Pt vs. Au in water-gas shift reaction

1  Pt vs. Au in Water Gas Shift Reaction M. Gonzalez Castaño*, T. Ramirez Reina, S. Ivanova, M. A. Centeno, J. A. Odriozola Departamento de Química Inorgánica e Instituto de Ciencia de Materiales de Sevilla (ICMS), Centro mixto CSIC-Universidad de Sevilla, Avda. AmericoVespucio 49, 41092 Sevilla, Spain Corresponding author email: [email protected] Abstract This work presents a comparison of the gold and platinum based catalysts behavior in the water gas shift (WGS) reaction. The influence of the support, e.g. its composition and electronic properties, studied in details by means of UV-VIS spectroscopy, of the metal nature and dispersion and of the stream composition has been evaluated. The catalytic performance of the samples is directly correlated to the electronic properties modification as a function of metal and/or support. Both metals present high activity in the selected reaction although in a different operation temperature window. Keywords: gold catalyst, platinum catalyst, mixed oxides, Water Gas Shift (WGS) 2  INTRODUCTION Nowadays, steam reforming is the main hydrogen source, especially when the final objective is the generation of pure hydrogen. The hydrogen yields generally increases via water gas shift (WGS) reaction, with the additional value of CO abatement. The WGS reaction (CO+H2O CO2+H2) is a classic industrial process commonly carried out in a series of adiabatic converters where the effluent from the reformer system is treated in two WGS reactors, a high temperature shift (HTS) and low temperature shift (LTS) ones. The industrial HTS catalysts are based on iron-chromium mixed oxides while the LTS catalysts on copper-zinc mixed oxides [1-3]. Since the WGS reaction is an exothermic and equilibrium-limited reaction, in which the H2 production decreases with the increase of the temperature, a LTS reactor is preferred for hydrogen yield rising. In spite of the wide use of the copper-zinc LTS catalysts, this catalytic system still requires specific reduction procedures and safety cautions on air exposure due to its pyrophoricity. These drawbacks are well controlled in the hydrogen production plants but when mobile applications are envisaged, the requirements for the WGS catalysts are quite different from those of the traditional catalysts [4].Thereby, i) a non-pyrophoric low temperature catalyst, ii) an essential reduction of the volume and weight of the reactor and iii) a sufficient withstanding to start-up/shut-down cycles are critical conditions for a successful techno-economic implementation. In response to these needs, noble metals (NM) based catalysts like Pt [1,5] or Au [3,6] are under intense development. Despite of the extended number of studies concerning platinum or gold-based catalysts for the WGS reaction, still at present in the scientific community no clear 3  preference for the use of one or another system exists. Burch [6] summarizes the advantages and drawbacks of gold catalysts in comparison to their platinum analogs for the WGS reaction. According to this review, gold catalysts can offer some advantages in the low temperature range (180-250ºC) where the platinum catalysts are insufficiently active. However, at higher temperatures the reported inherent activity of the gold catalysts is lower than that of properly activated platinum ones [7]. The pretreatment employed for the Pt based catalysts appears to be essential for their performance which is not the case of gold catalysts, which could be used as prepared. It is very difficult to draw a general conclusion about the importance of one or another NM from the data available in the literature based only on the discussion of the catalytic behavior of the catalysts. No matter the metal, a strong influence of the preparation method is generally reported [8,9]. The gold-based catalysts are only active when present in a highly dispersed state, which imposes, for a proper comparison with the platinum based catalysts, the use of a suitable preparation method. Moreover, the choice of the support is also of primordial importance for both metals [10-12]. In general, the most active formulations are commonly based on NMs supported on reducible oxides as CeO2 [13-15]. The enhanced activity of the CeO2 supported NM catalysts is frequently attributed to the ceria ability to undergo fast storage/release of oxygen cycles via a reversible conversion between Ce4+ and Ce3+ oxidation species [16, 17]. These redox cycles implies a mechanism of oxygen vacancies formation that is indirectly correlated with the catalyst ability to activate oxygen species through electronic transfer and hence to the catalytic activity in oxidation reactions. CeO2 lattice modifications, like the incorporation of trivalent metals ions, allow the improvement of the oxygen exchange ability and, thus the catalytic performances of the catalytic systems. The incorporation of metallic promoters, as for example Fe or Zr, was recently 4  reported to successfully modify the CeO2 electronic band structure thus increasing its reducibility and WGS activity [18,19]. A meaningful comparison for the reported catalytic activity in the WGS reaction of Au and Pt catalysts is difficult since the authors proposed different catalyst formulations that are tested in a variety of feedstock compositions and reaction conditions. For example, Trovarelli et al. [20] have compared Au and Pt supported on zirconia-promoted ceria concluding that the synergism between the precious metal and the oxide can be affected by the chosen synthesis parameters and structural and morphological properties of the support. On these bases, the aim of this paper is to compare the catalytic behavior of gold and platinum catalysts supported on ceria modified materials in the WGS reaction, correlating its activity to the nature of the support and of the NM. In order to avoid more variables, the same support and metal loading has been kept for all the catalysts. The preparation methods were chosen in order to obtain comparable metal particle size and dispersion during the reaction. The catalytic activity and stability have been measured in both model and simulated realistic post reforming WGS streams. EXPERIMENTAL: - Catalyst synthesis: γ-alumina powder (Sasol) has been employed as a main component of the evaluated materials. The supports (CeO2/Al2O3 and Ce0.8Fe0.2/Al2O3) used in this study have been prepared by incipient wetness impregnation using CeNO3·6H2O (Aldrich) and Fe(NO3)3·9H2O (Aldrich) as oxides precursors. Supports were designed to present 15 wt.% of CeO2, 2 wt.% of FeOx and 80-85 wt.% of γ-Al2O3. The impregnation was carried out in 50 ml ethanolic solution and evaporated at 50ºC at reduced pressure in a 5  rotatory evaporator till a dry solid was obtained. The solid was then treated with NH3 solution (10 mol L-1) for 30 min in order to assure the full conversion of the nitrates to hydroxides. The supports were then filtered, dried and finally calcined at 450ºC for 4h. Platinum based catalysts were obtained via aqueous impregnation using tetrammonium nitrate platinate (Johnson Matthey) as Pt precursor at reduced pressure in a rotatory evaporator at 80ºC. The as obtained solids were then dried and calcined at 350ºC for 8h at a heating rate of 5ºC/min. The gold based catalysts were prepared by the direct anionic exchange method, assisted by NH3 using HAuCl4 (Alfa Aesar) as gold precursor [21]. The calcination was carried out in the same conditions than for the platinum based ones but for 4h. The NM total loading was set to 2 wt%. As mentioned above, the preparation methods have been intentionally selected in order to achieve comparable precious metal particle sizes and dispersions. In the adopted nomenclature the oxygen and the stoichiometric coefficients into the mixed oxide were omitted for simplicity. For example, Pt (2%)/Ce0.8Fe0.2/Al2O3 is named as Pt/CeFeAl. - Catalyst characterization: X-ray microfluorescence spectrometry (XRMF) was used to determine the chemical compositions and the analysis was performed in an EDAX Eagle III spectrophotometer with a rhodium source of radiation. The textural properties were studied by N2 adsorption-desorption measurements at liquid nitrogen temperature. The experiments were carried out on a Micrometrics Tristar 6  II instrument. Before the analysis, the samples were degassed for 2h at 250ºC in vacuum. X-ray diffraction (XRD) analysis was carried out on a X`Pert Pro PANalytical instrument. Diffraction patterns were recorded using Cu Kα radiation (40 mA, 45kV) over a 2θ-range of 10 to 95º and a position-sensitive detector using a step size of 0.05º and a step time of 240 s. The Raman spectra were recorded in a dispersive Horiva Jobin Yvon LabRam HR800 spectrometer, with a 20 mW He–Ne green laser (532.14 nm) operating 5 mW. The microscope used a 50x objective and a confocal pinhole of 1000 µm. The UV-Vis spectra were recorder on a Varian spectroscopy model Cary 100, equipped with an integrating sphere using BaSO4. All spectra were collected in diffuse reflectance mode and transformed to a magnitude proportional to the extinction through the Kubelka-Munk function F(α). CO chemisorption measurements were performed on fresh and post-reacted platinum samples by using a Micromeritics Autochem II apparatus. Prior to chemisorption, the samples were heated with a ramping rate of 10ºC/min from ambient temperature to 350ºC and treated with flowing hydrogen (10% H2 in Helium) for 2 h at this temperature. The samples were then cooled to room temperature under vacuum. It was assumed that each surface platinum atom was one CO chemisorption site. Gold dispersion was estimated through High-Angle Annular Dark Field-Scanning Transmission Electron Microscopy (HAADF-STEM). The images were recorded on a JEOL2010F instrument. The HAADF-STEM studies were performed using an electron probe of 0.5 nm diameter and a diffraction camera length of 10 cm. It should be pointed that the chemical composition of the sample was studied in STEM mode using an 7  Energy-dispersive X-ray spectrometer (Oxford Instrument, Inca Energy-200). Loose powder samples were supported on a holey carbon-coated copper grid without using any liquid. The excess of sample was removed from the grids using a flow of N2. This approach allows us to obtain electron transparent thin regions. Catalytic activity The water gas shift reaction was carried out placing 1g of catalyst in a tubular flow reactor at atmospheric pressure in the 180-350 ºC temperature range using two reaction schemes named model and realistic conditions. Pt based catalysts were activated during 2h at 350 ºC in a H2 stream (10% vol.) while a pretreatment was not necessary for gold based samples. For the model WGS reaction just water and CO diluted in nitrogen were used while for the realistic WGS reaction a reformate gas surrogate containing hydrogen and carbon dioxide was selected. The experimental mixtures and conditions are summarized in the Table 1. The model reaction mixture is the one usually employed for testing catalysts for the clean H2 production via WGS. The conditions used to test the catalysts using the reformate gas surrogate simulate the outlet of a typical ethanol reformer and they are known as CO clean-up conditions. All the solids were sieved and the 600-800 μm fraction retained. The CO and CO2 content was analyzed with an ABB gas analyzer and the activity expressed in terms of CO conversion. RESULTS AND DISCUSSION The textural properties of the studied systems are summarized in Table 2. All the samples are mesoporous materials with textural properties mainly governed by the presence of the primary γ-Al2O3 support. Regarding the supports, a decrease on the specific surface area, pore volume and pore size was observed when ceria is added to alumina. This effect was enhanced with the inclusion of Fe as ceria doping agent, 8  leading to materials with smaller pores and lower specific surface areas than those of the CeAl sample. As reported elsewhere for gold catalysts, the NM addition increases the surface area in all the samples respect to their corresponding supports. The increase of the surface area is related to the observed increase in the pore volume. This effect has been previously explicated by Somorjai et al. [22] claiming that the inclusion of gold nanoparticles in the pore structure would lead to a pore widening and therefore to an increase in the specific surface area. On the contrary, the opposite effect is observed for the platinum based catalyst pointing to a slight blocking of the pores by the presence of platinum species. The chemical composition of the solids is presented in Table 3. All the samples show chemical compositions close to the target values. Nevertheless, in the case of bare alumina supports, a metal loading below the target one was obtained irrespectively of the NM (Au or Pt) pointing to a weak interaction between the γ-Al2O3 support and the metals. The presence of ceria and ceria mixed oxides leads to stronger metal-support contact, resulting in an enhanced metal loading. In any case, the obtained metal loadings and the target ones are closer for platinum than for gold and this must be related to the different synthetic procedure. Gold deposition was carried out under a high basic media. According to Moreau and Bond, the gold uptake depends on several factors such as pH, temperature and specific surface area of the support [23]. Working at pH above the isoelectric point of the solid (pH ~ 7.2 in our ceria containing samples) involves a change of charge in the support becoming increasingly more negative. This results in an electrostatic repulsion of gold containing anions limiting their deposition. 9  XRD diffraction patterns of the employed supports are shown in Fig. 1a. All the supports present the characteristic diffractions of the cubic CeO2 fluorite structure (JCPDS#00-004-0593) and the γ-Al2O3 phase (JCDPS# 00-048-0267). However, no diffraction peaks of any iron-containing crystalline phases were detected either for the small amount of the doping agent or for the possible inclusion of these cations in the ceria lattice resulting in a solid solution formation. Indeed, all the ceria diffraction peaks were shifted toward higher angles when iron was added suggesting the formation of CeO2-FeOx solid solution. The CeO2 lattice parameter was then estimated using the main (111) CeO2 diffraction line (where no contribution of the γ-alumina phase is observed), the λ factor of Cu filament (1.5405 Å) and the expression that relates the lattice parameter of the fluorite structure with the diffraction angle: The calculated parameter for the CeAl sample resulted to be 5.407 Å, which is very close to the reported value for a pure ceria powder sample in ref. [18]. The corresponding a parameter for the CeFeAl sample was 5.395 Å. The formation of a substitutional Ce-Fe solid solution must result in a lattice contraction for ceria since Ce4+ (0.97 Å ionic radius) ions are bigger than Fe3+(0.64Å ionic radius) ions. Recently, for a similar mixed oxide system, the formation of Ce-Fe solid solutions together with the detection of amorphous Fe2O3 segregated phases was described [24]. Actually, interstitial and substitutional solid solutions can be obtained when ceria and iron oxides are mixed together. The formation either of interstitial or substitutional solid solutions depends on the relative amount of each oxide component [18,25]. In our case the presence of certain amount of segregated iron species cannot be totally discarded since the small amount of iron limits its detection by XRD. As a secondary effect of the 16  formation of CO2 and H2. The thermal decomposition of this carboxyl species was proposed as responsible for CO2 evolution on supported gold catalysts as the simpler and more probable mechanism [53]. More recently, Rodriguez et al. [54] propose also carbonate or bicarbonate species as intermediates in the low temperature WGS. The main difference between the tested NM may reside on water dissociation. Thus, while CO activation only occurs on the metal particle, water dissociation may take place either on the support, as for gold catalysts, or in the support and the metal particle as in the case of platinum catalysts. This effect can be clearly seen for alumina-supported NMs, the activity for the platinum catalyst increases with temperature reaching the equilibrium conversion at ca. 300ºC; on the contrary the gold catalyst present a very low conversion (~10% CO conversion) hardly dependent on temperature. DFT calculations have found that water dissociation is highly endothermic on Au(111) and slightly endothermic on Pt(111) [55]. Therefore, in the absence of a support able of activating water dissociation only, the Pt catalysts will be active in the WGS. For that reason, the influence of the support nature in the WGS activity of gold catalysts might be far more important than for platinum based ones [8,10]. The modification of the alumina support alters the concentration of oxygen vacancies and hence, reduces the band gap and increases the Lewis basicity of the support. The catalytic activities of the gold catalysts follow parallel trends for the different supports, in such a way that for all the temperature range the CO conversions for the Au/CeFeAl and Au/CeAl catalysts are around 8,6±0,8 and 4,7±0,6 times that of the Au/Al catalyst respectively. Therefore on increasing the concentration of oxygen vacancies, the band gap decreases and the basicity increases favoring water dissociation and hence the CO conversion rate. 17  As mentioned above the existence of oxygen vacancies is not required for reaching the equilibrium conversion in the case of platinum catalysts. However, a detailed analysis evidences that the cerium-containing catalysts reach the equilibrium conversion at lower temperature (230ºC) than the un-doped Pt/Al catalyst. According to this, it might be stated that the presence of a support with enhanced redox properties is also required for achieving good performances with Pt systems. It must be noticed however that the support ability for water dissociation is the same for Pt or Au catalysts and therefore the differences in catalytic activity must be associated to the CO adsorption properties on the metal surface. The support band gap is the energy difference between the O(2p) band and the localized Ce(4f) states. The measured band gap in the NM supported catalysts, smaller than the corresponding to the pure support, accounts for the difference in energy between the NM 5d states and the 4f levels of cerium [27]. The catalytic activity of the samples is a function of this band gap, in such a way that the higher the band gap the higher the activity. This behavior can be explained in terms of separation between energy levels, showing that the higher catalytic activity is obtained when the energy levels introduced by the NM are closer to the O(2p) ones that means, when the easiness of the electron transfer between oxygen atoms of ceria and the NM is higher and therefore this excess electron density weakens the CO bond strength through backbonding and facilitates the formation of the carboxyl intermediate. For a proper comparison of the catalytic data, WGS specific reaction rates (expressed in molCOconv.gmetal-1.s-1) at two different temperatures have been calculated for the prepared materials, Table 5. Besides, in the Table 6 the turnover frequencies (TOFs) are presented in two temperatures. Two values are reported for Pt/CeFeAl, for fresh and post-reaction catalysts as a sinterization occurs during the reaction. For 18  Au/CeFeAl no agglomeration is observed and only one TOF value is reported for both fresh and post-reaction samples. Particle sizes calculated from TEM data and/or CO chemisorption were used to estimate NM dispersion. TEM data allows establishing a gold dispersion of 32%, corresponding to an average particle size of 4 nm, figure 4. CO chemisorption experiments led to a Pt dispersion of 70% for the platinum fresh catalyst corresponding to Pt nanoparticles of 2.2 nm, which is close to the average particle size estimated by TEM experiments (figure 4b). As stated in a previous work [56], the average particle size of gold nanoparticles after the WGS reaction remains while platinum nanoparticles sinter upon reaction resulting in 28% dispersion which accounts for an average particle size of 5.8 nm. Platinum sintering during the WGS reaction was also observed by Farrauto’s group [8] reporting values for pre- and postreaction particle sizes quite similar to those reported in this work. Once the platinum catalyst is stabilized, the particle size and metal dispersion of Au and Pt nanoparticles are quite similar and so, any effect on the differences in activity between both NM associated to particle size and/or dispersion may be discarded. Several points should be considered regarding the obtained results. First, gold based systems exhibit superior catalytic activity in terms of rates than the platinum ones in the low temperature range. However, the tendency changes at high temperature being more active the platinum based catalysts. Second, all the ceria containing solids presented higher activities compared to Au/Al and Pt/Al. The iron to ceria promotion is also evidenced by the increase of the reaction rates for both metals observed for the iron loaded solids once compared with their un-promoted parent catalysts. Confirming the specific rate trend, the TOF values for the catalysts at 180ºC show the superiority of the gold sample in this temperature range. However, at 250oC the platinum based catalysts doubled the TOF obtained for the gold system with a similar dispersion, thus converting 19  platinum in better choice in this temperature range. The evaluated gold-based catalysts presented higher CO rates and turnover frequencies at 180ºC than some well referenced Au/CeO2 and Au/CeO2-MOx systems tested in the same conditions [57,58]. Regarding the Pt materials, the specific reaction rates of our Pt based materials are superior compared to the results obtained by the group of Efstathiou using a similar WGS inlet stream [59]. As a general remark, no matter the support, the gold based catalysts behave always better at the lowest measured temperature (180ºC), and the platinum based catalysts takes the advantage at temperatures above 200ºC. Taking into account all the information presented above, the study of the behavior of the catalysts in the reformate gas surrogate is mandatory. Thus, the most active Au and Pt catalysts, those supported on the CeFeAl material, were selected for the catalytic evaluation in a stream with the composition shown in table 1. The catalytic performances of both samples are presented in Fig. 5b. Both activity curves show a shift to higher temperatures with respect to those obtained under model conditions, as expected on the bases of Le Chatelier’s principle. However, it must be pointed out that the gold-based catalyst does not reach the equilibrium conversion in the whole range of studied temperatures. As proposed by Rodriguez et al. [60] the redox mechanism with a carboxyl intermediate can explain the behavior of Au/CeO2based catalysts. In this mechanism, it is crucial to achieve some ceria reduction degree to promote the water splitting into the ceria oxygen vacancies. Recent in-situ XANES studies revealed that in a Au/CeO2-FeOx/Al2O3 system, the highest WGS performance was achieved for a Ce formal oxidation state of 3.3 [56]. However, in order to close the catalytic cycle, water has to be able to reoxidize the 20  partially reduced support in the studied temperature range. This redox behavior is easily maintained in the model mixture due to the softer reductive conditions. Nevertheless, the ability of water to reoxidize the ceria-iron support in the presence of a high amount of hydrogen is limited decreasing remarkably the reaction rate. Regarding the catalytic results presented in Table 6 in a realistic mixture, a general decrease of the TOF values compared with the model mixture is obtained indicating the negative effect of the presence of CO2 and H2 in the mixture. Again gold seems better at 180ºC as in the model mixture. However at temperatures higher than 250ºC, platinum showed not only higher rate but also better tolerance to the composition of the stream (same TOF in model and realistic mixture). Conclusions The catalytic activity of the samples is a function of the supports electronic properties and more precisely on the samples band gap, in such a way that the higher the band gap the higher the activity. The higher catalytic activity is obtained when the energy levels introduced by the NM are closer to the O(2p) ones that means, when the easiness of the electron transfer between oxygen atoms of ceria and the NM is higher. For all the systems (gold or platinum) the presence of ceria or mixed CeFe oxide is beneficial for the WGS activity, being the gold based systems more influenced by the electronic properties of the support. No matter the stream composition, the gold based catalysts present superior activity than the platinum ones at 180ºC, thus shifting its operation window to the lowest temperature range. Nevertheless, the platinum based systems are much more 21  versatile since they are able to withstand changes in the stream composition with higher TOF than its gold homologues and attained the equilibrium conversion at temperatures close to 250 ºC. Acknowledgments Financial support for this work has been obtained from the Spanish Ministerio de Ciencia e Innovación (ENE2012-374301-C03-01) co-financed by FEDER funds from the European Union and from Junta de Andalucía (TEP-8196). S. Ivanova acknowledges MEC for her Ramon y Cajal contract. T.R. Reina acknowledges CSIC for his JAE-Predoc fellowship and all the authors acknowledge Junta de Andalucía—‘‘TEP 106 group’’. Sasol is gratefully acknowledged for supplying alumina. References [1] C. Ratnasamy, J. Wagner, Catal. Rev. 51 (2009) 325-440. [2] D. Andreeva, V. Idakiev, T. Tabakova, A. Andreev, R. Giovanoli, Appl. Catal. A 134 (1996) 275-283. [3] T. Tabakova, V. Idakiev, D. Andreeva, I. Mitov, Appl. Catal. A 202 (2000) 91-97. [4] D.L. Trimm, Appl. Catal. A 296 (2005) 1-11. [5] R.J. Farrauto, Y. Liu, W. Ruettinger, O. Ilinich, L. Shore, T. Giroux, Catal. Rev. 49 (2007) 141–196. [6] R. Burch, Phys. Chem. Chem. Phys. 47 (2006) 5483-5500. [7] A.A. Phatak, N. Koryabkina, S. Rai, J.L. Ratts, W. Ruettinger, R.J. Farrauto, Catal. Today 123 (2007) 224-34. [8] W. Ruettinger, X. Liu, R.J. Farrauto, Appl. Catal. B 65 (2006) 135-141. 22  [9] G. Jacobs, S. Ricote, P.M. Patterson, U.M. Graham, A. Dozier, S. Khalid, E. Rhodus, B.H. Davis, Appl. Catal. A 292 (2005) 229-243. [10] T. Tabakova, L. Ilieva, I. Ivanov, R. Zanella, J.W. Sobczak, W. Lisowski, Z. Kaszkur, D. Andreeva, Appl. Catal. B 136-137 (2013) 70-80. [11] S. Ivanova, V. Pitchon, C. Petit, J. Mol. Catal. A: Chem. 256 (2006) 278–283. [12] C.M. Kalamaras, I.D. Gonzalez, R.M. Navarro, J.L.G. Fierro, A.M. Efstathiou, J. Phys. Chem. C 115 (2011) 11595-11610. [13] Q. Fu, H. Saltsburg, M. Flytzani-Stephanopolous, Science, 310 (2003) 935-938. [14] D. Andreeva, V. Idakiev, T. Tabakova, L. Ilieva, P. Falaras A. Bourlinos, A. Travlos, Catal. Today 72 (2002) 51–57. [15] G. Jacobs, P.M. Patterson, L. Williams, E. Chenu, D. Sparks, G. Thomas, B.H. Davis, Appl. Catal. A 262 (2004) 177-187. [16] T. R. Reina, S. Ivanova, V. Idakiev, J.J. Delgado, I. Ivanov, T. Tabakova, M.A. Centeno, J.A. Odriozola, Catal. Sci. Technol. 3 (2013) 779-787. [17] A. Holmgren, B. Andersson, D. Duprez, Appl. Catal. B 22 (1998) 215-230. [18] O.H. Laguna, F. Romero-Sarria, M.A. Centeno, J.A. Odriozola, J. Catal. 276 (2010) 360-370. [19] P. Panagiotopoulou, J. Papavasiliou, G. Avgouropoulos, T. Ioannides, D.I. Kondarides, Chem. Eng. J. 134 (2007) 16-22. [20] M. Boaro, M. Vicario, J. Llorca, C. de Leitenburg, G. Dolcetti, A. Trovarelli, Appl. Catal. B 88 (2009) 272–282. [21] S. Ivanova, C. Petit, V. Pitchon, Appl. Catal. A 267 (2004) 191-201. [22] Z. Konya, V.F. Puntes, I. Kiricsi, J. Zhu, J.W. Ager, M.K. Ko, H. Frei, P., Alivisatos, G.A. Somorjai, Chem. Mater. 15 (2003) 1242-1248. [23] F. Moreau, G.C. Bond, Appl. Catal. A 302 (2006) 110–117. 23  [24] F.J. Pérez-Alonso, M.L. Granados, M. Ojeda, P. Terreros, S. Rojas, T. Herranz, J. R. Gancedo, Chem. Mater. 17(9) (2005) 2329-2339. [25] S. Damyanova, B. Pawelec, K. Arishtirova, M. Huerta, J.L.G. Fierro, Appl. Catal. A 337 (2008) 86-96. [26] A.S. Reddy, C.Y. Chen, C.C. Chen, S.H. Chen, C.J. Lin, K.H. Lin, C.L. Chen, S.C. Chang, J. Mol. Cat. A: Chem. 318 (2010) 60-67. [27] M.A. Centeno, C. Portales, I. Carriozosa, J.A. Odriozola, Catal. Lett. 102 (2005) 289-297. [28] C.W.M. Castleton, J. Kullgren, K. Hermansson, J. Chem. Phys. 127 (2007) 244704 1-11. [29] V. Petrovsky, B.P. Gorman, H.U. Anderson, T. Petrovsky, J. Appl. Phys. 90 (2001) 2517-2521. [30] F. Marabelli, P. Wachter, Phys. Rev. B 36 (1987) 1238-1243. [31] P. Patsalas, S. Logothetidis, L. Sygellou, S. Kennou, Phys. Rev. B 68 (2003) 351041-13. [32] S. Deshpande, S. Patil, S. Kuchibhatla, S. Seal, App. Phys. Lett. 87 (2005) 133113 1-3. [33] S. Tsunekawa, T. Fukuda, A. Kasuya, Surf. Sci. 457 (2000) L437-L440. [34] L. Wu, H.J. Wiesmann, A.R. Moodenbaugh, Phys. Rev. B 69 (2004) 125415 1-9. [35] O.H. Laguna, M.A. Centeno, M. Boutonnet, J.A. Odriozola, Appl. Catal. B 106 (2011) 621-629. [36] T. Dhannia, S. Jayalekshmi, M.C.S. Kumar, T.P. Rao, A.C. Bose, J. Phys. Chem. Solids 71 (2010) 1020–1025. [37] P.C.A. Brito, D.A.A. Santos, J.G.S. Duque, M.A. Macedo, Physica B 405 (2010) 1821–1825. 24  [38] L. Yue, X.M. Zhang, J. Alloy Compd. 475 (2009) 702–705. [39] W.Y. Hernández, M.A. Centeno, F. Romero-Sarria, J.A. Odriozola, J. Phys. Chem. C 113 (2009) 5629–5635. [40] O.H. Laguna, M.A. Centeno, G. Arzamendi, L.M. Gandía, F. Romero-Sarria, J.A. Odriozola, Catal. Today 157 (2010) 155–159. [41] J.E. Spanier, R.D. Robinson, F. Zhang, S.W. Chan, I.P. Herman, Phys. Rev. B 64 (2001) 2454071–2454078. [42] G.W. Graham, W.H. Weber, C.R. Peters, R. Usmen, J. Catal. 130 (1991) 310-313. [43] Z.V. Popovic, Z.D. Dohcevic-Mitrovic, N. Paunovic, M. Radovic , Phys. Rev. B 85 (2012) 014302 1-6. [44] M.A. Centeno, M. Paulis, M. Montes, J. A. Odriozola, Appl. Catal. A 234 (2002) 65-78. [45] C.T. Campbell, C.H. Peden, Science 309 (2005) 713-714. [46] C. Zhang, A. Michaelides, D.A. King, S.J. Jenkins, Phys. Rev. B 79 (2009) 075433 1-11. [47] C. Zhang, A. Michaelides, D.A. King, S.J. Jenkins, J. Chem. Phys. 129 (2008) 194708 1-8. [48] Z. Yang, Z. Lu, G. Luo, Phys. Rev. B 76 (2007) 075421 1-7. [49] A. Martínez-Arias, J.M. Coronado, R. Cataluña, J.C. Conesa, J. Soria, J. Phys. Chem. B 102 (1998) 4357-4365. [50] A.A. Gokhale, J.A. Dumesic, M.M. Mavrikakis, J. Am. Chem. Soc. 130 (2008) 1402-1414. [51] F. Romero-Sarria, A. Penkova, L.M. Martínez,M.A. Centeno, K. Hadjiivanov, J.A. Odriozola, Appl. Catal. B 84 (2008) 119–124. 25  [52] L. C. Grabow, A.A. Gokhale, S.T. Evans, J.A. Dumesic, M. Mavrikakis, J. Phys. Chem. C, 112 (2008) 4608-4617. [53] G. Bond, Gold Bull 42 (2009) 337-342. [54] J.A. Rodriguez, J.C. Hanson, D. Stacchiola, S.D. Senanayake, Phys. Chem. Chem. Phys. 15 (2013) 12004-12025. [55] A.A. Phatak, W.N. Delgass, F.H. Ribeiro, W.F. Schneider, J. Phys. Chem. C 113 (2009) 7269–7276. [56] T.R. Reina, W. Xu, S. Ivanova, M.A. Centeno, J. Hanson, J.A. Rodriguez, J.A. Odriozola, Catal. Today 205 (2013) 41-48. [57] D. Andreeva, I. Ivanov, L. Ilieva, M.V. Abrashev, Appl. Catal. A 302 (2006) 127- 132. [58] Z.Y. Yuan, V. Idakiev, A. Vantomme, T. Tabakova, T.Z. Ren, B.L. Su, Catal. Today 131 (2008) 203-210. [59] K.C. Petallidou, A.M. Efstathiou, Appl. Catal. B 140–141 (2013) 333–347. [60] J.A. Rodriguez, P. Liu, J. Hrbek, J. Evans, M. Pérez, Angew. Chem. Int. Edit. 46 (2007) 1329–1332.