % v/v) solution. The solution initial pH was adjusted to 5 with perchloric acid. Prior to irradiation the system was purged with argon at 100 mL/min to ensure complete air removal. The reaction was allowed to proceed for two hours without co-catalyst. After that, Rh was in-situ photodeposited (0.05 wt.-% Rh loading) on the catalyst from Na3 RhCl6 (1.7 mL) introduced in the system with a syringe through a reactor inlet and rubber sealing (Fig. S2 Supplementary data). This procedure has the advantage that the reactor does not have to be opened, reducing the flushing time to remove air after precursor addition. Afterwards, the irradiation continued for another five hours, with a short induction period in H2 generation due to Rh photodeposition.
Table 1 Compound specific precursors and reaction temperatures. Compound
A Precursor
B Precursor
ACl
Weight ratio (precursors:ACl)
CsCa2 Nb3 O10 CsSr2 Nb3 O10 CsBa2 Nb3 O10 RbCa2 Nb3 O10 RbSr2 Nb3 O10 KCa2 Nb3 O10 KSr2 Nb3 O10
CsCO3 CsCO3 CsCO3 RbCO3 RbCO3 KCO3 KCO3
CaCO3 Sr(NO3 )2 BaCO3 CaCO3 Sr(NO3 )2 CaCO3 SrCO3
CsCl CsCl CsCl RbCl RbCl KCl KCl
1:2 1:2 1:2 1:2 1:2 1:2 1:2
Temperature 750 ◦ C 1000 ◦ C 850 ◦ C 900 ◦ C 900 ◦ C 1000 ◦ C 1200 ◦ C
N. Kulischow et al. / Catalysis Today 287 (2017) 65–69
67
Fig. 1. XRD patterns of a) CsCa2 Nb3 O10 b) CsSr2 Nb3 O10 c) CsBa2 Nb3 O10 d) RbCa2 Nb3 O10 e) RbSr2 Nb3 O10 f) KCa2 Nb3 O10 g) KSr2 Nb3 O10 .
3. Results and discussion 3.1. Characterization The X-ray powder diffraction patterns of the AB2 Nb3 O10 compounds are presented in Fig. 1. All powders show high crystallinity and reflections characteristic for the Dion-Jacobson group of compounds. The diffraction patterns show mainly a single phase assigned to the layered AB2 Nb3 O10 compounds, with the exception of CsBa2 Nb3 O10 that also includes small amounts of an impurity phase of Ba5 Nb4 O15 (ICSD 98-015-7477 [29]). Further analysis of the XRD patterns also reveals both a shift to lower angle, as well as space group changes. The shift to lower angle is due to the distortion of the NbO6 octahedra, which is attributed to the increasing metal-oxygen bond length with increasing B-cation radius [9]. The change in space groups from Pnma when A = Cs and B = Ca to P4/mmm when A = Cs, Rb and B = Ca, Sr, Ba; and to Cmcm when A = K and B = Ca, Sr, is due to a difference in stacking of the perovskite slabs of the different compounds and the distortion of the NbO6 octahedral units [19,27,28]. The adjacent slabs are mutually displaced by a/2 in the A = K compounds, while the slabs of A = Cs and Rb are stacked without displacement, resulting in different coordination environments for the interlayer alkali cations [19,27]. Lattice parameters are given in the Supporting information. The local structure was closely examined by Raman spectroscopy. Fig. 2 shows the Raman spectra collected between 1000 cm−1 and 400 cm−1 . All samples show the bands characteristic for a perovskite structure. The bands at 930–940 cm−1 represent the Nb = O terminal stretching mode, the bands at 680–770 cm−1 and 530–570 cm−1 represent the edge-shared octahedral NbO6 symmetric stretching modes [30]. The bands at 430–500 cm−1 are assigned to the edge-shared octahedral NbO6 antisymmetric stretching modes [30]. The spectra of KCa2 Nb3 O10 and KSr2 Nb3 O10 show another Nb = O terminal stretching mode represented by the bands at 890–910 cm−1 that arise due to unequal Nb = O bond lengths [30]. The spectrum of CsBa2 Nb3 O10 shows further bands at 850 cm−1 , 770 cm−1 and 490 cm−1 which can be attributed to the impurity Ba5 Nb4 O15 present in the sample [31]. The shift to lower wavenumbers with the increasing B-cation radius is attributed to the increasing metal-oxygen bond lengths [32]. It can be also observed that the A-cation has no influence on the band positions in the corresponding compounds. The results of the energy dispersive X-ray spectroscopy measurements are presented in Table S1 (Supplementary data). The averaged atomic percentages confirm the expected stoichiometric composition of the synthesized materials, despite slight deviations from the expected ideal values. The deviations can be explained by the imprecision of the EDX measurements and the presence of a Ba5 Nb4 O10 by-phase in the CsBa2 Nb3 O10 sample.
Fig. 2. Raman spectra of a) CsCa2 Nb3 O10 b) CsSr2 Nb3 O10 c) CsBa2 Nb3 O10 d) RbCa2 Nb3 O10 e) RbSr2 Nb3 O10 f) KCa2 Nb3 O10 g) KSr2 Nb3 O10 .
Fig. 3. SEM image of CsCa2 Nb3 O10 and the schematic crystal structure, shown exemplarily for this family of compounds.
Fig. 3 shows the SEM image of CsCa2 Nb3 O10 representative for the prepared layered niobium oxides. In addition, SEM images of the compounds can be seen in Fig. S3 in the Supporting information. The lamellar structure of the layered compounds can be clearly seen, which is consistent with the structures previously reported in the literature [3,33]. The observed particles sizes are in some cases several hundred nanometers large, showing crystal growth in two dimensions, as expected on the basis of the crystal structure.
68
N. Kulischow et al. / Catalysis Today 287 (2017) 65–69
Fig. 4. TEM image and SAED pattern of CsCa2 Nb3 O10 .
Fig. 6. Hydrogen evolution rates of the synthesized materials. Fig. 5. Tauc plots and derived band gap energies of layered niobates.
Thermal gravimetric analysis (Fig. S4 Supplementary data) showed no mass loss, beyond that of water, indicating thermal stability of the synthesized materials at temperatures up to 1000 ◦ C. BET measurements showed very low surface areas (1–7 m2 /g) for all of the compounds, which is expected for this type of materials. For further references see Table S2 (Supplementary data). The TEM image of CsCa2 Nb3 O10 and the corresponding selected area electron diffraction (SAED) pattern presented in Fig. 4 is indicative of the single crystal nature of the synthesized particles. The highly ordered diffraction spots can be indexed to an orthorhombic crystal system [34]. The image can be considered representative for the entire group of materials. Fig. 5 shows the Tauc plots calculated from Kubelka-Munk transformations of the absorption spectra of the synthesized compounds. The band gap energies were estimated from the intersection point of the linear part of the graph (F(R)·h)0.5 with the extrapolated baseline, (Fig. S5 Supplementary data). The band gaps of AB2 Nb3 O10 varied with B-cation incorporation, being 3.6 eV in case of Ca, 3.2 (3.3) eV in case of Sr, and 3.0 eV in case of Ba, following the order Ca > Sr > Ba. The observed band gap narrowing suggests a dependence on the ionic radii of the Bcations that fill the spaces between the NbO6 octahedral units. It is known that the band gap energy of a semiconductor is influenced by the overlap of the conduction band forming orbitals [34,35]. The conduction band for these niobates consists of the Nb 4d orbitals.
As the ionic radius of the B-cation increases, the overlap of the Nb 4d orbitals is greater stabilizing the conduction band energy and consequently the band is shifted negatively [35], leading to the observed band gap narrowing. On the other hand, the ionic radii of the A-cations at the interlayer for the same B- cation has no influence on the band gap energy. 3.2. Hydrogen production Fig. 6 shows the time course of photocatalytic hydrogen evolution with the presented layered niobates from water/MeOH solution. All samples show some photocatalytic activity even without the addition of co-catalyst. The hydrogen generation rates are strongly improved for all compounds after the addition of the cocatalyst, reaching a steady state after a certain time, depending on the material, without significant loss of photocatalytic activity. The addition of Rh co-catalyst is known to improve the hydrogen generation rates due to charge separation, the reduction of activation energies and creation of catalytic active sites for H2 production [36,37]. The highest overall hydrogen production for this class of compounds was achieved by CsCa2 Nb3 O10 , and the lowest activity was exhibited by CsBa2 Nb3 O10. The Sr2+ containing compounds showed moderate to low activity. The observed hydrogen evolution rates suggest that the type of the A and B cations and combination thereof has an impact on the photocatalytic activity of these materials.
N. Kulischow et al. / Catalysis Today 287 (2017) 65–69
From Fig. 6, it can be observed that the photocatalytic activity is influenced mainly by the type of the A-cation at the interlayer as follows: Cs > Rb > K, suggesting a dependence on the ionic radii of these cations. Same observation was made by Miseki et al. studying the photocatalytic activity of layered perovskite compounds A5 Nb4 O15 (A = Sr, Ba), Ba3 LaNb3 O12 , ALa4 Ti4 O15 (A = Cs, Sr, Ba) and La4 Ti3 O12 [34]. According to the authors, the observed photocatalytic behavior is due to the conduction band level and the structural anisotropy that are caused by the A-cations. Furthermore, the different distribution and size of the A-cations at the interlayer and the caused anisotropy promotes charge separation [34], for which we have no direct evidence yet. Moreover, it can also influence the distribution of the co-catalyst by photodeposition inside the interlayers spacing, due to better accessibility for the precursors, leading to the observed photocatalytic activity order. RbSr2 Nb3 O10 however does not follow the trend Cs > Rb > K when B = Sr, but rather Rb > Cs > K order. The reason for this behavior is yet unknown warranting further investigation. The type of the B-cations also influences the photocatalytic activity although to a lesser extent, the observed order in the case A = Cs being Ca > Sr > Ba after reaching steady state, for A = K being Ca > Sr. According to Scaife [38], the O 2p orbitals in oxides not containing partly-filled d-levels can be estimated to be indifferent. Therefore, an increase in band gap energy, as observed in absorption spectroscopy in the order B = Ca > Sr > Ba, indicates a cathodic shift of the conduction band edge with smaller B-cation. Thus, the layered niobates containing Ca as B-cation have the highest conduction band minimum, absorbing the fewest number of photons, but exhibiting the largest driving force for proton reduction. Enhancement of hydrogen production by increasing the conduction band edge is a typical phenomenon known in literature [22,39]. Combined, both observed trends can explain the highest activity for CsCa2 Nb3 O10 . 4. Conclusions
Acknowledgements We thank Pascal Voepel for TEM measurements, Hubert Woerner for TGA measurements, and Ruediger Ellinghaus for performing the BET measurements (all Justus-Liebig-University Giessen). We acknowledge financial support by the BMBF (Bun¨ Bildung und Forschung), research project desministerium fur DuaSol (03SF0482D). R.M. gratefully acknowledges funding in the Emmy-Noether program (MA 5392/3-1) of the German Research Foundation DFG. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cattod.2016.10. 009. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]
We have successfully prepared a class of Dion-Jacobson type layered perovskite niobium oxides (AB2 Nb3 O10 where A = K, Rb, Cs and B = Ca, Sr, Ba) using for the first time a molten salt method. This synthesis route enabled the preparation of highly crystalline layered niobates needing only two hours of synthesis time compared to conventional methods such as solid state or PC. XRD, Raman and EDX analyses showed mainly phase pure characteristic DionJacobson type structures and the expected elemental composition of the materials. SEM and TEM images revealed a lamellar structure and a single crystal nature of the compounds. TGA indicates thermal stability up to 1000 ◦ C with negligible mass loss. All compounds have low but similar BET surface area typical for this class of perovskites. The estimated band gap energies of the prepared compounds showed a dependence on the ionic radii of the B-cations due to the stabilization of the conduction band and consequent negative shift as the ionic radius increases. The A-cations had no influence on the band gap energy. All samples showed some photocatalytic activity even without the co-catalyst. After in-situ photodeposition of Rh co-catalyst, the photocatalytic behavior is influenced on one hand by the type of the A-cation at the interlayer, suggesting a dependence on the ionic radii of these cations in the order Cs > Rb > K, with the exception of RbSr2 Nb3 O10 . On the other hand, the B-cation influences the hydrogen generation rates due to more cathodic conduction band edges in the order Ca > Sr > Ba. Further work is in progress to elucidate the behavior of RbSr2 Nb3 O10 and to investigate the photocatalytic activity of these compounds after doping, ion exchange and exfoliation.
69
[17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39]
R. Marschall, L. Wang, Catal. Today 225 (2014) 111–135. N. Armaroli, V. Balzani, Chem.Sus.Chem 4 (2011) 21–36. Y. Hu, P. Guo, L. Guo, Int. J. Hydrogen Energy 37 (2012) 1007–1013. K. Honda, A. Fujishima, Nature 238 (1972) 37–38. M. Kato, Y. Imai, T. Kajita, Y. Takarabe, T. Minakawa, K. Nemoto, H. Tezuka, T. Nojia, Y. Koike, Mater. Sci. Eng. B 148 (2008) 53–57. M. Fang, C.-H. Kim, T. Mallouk, Chem. Mater. 11 (1999) 1519–1525. V. Thangadurai, P. Schmid-Beurmann, W. Weppner, J. Solid State Chem. 158 (2001) 279–289. S. Ida, C. Ogata, M. Eguchi, W. Youngblood, T. Mallouk, Y. Matsumoto, J. Am. Chem. Soc. 130 (2008) 7052–7059. M. Dion, M. Ganne, M. Tournoux, Mater. Res. Bull. 16 (1981) 1429–1435. A. Jacobson, J. Johnson, J. Lewandowski, Inorg. Chem. 24 (1985) 3727–3729. T. Takata, A. Tanaka, M. Hara, N. Junko, Catal. Today 44 (1998) 17–26. A. Tanaka, J. Kondo, K. Domen, Crit. Rev. Surf. Chem. 5 (1995) 305–326. Y. Huang, Y. Xie, L. Fan, Y. Li, Y. Wei, J. Lin, Int. J. Hydrogen Energy 33 (2008) 6432–6438. H. Hata, Y. Kobayashi, V. Bojan, W. Youngblood, T. Mallouk, Nano Lett. 8 (2008) 794–799. Y. Ebina, T. Sasaki, M. Harada, M. Watanabe, Chem. Mater. 14 (2002) 4390–4395. E. Sabio, R. Chamousis, N. Browning, F. Osterloh, J. Phys. Chem. C 116 (2012) 3161–3170. T. Oshima, O. Ishitani, K. Maeda,Adv. Mater. Interfaces 1 (2014), 1400131 1–4. K. Maeda, M. Eguchi, W. Youngblood, T. Mallouk, J. Mater. Chem. 19 (2009) 4813–4818. V. Thangadurai, P. Schmid-Beurmann, W. Weppner, J. Solid State Chem. 158 (2001) 279–289. J. Schottenfeld, A. Benesi, P. Stephens, G. Chen, P. Eklund, T. Mallouk, J. Solid State Chem. 178 (2005) 2313–2321. P. Wang, P. Chen, A. Kostka, R. Marschall, M. Wark, Chem. Mater. 25 (2013) 4739–4745. P. Wang, L. Schwertmann, R. Marschall, M. Wark, J. Mater. Chem. A 2 (2014) 8815–8822. T. Kimura, Advances in ceramics—synthesis and characterization, processing and specific application, Prof. Costas Sikalidis Ed. 4 (2011) 75–100. M. Dion, M. Ganne, M. Tournoux, Rev. de Chim. Miner. 21 (1984) 92–103. L. Zhenhua, T. Kaibin, C. Qianwang, Z. Huagui, Acta Crystallogr. E Struct. Rep. Online 65 (2009) 0. V. Thangadurai, P. Schmid-Beurmann, W. Weppner, Golden Book Phase Transit. 1 (2002) 1–123. H. Fukuoka, T. Isami, S. Yamanaka, J. Solid State Chem. 151 (2000) 40–45. G. Yang, Y. Kong, W. Hou, Q. Yan, J. Phys. Chem. B 109 (2005) 1371–1379. T. Vanderah, T. Collins, W. Wong-Ng, R. Roth, L. Farber, J. Alloys Compd. 346 (2002) 116–128. J.-M. Jehng, I. Wachs, Chem. Mater. 3 (1991) 100–107. N. Massa, S. Pagola, R. Carbonio, Phs. Rev. B 53 (1996) 8148–8150. K. Maeda, M. Eguchi, T. Oshima, Angew. Chem. Int. Ed. 53 (2014) 13164–13168. V. Thangadurai, W. Weppner, Chem. Mater. 14 (2002) 1136–1143. Y. Miseki, H. Kato, A. Kudo, Energy Environ. Sci. 2 (2009) 306–314. T. Oshima, M. Eguchi, K. Maeda, Chem. Sus. Chem 9 (2016) 396–402. K. Kamata, K. Maeda, D. Lu, Y. Kako, K. Domen, Chem. Phys. Lett. 470 (2009) 90–94. R. Marschall, Adv. Funct. Mater. 24 (2014) 2421–2440. D. Scaife, Sol. Energy 25 (1980) 41–54. M. Yoshino, M. Kakihana, W. Cho, H. Kato, A. Kudo, Chem. Mater. 14 (2002) 3369–3376.
Contents lists available at ScienceDirect
Catalysis Today journal homepage: www.elsevier.com/locate/cattod
Layered Dion-Jacobson type niobium oxides for photocatalytic hydrogen production prepared via molten salt synthesis Natalia Kulischow a , Calin Ladasiu a,b , Roland Marschall a,∗ a b
Institute of Physical Chemistry, Justus-Liebig-University Giessen, D-35392 Giessen, Germany Vasile Goldis Western University of Arad, Romania
a r t i c l e
i n f o
Article history: Received 15 June 2016 Received in revised form 22 September 2016 Accepted 10 October 2016 Available online 22 October 2016 Keywords: Dion-Jacobson type niobates Molten salt synthesis Photocatalysis Hydrogen evolution
a b s t r a c t A class of Dion-Jacobson type layered perovskite niobium oxides (AB2 Nb3 O10 with A = K, Rb, Cs and B = Ca, Sr, Ba) was prepared via molten salt method for the first time. By dissolving oxide and carbonate precursors in A-cation chloride melts, this type of synthesis needs only two hours of preparation time resulting in highly crystalline layered niobates at a much shorter synthesis time compared to conventional syntheses, like the solid state reaction. Beside detailed materials characterization, we investigated the influence of A-cation and B-cation variation on the band gap and photocatalytic activity for hydrogen production. It was found that the band gap decreases with the increase in size of the B-cation. A strong dependence on interlayer spacing (influenced by the size of the A-cation) and lattice relaxation can be derived from the steady-state hydrogen evolution rates. © 2016 Elsevier B.V. All rights reserved.
1. Introduction The rising global population, together with the increasing demand for clean energy and shortening fossil fuel resources, requires the development of technologies allowing the conversion of solar light into energy and solar fuels. Water splitting using photocatalytic materials active both in UV and visible light is often discussed as a promising way of hydrogen production as a clean fuel [1,2]. To achieve this, stable photocatalysts with high activity, suitable band positions and band gap energy are needed [3]. Specifically, the conduction band minimum of such catalysts has to be more negative than the reduction potential of H+ /H2 to reduce protons, and the valence band edge has to be more positive than the H2 O/O2 redox potential to oxidize water [4]. Layered perovskite materials, like Dion-Jacobson phase niobium oxides AB2 Nb3 O10 (A = K, Rb, Cs and B = Ca, Sr) are known for their high photocatalytic activity under UV light irradiation [5–10]. Other properties of this group of materials include superconductivity, dielectric behavior, good electron conductivity, photoluminescence and the possibility of easy ion exchange [5–10]. Their structure consists of negatively charged niobium oxide sheets, formed by corner-sharing NbO6 octahedra with B-cations in the
∗ Corresponding author. E-mail address: [email protected] (R. Marschall). http://dx.doi.org/10.1016/j.cattod.2016.10.009 0920-5861/© 2016 Elsevier B.V. All rights reserved.
gaps between them. The positively charged A-cations are located in the interlayer space [11]. It is known that the layered structure of this type of materials is promoting the separation of photogenerated electrons and holes and thus is improving the photocatalytic activity [12]. The relatively large interlayer space allows the exchange of the A-cations by H+ , which dramatically increases the photocatalytic activity [11]. Doping and other types of modification are also more effective compared to non-layered materials. This is because the tunnel like interlayer gaps allowing for a better and more homogenous dopant distribution in the material [13–15]. Another advantage of the layered materials is the possibility to be exfoliated, leading to the production of two-dimensional crystals, also known as “nanosheets”. Concerning the photocatalytic activity, the superiority of “nanosheets” compared to the corresponding bulk materials has been already reported [15–17]. Layered Dion-Jacobson niobium oxides are conventionally prepared by the solid state reaction [9]. The synthesis by the polymerizable complex (PC) method is also known [18]. The disadvantages of these methods are the high reaction temperatures, the exceedingly long reaction times, and the necessity of adding an excess of the A-cation precursor, due to its high volatility during the synthesis process [19,20]. Thus, such methods are unsuitable for industrial application. This work presents the molten salt method as a much more attractive synthesis process for this group of materials.
66
N. Kulischow et al. / Catalysis Today 287 (2017) 65–69
The molten salt method has been previously utilized for the preparation of the calcium tantalate composite photocatalysts and tetragonal tungsten bronze-type nanorods [21,22]. A salt (in our case chlorides) with a lower melting temperature than that of the precursors is added in excess, acting as a solvent. In this respect, the molten salt method differs from the flux method, where the salt is added in a low amount (a few percent of the total weight) in order to enhance the reaction rate. The role of the molten salt is to lower the reaction temperature, increase the reaction rate and the homogeneity of the reaction mixture, as well as to control the particle size, shape and agglomeration rate [23]. We herein report for the first time the successful application of this method for the preparation of the family of layered DionJacobson perovskite type materials AB2 Nb3 O10 (A = K, Rb, Cs and B = Ca, Sr, Ba). The physical properties of the compounds synthesized in this manner, as well as their photocatalytic activity for hydrogen production will be presented, the latter being strongly influenced by interlayer spacing and lattice relaxation. 2. Experimental 2.1. Reagents and materials All chemicals were of analytical grade and used as received. BaCO3 (Alfa Aesar, 99.9%), CaCO3 (Gruessing, 99.0%), CsCl (Alfa Aesar, 99.9%), Cs2 CO3 (Alfa Aesar, 99.9%), KCl (Gruessing, 99.0%), K2 CO3 (Grüssing, 99.0%), methanol (J.T. Baker, a. g.), Na3 RhCl6 (Sigma-Aldrich, 99.9%), Nb2 O5 (Fluka, 99.9%), perchloric acid (Sigma-Aldrich, 70%), RbCl (Acros Organics, 99.9%), Rb2 CO3 (SigmaAldrich, 99.8%), SrCO3 (Gruessing, 99.0%), Sr(NO3 )2 (Acros Organics, 99.9%) 2.2. Catalysts preparation As shown schematically in Fig. S1 (Supplementary data), the AB2 Nb3 O10 compounds were prepared via molten salt method by mixing BCO3 or B(NO3 )2 , (B = Ca, Sr, Ba), Nb2 O5 , A2 CO3 with ACl (A = K, Rb, Cs) at a weight ratio 1:2 and heating the mixture in an alumina crucible at a compound specific reaction temperature (Table 1) for 2 h, using a Carbolite CWF 1300 muffle furnace. The obtained product was washed several times with distilled water and dried at 100 ◦ C overnight. About 65% of the A-cation chlorides (ACl) could be recovered out of the washing water. The compound A = Rb and B = Ba was also synthesized. However, since the impurity phases still dominate the product, it was left out of the following discussion. The synthesis of the compound A = K and B = Ba was not successful with this method. 2.3. Characterization X-ray diffraction patterns were measured on a PANalytical MPD diffractometer using Cu-K␣ radiation ( = 0.1541 nm) in the 2 range from 5◦ to 55◦ . The phase purity was confirmed using the HighScore Plus software version 3.0e (3.0.5) and the ICSD database (CsCa2 Nb3 O10 98-020-1425 [24], CsSr2 Nb3 O10 98-009-3675 [7],
CsBa2 Nb3 O10 98-009-3676 [7], RbCa2 Nb3 O10 98-026-0289 [25], RbSr2 Nb3 O10 98-009-3674 [26], KCa2 Nb3 O10 98-009-1098 [27]). The phase purity of KSr2 Nb3 O10 was confirmed by comparing the XRD pattern of the compound with the current literature [28]. Raman spectra were acquired on a SENTERRA dispersive Raman microscope from Bruker Optics equipped with an Olympus (MPlanN 50x) objective and a Nd:YAG laser ( = 532 nm, P = 2.0 mW). Scanning electron microscopy (SEM) images were recorded on a Philips LEO Gemini 982 field emission SEM at an operating voltage of 3 kV. The EDX elemental analysis was conducted on the same instrument at 20 kV accelerating voltage. Transmission electron microscopy (TEM) and selected area electron diffraction (SAED) were performed on a Philips CM30 with 300 kV acceleration voltage and carbon-filmed copper mesh grids for sample preparation. Thermogravimetric analysis (TGA) was performed on a NETZSCH STA 409 PC/PG instrument using an Al2 O3 crucible under the following conditions: 30 mL/min. Ar gas flow, temperature range between 25 ◦ C and 1000 ◦ C. BET measurements were conducted on a Quantachrome Quadrasorb evo instrument at the temperature of liquid nitrogen (−196 ◦ C). UV–vis diffuse reflectance measurements were performed on a Perkin Elmer Lambda 750 spectrometer equipped with a Praying-Mantis accessory. The reflectance spectra were recorded between 200 and 500 nm using BaSO4 as reference. 2.4. Hydrogen production—experimental setup and procedure The photocatalytic hydrogen production using the synthesized materials was performed in a double-walled inner irradiation-type quartz reactor connected to a homemade closed gas evolution system (Fig. S2 Supplementary data). To prevent any thermal catalytic effect, the reactor was cooled down to 10 ◦ C using a double walled quartz jacket, through which cooling water was circulated from a thermostat (Lauda). A 350 W Hg immersion lamp (UV-Consulting, Peschl) was used as light source. High purity argon was used as the carrier gas for the reaction products. The flow rate was set at 25 mL/min. and controlled by a Bronkhorst mass flow controller. The evolved hydrogen was detected using a Shimadzu GC-2014 gas chromatograph equipped with a thermal conductivity (TCD) detector and RESTEK ShinCarbon ST 100/120 column. The column was kept at 35 ◦ C throughout the measurement. The elution time for H2 was 1 min. In a typical experiment 0.3 g photocatalyst was suspended (under sonication) in 600 mL aqueous methanol (10 % v/v) solution. The solution initial pH was adjusted to 5 with perchloric acid. Prior to irradiation the system was purged with argon at 100 mL/min to ensure complete air removal. The reaction was allowed to proceed for two hours without co-catalyst. After that, Rh was in-situ photodeposited (0.05 wt.-% Rh loading) on the catalyst from Na3 RhCl6 (1.7 mL) introduced in the system with a syringe through a reactor inlet and rubber sealing (Fig. S2 Supplementary data). This procedure has the advantage that the reactor does not have to be opened, reducing the flushing time to remove air after precursor addition. Afterwards, the irradiation continued for another five hours, with a short induction period in H2 generation due to Rh photodeposition.
Table 1 Compound specific precursors and reaction temperatures. Compound
A Precursor
B Precursor
ACl
Weight ratio (precursors:ACl)
CsCa2 Nb3 O10 CsSr2 Nb3 O10 CsBa2 Nb3 O10 RbCa2 Nb3 O10 RbSr2 Nb3 O10 KCa2 Nb3 O10 KSr2 Nb3 O10
CsCO3 CsCO3 CsCO3 RbCO3 RbCO3 KCO3 KCO3
CaCO3 Sr(NO3 )2 BaCO3 CaCO3 Sr(NO3 )2 CaCO3 SrCO3
CsCl CsCl CsCl RbCl RbCl KCl KCl
1:2 1:2 1:2 1:2 1:2 1:2 1:2
Temperature 750 ◦ C 1000 ◦ C 850 ◦ C 900 ◦ C 900 ◦ C 1000 ◦ C 1200 ◦ C
N. Kulischow et al. / Catalysis Today 287 (2017) 65–69
67
Fig. 1. XRD patterns of a) CsCa2 Nb3 O10 b) CsSr2 Nb3 O10 c) CsBa2 Nb3 O10 d) RbCa2 Nb3 O10 e) RbSr2 Nb3 O10 f) KCa2 Nb3 O10 g) KSr2 Nb3 O10 .
3. Results and discussion 3.1. Characterization The X-ray powder diffraction patterns of the AB2 Nb3 O10 compounds are presented in Fig. 1. All powders show high crystallinity and reflections characteristic for the Dion-Jacobson group of compounds. The diffraction patterns show mainly a single phase assigned to the layered AB2 Nb3 O10 compounds, with the exception of CsBa2 Nb3 O10 that also includes small amounts of an impurity phase of Ba5 Nb4 O15 (ICSD 98-015-7477 [29]). Further analysis of the XRD patterns also reveals both a shift to lower angle, as well as space group changes. The shift to lower angle is due to the distortion of the NbO6 octahedra, which is attributed to the increasing metal-oxygen bond length with increasing B-cation radius [9]. The change in space groups from Pnma when A = Cs and B = Ca to P4/mmm when A = Cs, Rb and B = Ca, Sr, Ba; and to Cmcm when A = K and B = Ca, Sr, is due to a difference in stacking of the perovskite slabs of the different compounds and the distortion of the NbO6 octahedral units [19,27,28]. The adjacent slabs are mutually displaced by a/2 in the A = K compounds, while the slabs of A = Cs and Rb are stacked without displacement, resulting in different coordination environments for the interlayer alkali cations [19,27]. Lattice parameters are given in the Supporting information. The local structure was closely examined by Raman spectroscopy. Fig. 2 shows the Raman spectra collected between 1000 cm−1 and 400 cm−1 . All samples show the bands characteristic for a perovskite structure. The bands at 930–940 cm−1 represent the Nb = O terminal stretching mode, the bands at 680–770 cm−1 and 530–570 cm−1 represent the edge-shared octahedral NbO6 symmetric stretching modes [30]. The bands at 430–500 cm−1 are assigned to the edge-shared octahedral NbO6 antisymmetric stretching modes [30]. The spectra of KCa2 Nb3 O10 and KSr2 Nb3 O10 show another Nb = O terminal stretching mode represented by the bands at 890–910 cm−1 that arise due to unequal Nb = O bond lengths [30]. The spectrum of CsBa2 Nb3 O10 shows further bands at 850 cm−1 , 770 cm−1 and 490 cm−1 which can be attributed to the impurity Ba5 Nb4 O15 present in the sample [31]. The shift to lower wavenumbers with the increasing B-cation radius is attributed to the increasing metal-oxygen bond lengths [32]. It can be also observed that the A-cation has no influence on the band positions in the corresponding compounds. The results of the energy dispersive X-ray spectroscopy measurements are presented in Table S1 (Supplementary data). The averaged atomic percentages confirm the expected stoichiometric composition of the synthesized materials, despite slight deviations from the expected ideal values. The deviations can be explained by the imprecision of the EDX measurements and the presence of a Ba5 Nb4 O10 by-phase in the CsBa2 Nb3 O10 sample.
Fig. 2. Raman spectra of a) CsCa2 Nb3 O10 b) CsSr2 Nb3 O10 c) CsBa2 Nb3 O10 d) RbCa2 Nb3 O10 e) RbSr2 Nb3 O10 f) KCa2 Nb3 O10 g) KSr2 Nb3 O10 .
Fig. 3. SEM image of CsCa2 Nb3 O10 and the schematic crystal structure, shown exemplarily for this family of compounds.
Fig. 3 shows the SEM image of CsCa2 Nb3 O10 representative for the prepared layered niobium oxides. In addition, SEM images of the compounds can be seen in Fig. S3 in the Supporting information. The lamellar structure of the layered compounds can be clearly seen, which is consistent with the structures previously reported in the literature [3,33]. The observed particles sizes are in some cases several hundred nanometers large, showing crystal growth in two dimensions, as expected on the basis of the crystal structure.
68
N. Kulischow et al. / Catalysis Today 287 (2017) 65–69
Fig. 4. TEM image and SAED pattern of CsCa2 Nb3 O10 .
Fig. 6. Hydrogen evolution rates of the synthesized materials. Fig. 5. Tauc plots and derived band gap energies of layered niobates.
Thermal gravimetric analysis (Fig. S4 Supplementary data) showed no mass loss, beyond that of water, indicating thermal stability of the synthesized materials at temperatures up to 1000 ◦ C. BET measurements showed very low surface areas (1–7 m2 /g) for all of the compounds, which is expected for this type of materials. For further references see Table S2 (Supplementary data). The TEM image of CsCa2 Nb3 O10 and the corresponding selected area electron diffraction (SAED) pattern presented in Fig. 4 is indicative of the single crystal nature of the synthesized particles. The highly ordered diffraction spots can be indexed to an orthorhombic crystal system [34]. The image can be considered representative for the entire group of materials. Fig. 5 shows the Tauc plots calculated from Kubelka-Munk transformations of the absorption spectra of the synthesized compounds. The band gap energies were estimated from the intersection point of the linear part of the graph (F(R)·h)0.5 with the extrapolated baseline, (Fig. S5 Supplementary data). The band gaps of AB2 Nb3 O10 varied with B-cation incorporation, being 3.6 eV in case of Ca, 3.2 (3.3) eV in case of Sr, and 3.0 eV in case of Ba, following the order Ca > Sr > Ba. The observed band gap narrowing suggests a dependence on the ionic radii of the Bcations that fill the spaces between the NbO6 octahedral units. It is known that the band gap energy of a semiconductor is influenced by the overlap of the conduction band forming orbitals [34,35]. The conduction band for these niobates consists of the Nb 4d orbitals.
As the ionic radius of the B-cation increases, the overlap of the Nb 4d orbitals is greater stabilizing the conduction band energy and consequently the band is shifted negatively [35], leading to the observed band gap narrowing. On the other hand, the ionic radii of the A-cations at the interlayer for the same B- cation has no influence on the band gap energy. 3.2. Hydrogen production Fig. 6 shows the time course of photocatalytic hydrogen evolution with the presented layered niobates from water/MeOH solution. All samples show some photocatalytic activity even without the addition of co-catalyst. The hydrogen generation rates are strongly improved for all compounds after the addition of the cocatalyst, reaching a steady state after a certain time, depending on the material, without significant loss of photocatalytic activity. The addition of Rh co-catalyst is known to improve the hydrogen generation rates due to charge separation, the reduction of activation energies and creation of catalytic active sites for H2 production [36,37]. The highest overall hydrogen production for this class of compounds was achieved by CsCa2 Nb3 O10 , and the lowest activity was exhibited by CsBa2 Nb3 O10. The Sr2+ containing compounds showed moderate to low activity. The observed hydrogen evolution rates suggest that the type of the A and B cations and combination thereof has an impact on the photocatalytic activity of these materials.
N. Kulischow et al. / Catalysis Today 287 (2017) 65–69
From Fig. 6, it can be observed that the photocatalytic activity is influenced mainly by the type of the A-cation at the interlayer as follows: Cs > Rb > K, suggesting a dependence on the ionic radii of these cations. Same observation was made by Miseki et al. studying the photocatalytic activity of layered perovskite compounds A5 Nb4 O15 (A = Sr, Ba), Ba3 LaNb3 O12 , ALa4 Ti4 O15 (A = Cs, Sr, Ba) and La4 Ti3 O12 [34]. According to the authors, the observed photocatalytic behavior is due to the conduction band level and the structural anisotropy that are caused by the A-cations. Furthermore, the different distribution and size of the A-cations at the interlayer and the caused anisotropy promotes charge separation [34], for which we have no direct evidence yet. Moreover, it can also influence the distribution of the co-catalyst by photodeposition inside the interlayers spacing, due to better accessibility for the precursors, leading to the observed photocatalytic activity order. RbSr2 Nb3 O10 however does not follow the trend Cs > Rb > K when B = Sr, but rather Rb > Cs > K order. The reason for this behavior is yet unknown warranting further investigation. The type of the B-cations also influences the photocatalytic activity although to a lesser extent, the observed order in the case A = Cs being Ca > Sr > Ba after reaching steady state, for A = K being Ca > Sr. According to Scaife [38], the O 2p orbitals in oxides not containing partly-filled d-levels can be estimated to be indifferent. Therefore, an increase in band gap energy, as observed in absorption spectroscopy in the order B = Ca > Sr > Ba, indicates a cathodic shift of the conduction band edge with smaller B-cation. Thus, the layered niobates containing Ca as B-cation have the highest conduction band minimum, absorbing the fewest number of photons, but exhibiting the largest driving force for proton reduction. Enhancement of hydrogen production by increasing the conduction band edge is a typical phenomenon known in literature [22,39]. Combined, both observed trends can explain the highest activity for CsCa2 Nb3 O10 . 4. Conclusions
Acknowledgements We thank Pascal Voepel for TEM measurements, Hubert Woerner for TGA measurements, and Ruediger Ellinghaus for performing the BET measurements (all Justus-Liebig-University Giessen). We acknowledge financial support by the BMBF (Bun¨ Bildung und Forschung), research project desministerium fur DuaSol (03SF0482D). R.M. gratefully acknowledges funding in the Emmy-Noether program (MA 5392/3-1) of the German Research Foundation DFG. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cattod.2016.10. 009. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]
We have successfully prepared a class of Dion-Jacobson type layered perovskite niobium oxides (AB2 Nb3 O10 where A = K, Rb, Cs and B = Ca, Sr, Ba) using for the first time a molten salt method. This synthesis route enabled the preparation of highly crystalline layered niobates needing only two hours of synthesis time compared to conventional methods such as solid state or PC. XRD, Raman and EDX analyses showed mainly phase pure characteristic DionJacobson type structures and the expected elemental composition of the materials. SEM and TEM images revealed a lamellar structure and a single crystal nature of the compounds. TGA indicates thermal stability up to 1000 ◦ C with negligible mass loss. All compounds have low but similar BET surface area typical for this class of perovskites. The estimated band gap energies of the prepared compounds showed a dependence on the ionic radii of the B-cations due to the stabilization of the conduction band and consequent negative shift as the ionic radius increases. The A-cations had no influence on the band gap energy. All samples showed some photocatalytic activity even without the co-catalyst. After in-situ photodeposition of Rh co-catalyst, the photocatalytic behavior is influenced on one hand by the type of the A-cation at the interlayer, suggesting a dependence on the ionic radii of these cations in the order Cs > Rb > K, with the exception of RbSr2 Nb3 O10 . On the other hand, the B-cation influences the hydrogen generation rates due to more cathodic conduction band edges in the order Ca > Sr > Ba. Further work is in progress to elucidate the behavior of RbSr2 Nb3 O10 and to investigate the photocatalytic activity of these compounds after doping, ion exchange and exfoliation.
69
[17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39]
R. Marschall, L. Wang, Catal. Today 225 (2014) 111–135. N. Armaroli, V. Balzani, Chem.Sus.Chem 4 (2011) 21–36. Y. Hu, P. Guo, L. Guo, Int. J. Hydrogen Energy 37 (2012) 1007–1013. K. Honda, A. Fujishima, Nature 238 (1972) 37–38. M. Kato, Y. Imai, T. Kajita, Y. Takarabe, T. Minakawa, K. Nemoto, H. Tezuka, T. Nojia, Y. Koike, Mater. Sci. Eng. B 148 (2008) 53–57. M. Fang, C.-H. Kim, T. Mallouk, Chem. Mater. 11 (1999) 1519–1525. V. Thangadurai, P. Schmid-Beurmann, W. Weppner, J. Solid State Chem. 158 (2001) 279–289. S. Ida, C. Ogata, M. Eguchi, W. Youngblood, T. Mallouk, Y. Matsumoto, J. Am. Chem. Soc. 130 (2008) 7052–7059. M. Dion, M. Ganne, M. Tournoux, Mater. Res. Bull. 16 (1981) 1429–1435. A. Jacobson, J. Johnson, J. Lewandowski, Inorg. Chem. 24 (1985) 3727–3729. T. Takata, A. Tanaka, M. Hara, N. Junko, Catal. Today 44 (1998) 17–26. A. Tanaka, J. Kondo, K. Domen, Crit. Rev. Surf. Chem. 5 (1995) 305–326. Y. Huang, Y. Xie, L. Fan, Y. Li, Y. Wei, J. Lin, Int. J. Hydrogen Energy 33 (2008) 6432–6438. H. Hata, Y. Kobayashi, V. Bojan, W. Youngblood, T. Mallouk, Nano Lett. 8 (2008) 794–799. Y. Ebina, T. Sasaki, M. Harada, M. Watanabe, Chem. Mater. 14 (2002) 4390–4395. E. Sabio, R. Chamousis, N. Browning, F. Osterloh, J. Phys. Chem. C 116 (2012) 3161–3170. T. Oshima, O. Ishitani, K. Maeda,Adv. Mater. Interfaces 1 (2014), 1400131 1–4. K. Maeda, M. Eguchi, W. Youngblood, T. Mallouk, J. Mater. Chem. 19 (2009) 4813–4818. V. Thangadurai, P. Schmid-Beurmann, W. Weppner, J. Solid State Chem. 158 (2001) 279–289. J. Schottenfeld, A. Benesi, P. Stephens, G. Chen, P. Eklund, T. Mallouk, J. Solid State Chem. 178 (2005) 2313–2321. P. Wang, P. Chen, A. Kostka, R. Marschall, M. Wark, Chem. Mater. 25 (2013) 4739–4745. P. Wang, L. Schwertmann, R. Marschall, M. Wark, J. Mater. Chem. A 2 (2014) 8815–8822. T. Kimura, Advances in ceramics—synthesis and characterization, processing and specific application, Prof. Costas Sikalidis Ed. 4 (2011) 75–100. M. Dion, M. Ganne, M. Tournoux, Rev. de Chim. Miner. 21 (1984) 92–103. L. Zhenhua, T. Kaibin, C. Qianwang, Z. Huagui, Acta Crystallogr. E Struct. Rep. Online 65 (2009) 0. V. Thangadurai, P. Schmid-Beurmann, W. Weppner, Golden Book Phase Transit. 1 (2002) 1–123. H. Fukuoka, T. Isami, S. Yamanaka, J. Solid State Chem. 151 (2000) 40–45. G. Yang, Y. Kong, W. Hou, Q. Yan, J. Phys. Chem. B 109 (2005) 1371–1379. T. Vanderah, T. Collins, W. Wong-Ng, R. Roth, L. Farber, J. Alloys Compd. 346 (2002) 116–128. J.-M. Jehng, I. Wachs, Chem. Mater. 3 (1991) 100–107. N. Massa, S. Pagola, R. Carbonio, Phs. Rev. B 53 (1996) 8148–8150. K. Maeda, M. Eguchi, T. Oshima, Angew. Chem. Int. Ed. 53 (2014) 13164–13168. V. Thangadurai, W. Weppner, Chem. Mater. 14 (2002) 1136–1143. Y. Miseki, H. Kato, A. Kudo, Energy Environ. Sci. 2 (2009) 306–314. T. Oshima, M. Eguchi, K. Maeda, Chem. Sus. Chem 9 (2016) 396–402. K. Kamata, K. Maeda, D. Lu, Y. Kako, K. Domen, Chem. Phys. Lett. 470 (2009) 90–94. R. Marschall, Adv. Funct. Mater. 24 (2014) 2421–2440. D. Scaife, Sol. Energy 25 (1980) 41–54. M. Yoshino, M. Kakihana, W. Cho, H. Kato, A. Kudo, Chem. Mater. 14 (2002) 3369–3376.