Synergy between gold and oxygen vacancies in gold supported on Zr-doped ceria catalysts for the CO oxidation

1  Synergy between gold and oxygen vacancies in gold supported on Zr-doped ceria catalysts for the CO oxidation O.H. Laguna*, A. Pérez, M.A. Centeno, J.A. Odriozola Instituto de Ciencia de Materiales de Sevilla. Calle Américo Vespucio, 49. CP: 41092 – Seville - Spain (*) Corresponding author: [email protected] Abstract The CO oxidation activity of 1 wt.% gold catalysts prepared by depositionprecipitation on a series of ceria doped with Zr supports was studied. The supports (10, 25 and 50 Zr At. %) were synthesized by a pseudo sol-gel method through the thermal decomposition of the corresponding metallic propionates. All the prepared solids were characterized by means of XRF, BET, XRD, Raman spectroscopy, SEM, and H2-TPR. Solid solution was obtained in all mixed systems, while the segregation of different Ce-Zr oxides was observed for the solid with the 50 Zr At. %. The oxygen vacancies population and the amount of easier reducible Ce4+ species in the solids increase with the Zr content. No major textural or structural modifications were detected after gold deposition, although a strong Au-support interaction was generated. Such interaction is strongly influenced by the nucleation of gold deposits on the oxygen vacancies and consequently the amount of Zr inserted in the ceria network also determines the dispersion of gold. The presence of gold eases the surface reduction at lower temperatures, and as higher the amount of Zr in the gold catalysts, higher the CO conversion at low temperatures, probably due to the enhancement of the electronic transfer at the surface of the catalysts. Keywords Ce-Zr; CO oxidation, oxygen vacancies, gold nanoparticles 2  1. Introduction CeO2-ZrO2 mixed oxides has been widely studied because its application as three-way catalyst (TWC) promoters [1]. Apart from this use, materials based on these mixed systems have been reported in H2 production from hydrocarbons [2], sulfur abatement in the fluid catalytic cracking (FCC) process [3], water gas shift reaction (WGS) [4], and as solid oxide fuel cell anodes [5]. The wide range of applications of CeO2-ZrO2 mixed oxides is based on their ability to promote noble metal dispersion, and to enhance oxidation reaction processes by employing lattice oxygen. The last feature is allowed by the rapid reduction/oxidation cycles according to the redox reaction: Ce4+ + e-  Ce 3+. This results in the ability to store oxygen under lean conditions and to release oxygen under rich conditions, which is known as oxygen storage capacity (OSC) [6]. The intrinsic OSC of CeO2 is the key for its application as an ionic conductor. The generation of Ce3+ species in the reduction process is accompanied with the formation of oxygen vacancies in order to maintain the charge equilibrium. These may be located at the bulk or at the surface of the material, presenting differences of only one order of magnitude between the oxygen diffusivity in the bulk and over the surface. Oxygen vacancies are punctual defects that may alter the geometric and electronic structure as well as the chemical properties of the oxide. Kosacki et al. [7] have studied lattice defects in nanocrystalline CeO2 thin films and pointed out that the formation enthalpy of the oxygen vacancies can be diminished by lowering the crystallite size, promoting the formation of Ce3+ species, especially at the surface of the material, since these punctual defects tend to migrate there in order to reduce the free Gibbs energy of the system. Considering the relevance of the oxygen vacancies in the ionic mobility in ceria, several efforts have been devoted to increase the population of these defects and the doping with Zr has demonstrated to allow it [8-11]. The use of Zr as modifier of ceria has been widely reported not only by the improvement of the oxygen mobility but also for the enhancement of the thermal stability. Actually, 3  the thermal stability and the modification of the textural properties have been closely related to the OSC promotion. Besides this, Di Monte et al. [1] drew attention to a number of relevant questions related to Ce-Zr mixed systems that still remain open such as which is the adequate amount of Zr for obtaining an optimal redox and thermal stability. One of the most relevant remarks not usually considered is that the limited solubility of ZrO2 in the CeO2 lattice may produce metastable phases. Thus, several domain-type structures hardly detectable by means of conventional characterization techniques could be obtained during the synthesis. Consequently, the no detection of such inhomogeneities may drive to errors when comparing materials with similar compositions [1]. For that reason, there is not yet a general agreement in the optimal content of Zr for achieving the highest thermal stability or OSC in doped ceria solids. For instance, Trovarelli et al. [12] established Ce0.8Zr0.2O2 as the most stable composition in CeXZr1-XO2 materials. Kaspar et al. [13] suggested that the increase of CeO2 content in CeXZr1-XO2 mixed oxides produces the increase of BET area and solids with a ceria content superior than that of Ce0.4Zr0.6O2 may be are candidates for low to medium temperature catalytic applications, avoiding the phase segregation. On the other hand, Masui et al. [14] presented the synthesis of Ce0.5Zr0.5O2 employing different experimental procedures and the evolution of a cubic pyrochlore-type structure was detected in some cases. This is an interesting result because such materials presents ordered structural oxygen vacancies preferably making them more stable. From these studies, and taking into account the different discussions about the structural modifications in such oxides, the challenge of finding a way to control the oxygen mobility and the thermal stability is open. Despite this, it is clear that the Ce-Zr mixed oxides are good candidates for the deposition of gold nanoparticles, since their oxygen vacancies may act as gold nucleation centers [15-17]. Since Haruta’s pioneering work, it has been widely probed that when depositing gold on different supports as ultra-fine particles its chemistry changes 4  dramatically, becoming extraordinarily active in oxidation reactions at low temperature [18, 19]. Doped ceria oxides have been used for the preparation of gold catalysts looking for small sizes of the gold nanoparticles, and several authors have demonstrated a direct relationship between the oxygen vacancies population in the support and the achieved metallic dispersion of gold [10, 20- 23]. Gold supported on Zr-doped ceria has been reported as active catalysts in different processes, not only oxidation reactions, where the electronic interaction between the support and the gold nanoparticles plays a determinant role. For instance Pojanavaraphan et al. [16] have tested bimetallic Au- Cu/Ce0.75Zr0.25O2 catalysts in the steam reforming of methanol (SRM). The analyzed homogeneous Au-Cu alloys (achieved with 1/3 Au/Cu molar ratio) were active in the SRM but only the catalyst with a 7 wt.% Au-Cu presented a total methanol conversion at 300 ºC. The presence of oxygen vacancies was reported to be relevant not only for the improvement of the dispersion of Au and Cu, but also for the ability to activate oxygen which is crucial for the carbon gasification. The authors hypothesized that in the Au-Cu/Ce0.75Zr0.25O2 catalysts, the coke formation is reduced during reaction, since Cu is able to activate molecular O2 close to Au sites, where the oxygen molecules react with the coke. This results in an important inhibition of the principal deactivation process of the catalyst during the SRM. On the other hand, Perret et al. [24] have synthesized Au/Ce0.62Zr0.38O2 for the selective hydrogenation of benzoic acid in order to produce benzyl alcohol. The easy production of oxygen vacancies by the inclusion of Zr drives to the hydrogenation of an aqueous benzoic acid feed towards benzaldehyde and benzyl alcohol with a high rate because of the activation of the carboxyl function for hydrogen attack by the oxygen vacancies. For a series of Ce-Zr solids tested in the water gas shift reaction, Vindigni et al. [17] observed the following sequence of catalytic activity: AuCe50Zr50 > AuCe80Zr20 > AuCe. The higher catalytic activity showed for the AuCe50Zr50 sample was attributed not only to the higher dispersion of the deposits of gold, 5  but also to its increased acid/base properties. The strong interaction between gold clusters and oxygen vacancies may generate some partially negative charged gold nanoparticles that may participate as active sites in the reaction. Considering the scenario presented above, a wide number of papers devoted to the study of Ce-Zr mixed oxides may be found in literature. Nevertheless the existence of an optimal Ce/Zr molar ratio for achievement of the maximum improvement of the oxygen mobility in Zr-doped ceria is a question that remains unresolved. In this sense, this article aims fist to share different points of view presented in literature and to correlate them through a comparison with our own experimental results. Secondly, the influence of the structural modifications of Zr-doped ceria over the promotion of gold nanoparticles and their applying as CO oxidation catalysts is also discussed. For that purpose, the synthesis of Zrdoped ceria supports, varying the Zr atomic percentage (10, 25 and 50) is proposed. These solids are employed for the preparation by depositionprecipitation of 1 wt.% gold catalysts. The obtained materials are characterized by means of X-ray fluorescence (XRF), N2 adsorption, X-ray diffraction, Raman spectroscopy, scanning electronic microscopy, and H2-TPR. The generated results allows establishing clear trends of the structural modifications and the reducibility of the materials, that will be correlated with the catalytic activity of the solids, evaluated in the CO oxidation reaction. 2. Materials and methods 2.2. Synthesis of the supports A series of Ce-Zr mixed oxides, with different Zr atomic percentages (Zr At.% = 10, 25 and 50 calculated according Eq. 1) were synthesized by modifying a reported pseudo sol-gel method involving the thermal decomposition of the propionates of the respective cations. The adequate amounts of Ce (III) acetate (Aldrich®), and Zr (IV) acetyl acetonate (Aldrich®) were dissolved in propionic acid (Alpha Aesar®) (0.12 M). The propionic acid excess was distilled until obtaining a resin-like solid that was calcined at 500 °C (2 °C/min) for 2 h. Pure CeO2 was synthesized by the same procedure for comparative purposes. 6   100 )()( )( .%   atomsZratomsCe atomsZr ZrAt Eq. 1 Gold catalysts were prepared by deposition–precipitation. The adequate amount of HAuCl4·3H2O to obtain 1% w/w of gold in the support, was dissolved in deionized water (6.0 x 10-4 M) and the pH of the solution was adjusted to 8 by addition of NaOH 0.1 M with an automatic system of measurement and titration (CRISON® pH-Burette 24). This pH value has been reported as adequate for a successful deposition of gold on ceria surfaces [25]. The solution was heated to 70 °C and then the support was added and kept under continuous stirring for 20 minutes. After that, the solid was filtered and washed with distilled water until the disappearance of Cl- and Na+ ions, dried overnight at 100 °C and finally calcined for 2 h at 300 °C. The nomenclature used for the obtained materials indicates the Zr At.% in every case and the presence of gold for the gold catalysts (CeO2, CeZr10, CeZr25 and CeZr50 for the supports and Au/CeO2, Au/CeZr10, Au/CeZr25 and Au/CeZr50). 2.3. Characterization The elemental composition of the samples was determined by X-ray fluorescence (XRF) spectrometry in a Panalytical AXIOS PW4400 sequential spectrophotometer with Rh tube as source of radiation. The measurements were performed onto pressed pellets containing 6 wt.% wax. The textural properties were studied by N2 adsorption/desorption measurements at liquid nitrogen temperature in a Micromeritics ASAP 2010 apparatus. Before analysis, the samples were degassed at 150 ºC for 2 hours under vacuum. 7  Powder X-ray diffraction (XRD) patterns were recorded on a Siemens D500 diffractometer, using Cu K radiation (40 mA, 40 kV), with 0.017° step size and 275 s of step time, over a 2θ range from 10 to 90°. Raman spectra were recorded in a dispersive Horiva Jobin Yvon LabRam HR800 Confocal Raman Microscope with a green laser (532.14 nm) working at 5 mV power and using a 600 grooves/mm grating. The microscope used a 50x objective with a confocal pinhole of 1000 μm. Scanning Electron Microscopy (SEM) analyses of the supports were carried out in a SEM-FEG Hitachi® S4800 instrument equipped with an energy dispersive X-ray analyzer (EDX). Concerning the gold catalysts, these were studied with a Zeiss ULTRA 55 high-resolution FESEM equipped with lens, secondary and back-scattered detectors. Temperature Programmed Reduction (TPR) experiments were carried out in a conventional quartz reactor connected to a TCD. The reactive gas stream (5% H2 in Ar) was flowed at 50 mL/min over 50 mg of sample and the temperature raised at 10 °C/min from room temperature to 900 °C. A molecular sieve 13X was used to retain the produced H2O and CO2. Quantitative analysis was done by integration of the TCD signal and comparison with hydrogen consumption of a CuO reference. 2.4. Catalytic activity For the oxidation of CO, the solids (80 mg with 100 μm <ø<200 μm particle size) were pretreated for 1 h at 300 ºC in a 30 mL/min activation flow of 21vol% O2 balanced in He. The light-off curves (from room temperature to 400 ºC, 5ºC/min) were obtained passing a 42 mL/min feed-stream of 3.4 vol.% CO and 21 vol.% O2 balanced in He. The reaction was carried out in a conventional continuous flow U-shaped quartz reactor working at atmospheric pressure placing the sample between glass wools. The reaction products were followed by mass spectrometry (Balzers® Thermostar) equipped with software Balzers® [26]. 8  This software allows quantification of CO, O2 and CO 2 by continuously monitoring the selected m/z signals (28, 32 and 44, respectively, previously calibrated) as a function of time and related the observed intensity to that of helium (m/z=4) used as an internal standard. 3. Results and discussion The chemical composition and the BET surface areas of the prepared supports and the gold catalysts are presented in table 1. In all solids, supports and catalysts, the measured Zr loading is very close to the intended one. In the same way, the gold loading in the catalysts is close to the intended one of 1 wt %. The N2 adsorption-desorption experiments (not presented) exhibited IV type isotherms with H2 hysteresis in all cases. These isotherms are typical for mesoporous materials with heterogeneous distribution of the mesopores size. The mixed systems exhibited superior surface areas respect to that of the bare CeO2, presenting values around 70 m2/g. This agrees with similar results previously reported for Ce-Zr mixed oxides prepared by coprecipitation [27]. The BET surface area of the Au/CeO2 is similar to that of the support while those catalysts prepared from the mixed systems exhibited a slight decrease of this value respect to the supports (60 vs.70 m2/g). The Au/CeZr50 showed a decreasing of its specific BET area after the deposition of gold, even below than that of the CeZr50 support. The XRD patterns of the synthesized supports and the crystallite sizes calculated with the Scherrer equation are presented in Figure 1. The CeO2 profile corresponds to the c-CeO2 fluorite structure Fm3m (JCPDS 34394) [28] and the mixed oxides exhibit the same profile. Nevertheless, for these materials, a shift of the diffraction lines to higher 2θ values is observed with the Zr loading. Such behavior can be seen more clearly in Figure 1 inset, where the 9  shift of the diffraction line associated to the (111) crystallographic plane of CeO2 is presented. The modification of the position of the XRD lines of CeO2 has been widely described for Ce-Zr mixed systems with different Ce/Zr ratios [27, 29-32] and associated to the solid solution formation since the high solubility of Zr cations in the CeO2 network. A Ce1Zr1-XO2 solid solution (0 < X < 1) is obtained through the isomorphic replacement of Ce4+ cations by Zr4+ into the structure of ceria. Consequently a contraction of the framework would be expected due to the lower ionic radii of the Zr4+ compared to Ce4+ (0.84Å and 0,97Å respectively) [33]. In this sense, the distortion of the cell parameters by the isomorphic substitution results in a modification of the cubic symmetry of the CeO2. Actually the symmetry of the Ce-Zr mixed oxide depends on the Ce/Zr atomic ratio as proposed by Fornasiero et al. [29]. A classification of the crystallographic phase and the space group of Ce-Zr mixed oxides as a function of the Ce At.%, previously proposed by Yashima et al. [34, 35] is presented in Table 2. From this classification, it can be inferred that tetragonal phases are the intermediate structures between the monoclinic and the cubic ones, respectively associated to the bare ZrO2 and CeO2 oxides. The crystallite size values (Figure 1) decrease with the incorporation of Zr until a minimum value obtained for solids with 25 and 50 Zr At.%. Assuming that, smaller the crystallite size greater the exposed area of the crystals, these results agree with the textural ones for the solids doped with 10 and 25 Zr At.%, where higher BET surface areas are observed for the mixed systems compared to that of the CeO2 (Table 1). Nevertheless the CeZr50 did not present a superior specific area than that of the CeO2, while no decreasing of its crystalline size below such of the CeZr25 was observed. The decreasing of the crystallite sizes through the doping with Zr evidences that the solid solution formation stabilizes the CeO2 framework avoiding the sintering during the calcination. The inhibition of the CeO2 crystallite size growth during the calcination step due to the doping, not only with Zr4+ but also with other cations, has been previously evidenced [8, 36, 37]. From these works, it can be inferred 16  The observed interaction of gold and the oxygen vacancies allows considering the possibility that higher the population of oxygen vacancies in the support, higher the gold dispersion in the catalysts. However, the vanishing of the Ov signal after the deposition of gold suggests that the most oxygen vacancies detected were located preferentially at the surface of the supports, and in our case, a bulk to surface oxygen vacancies low ratio may occurring. This is in good agreement with the fact that formation of oxygen vacancies within the bulk is also feasible but less thermodynamically favored that the formation of surface oxygen vacancies. Moreover, despite the possible creation of oxygen vacancies in the bulk, these punctual defects tend to migrate to the surface in order to minimize the Gibbs free energy [53], therefore a homogeneous distribution of these defects along the extension of the material is hardly achievable. In this sense, a superior energetic barrier for the creation of oxygen vacancies at the bulk in CeO2 than in the surface of such material has been recently proposed by Plata et al. [54] employing DFT calculations for a model of a multilayer structure of CeO2 linked to a Yttrium-stabilized zirconia (YSZ), being 2.42 eV for the vacancies at the bulk and 1.72 eV at the surface of the CeO2 phase. The surface character of most of oxygen vacancies which were detected for the Ce-Zr mixed oxides through the Raman spectra has been previously suggested for a similar system (Eu-doped ceria [43]) evaluated under the same experimental conditions that our material. In this work it was considered that using a laser line where the optical absorption exists, the majority of excitation laser and scattering light were adsorbed by the sample. Then, only a portion of the radiation escaped, and the Raman signal got weak, containing information preferentially of the surface of the solids. Representative SEM micrographs of the supports and the average chemical composition determined by EDX from 10 measurements at different zones of the sample are presented in Figure 7. Pure CeO2 exhibited a laminated morphology with layers being composed by wormlike aggregates. Some gaps, that give rise to mesopores and macropores, remain at the junction of the aggregates (Figure 7a). On the other hand, the Ce- 17  Zr mixed systems present a randomly junction of the aggregates, without the layered structure of the bare CeO2. Zr At.% values calculated by EDX analysis agree with XRF ones (see Table 1). It means that there is an acceptable homogeneity of the Zr species along the materials. Despite the good correlation, a slight discrepancy is detected in the case of the CeZr50 solid. This is probably related to the segregation of different Ce-Zr mixed phases, with variable Ce/Zr molar ratios. In order to analyze the dispersion of the gold nanoparticles over the catalysts, a high resolution field emission scanning microscopy (FESEM) study was carried out. Selected micrographs are presented in Figure 8. The main change in morphology is observed in Au/CeO2, where the layered structure of the support is not detected. In addition, this material exhibits bright spots with diameters between 20 and 30 nm, probably associated to Au particles. These bright spots are not observed for the catalysts containing Zr. Since all catalysts present the same loading of gold, the absence of a high density of appreciable bright spots in the materials with Zr (Figures 8b and 8c) must be related with a higher metallic dispersion. However, the low contrast between Au and Ce due to their similar molar weight makes hardly observable by SEM the differences in metallic dispersion between the gold catalysts. This observation agrees with the higher interaction between the gold nanoparticles and the surface of the supports, especially with the oxygen vacancies, evidenced and discussed in the Raman results. Therefore, a higher oxygen vacancies concentration at the surface of the support results in a higher dispersion of the gold nanoparticles. Similar results have been recently presented for similar gold catalysts prepared over doped-ceria supports [8, 55], confirming the role of preferential sites for the nucleation of gold nanoparticles of the oxygen vacancies. The H2-TPR profiles of the bare CeO2 and the mixed oxides are presented in Figure 9. Pure CeO2 presents two main signals that have been widely described for such oxide. The initial occurrence of the low temperature signal (centered at 18  470 °C in our case) is due to the surface reduction of Ce4+ cations into Ce3+, while the high temperature signal (with maximum at 754 °C) is associated to Ce4+ to Ce3+ reduction in the bulk [8, 29, 30, 56, 57]. The high temperature signal also presents a shoulder at 884 °C that may be associated to the reduction of some Ce4+ cations in a deeper position into the bulk less exposed. The presence of different reduction events for the CeO2 at high temperature has been previously described and related with different reductions kinetics in the bulk [29, 58]. Ce-Zr mixed oxides also present two main reduction events at low and high temperatures, only due to the reduction of Ce4+ cations because the Zr4+ ones are not reducible [28]. Nevertheless, the intensity of the low temperature event is considerably higher than that of the high temperature reduction one for all the mixed oxides, which is in opposite way to that observed for the bare CeO2. On the other hand, the maximum intensity of the low temperature reduction process shows a slight shift to higher temperatures with increasing the Zr At.%. This suggests that the doping with Zr increases the population of more easily reducible Ce4+ cations but the required energy for the reduction of the additional Ce4+ cations progressively increases. Since the reduction of Ce4+ results in the creation of new oxygen vacancies, the observed shift of the low temperature peak may be indicative of the enhancement of the energetic barrier for the creation additional oxygen vacancies, as was proved by Krcha et al. [59] in their theoretical study applying DFT+U methods on transition metal-doped CeO2. They demonstrated the decreasing of the energy vacancy formation (Evac) in CeO2 in the nearest neighbor, by doping it with Zr (from 2.76 to 1.63 eV). On the other hand, even after the doping with Zr, the Evac for the creation of new oxygen vacancies in the next nearest neighbor was superior (2.61 eV), which may be the reason for the progressive shift to higher temperatures of the low temperature reduction peak in the TPR profiles of our supports (Figure 9). 19  The improvement of oxygen storage capacity, redox properties and oxygen mobility in ceria by the Ce4+/Zr4+ isomorphous substitution, which is in good agreement with the TPR results presented in Figure 9, has been also reported [14, 60]. Such enhancement of the redox properties in the mixed oxides also correlates with the strong interaction between Ce4+ and Zr4+ cations deduced by the XRD patterns and with the alteration of the symmetry observed through the Raman spectroscopy. An inadequate discussion of some nuances in explaining the results of reducibility may result in controversial discussions about the enhancement or detriment of the reducibility in Ce-Zr mixed oxides depending on the loading of Zr, as was pointed out by Vidal et al. [56]. Nevertheless, most authors claim a beneficial effect in the ceria reducibility by the addition of Zr because of the increasing of the population of surface Ce4+ reducible cations, even if the maximum temperature of the low temperature reduction becomes higher than for the pure CeO2. The TPR profiles of gold catalysts are presented in Figure 10. For the Au/CeO2 solid, two main peaks at low and high temperature are observed as for pure CeO2. However, the low temperature one occurs at a lower temperature and is more intense than the high temperature peak. Considering that gold must be present on the surface of the support in close interaction with the oxygen vacancies, as was demonstrated by Raman, such interaction could be also responsible of the improvement of the redox properties, not only by increasing the population of reducible Ce4+ cations at the surface of the CeO2, but also for carrying out this process at less temperature as will be discussed below. The gold catalysts synthesized with the mixed oxides also present both low and high temperature reduction events. The low temperature signals become broader as the amount of Zr increases, even becoming noticeable the contribution of more than one reduction process. At high temperatures, a broad zone of reduction is observed between 400 and 900 ºC. The area of this high peak clearly decreases by increasing the amount of Zr in the solid, which must 20  be related with the improvement of the reducibility due to the Au-Ce-Zr not only at the surface but also in the bulk. This enhancement of the reducibility of CeO2 and Ce-Zr mixed oxides by the deposition of gold has been reported previously by Fonseca et al. [61]. In their work, the ability of Au and Pt to promote the reducibility of CeO2 and CeZrO4 is compared. Although both noble metals showed similar shifts in the reduction profile of CeO2, they suggested that gold was less effective in the modification of the reduction degree of CeO2, because it only enhanced the reduction of surface and near-surface Ce4+, while Pt allows an easy reduction of both surface and bulk Ce4+. Our results partially agree with this observation because, although the main contribution on the reducibility after the deposition of gold seems to be over the surface Ce4+, the reduction bulk cations seems to be progressively affected also with the increasing amount of Zr. The generation of oxygen vacancies by the inclusion of Zr in the CeO2 framework and the strong interaction of such defects with gold nanoparticles influence features such as the final gold particle size in this type of catalysts, through the enhancement of the electronic transfer between the support and the noble metal clusters allowing a higher dispersion, as proposed from SEM (Figure 8). The electronic transfer between gold and CeO2 surfaces has been previously proposed by DFT calculations [23]. Au may weaken the bond of the surrounding oxygen atoms, making ceria a better oxidant, which in our case may be understood as better reducible material. In terms of the improvement of the reducibility of ceria through the creation of oxygen vacancies, the electronic transferring, promoted between the surface and the gold deposits, may reduce the energy barrier for creating more of such punctual defects at the surface [15]. On the other hand, the presence of several reduction events not only for gold catalysts but also for the supports (Figures 9 and 10) resulted in broad TPR profiles where the deconvolution is hardly achievable, and differentiation between surface or bulk reductions is difficult. Nevertheless, the total amount of 21  H2 consumption during each complete experiment, normalized by the loading of sample or the amount of reducible material, allows establishing a semiquantification of the reducible species population, which is an additional criterion for comparing the reducibility of the prepared materials. For that analysis, the reducibility percentage (RP) was calculated according to Eq. 2, defined as the ratio between the experimental H2 consumption (EHC) measured during the TPR analysis and the theoretical H2 consumption (THC) if all the Ce4+ cations become reduced (Figure 11). THC EHC RP 100   Eq. 2 In the case of gold catalysts, the possible contribution of cationic gold species was not considered since the Ce/Au atomic ratio in any case is close to 100 times. For the supports, the RP increases with the loading of Zr. This agrees with the fact that the structure of the mixed oxide is able to stabilize more Ce3+ cations than pure ceria. Considering that the ionic radii of Ce3+ is higher than those of Ce4+ and Zr4+, the possible expansion produced by the Ce3+ cations is competing with the contraction produced by the inclusion of Zr4+ in the framework of CeO2 [54]. No matter which of the two effects was more important, the electronic distortion is a phenomenon that greatly favors the ionic mobility. This explains the ability of the doped materials to generate and stabilizing more Ce3+ cations. Concerning the gold catalysts, the RP values present the same trend than for the supports. Although a slight superior amount of reducible species was observed, no considerable differences are evident between each support and its respective gold catalyst. According to these results, the principal contribution of gold to the reducibility of the materials is to facilitate the reduction of reducible species existing at lower temperatures. The more easily reducible species must be those located at the surface of the support, considering that 22  this is the place where the interaction between the noble metal clusters and the support is happening. The enhancement of the reducibility is another probe of the interaction between gold and the surface, especially with their oxygen vacancies as was proposed from Raman results. Catalytic activity: CO oxidation reaction The evaluation of the prepared supports and gold catalysts in the oxidation of CO is presented in Figure 12. The supports did not present any CO conversion until temperatures above 225 ºC and the highest performance was observed for the CeZr10 and CeZr25 materials, being their catalytic activity very similar between them. Gold catalysts presented a significant improvement of the CO conversion and total CO conversion was achieved at least at 150 ºC in all cases. Concerning the effect of the doping agent, all the mixed systems exhibited a superior catalytic activity than the Au/CeO2 one at low temperatures. In addition, as the amount of Zr increases, a higher catalytic activity is observed at room temperature (25 ºC). It is clear that in gold catalysts two effects cooperate for the observed improvement of the CO oxidation ability; i) the creation and stabilization of additional oxygen vacancies, and ii) the promotion of the electronic transfer between the gold nanoparticles and the support. The cooperation of these effects could be understood in terms of the dispersion of gold, because, higher the population of oxygen vacancies, higher the dispersion of the noble metal and higher the possibility for such electronic transfer. However, Chen and Goodman [62] have observed that the CO oxidation ability increases until a maximum value by decreasing the particle sizes of the deposits, near to 3 nm for Au-based catalysts for the CO oxidation reaction. Consequently, the enhancement of the oxygen vacancies population with the amount of Zr and the decreasing of the gold nanoparticles sizes for achieving a higher CO conversion at low temperature must be carefully associated, because depending on the size of the metallic clusters, the CO conversion would be drastically altered. 23  In good agreement with our results, the enhancement of the catalytic activity in CO oxidation reactions by doping the structure of CeO2 has been widely evidenced by many authors applying several doping agents: Zn [63], Sm [22], Fe [64], Co, Mn [65], Cu [66, 67] and Eu [43], among others. In these works, the creation of oxygen vacancies by the strong interaction of the CeO2 framework and the doping cation is commonly proposed as a determinant structural effect that facilitates the adsorption and activation of the CO and O2 molecules during the catalytic reaction. Besides this positive effect during the CO oxidation reaction, oxygen vacancies became determinant when gold is deposited on the surface because the strong interaction between them, decreases the required energy for the reduction of the catalyst. The easy reducibility promoted in the gold catalysts can be exploited as high capacity to capture and oxidize CO. On the other hand, the capital relevance of the gold-oxygen vacancy interaction can be observed in other systems also applied for the CO oxidation reaction. For instance, we have recently reported the enhancement of the electronic transfer in gold catalysts prepared on Y-doped TiO2 in the CO oxidation reaction [68]. When gold nanoparticles were placed over oxygen vacancies of the support, a large electronic transfer took place from gold to the gaseous oxygen molecule, activating it for the CO oxidation. The activation of the O2 molecule was produced because the electronic density was transferred to antibonding orbitals, weakening and lengthening the O-O bond. These results are in agreement with a previous study presented by Chen and Goodman [62] which proposed that electron-rich Au particles, whose present a strong interaction with the support, are crucial for activating molecular O2 during the CO oxidation. In a similar way, Boronat and Corma [69] studied the role of Au- and Fe-doped anatase TiO2 in the oxygen activation during the CO oxidation. By means of DFT methods, the authors observed that during the activation of molecular O2 on isolated and supported gold nanoparticles, the activation energy involved in the O2 dissociation was directly related with the degree of electron density transfer from the catalyst to the * molecular orbitals 24  of adsorbed O2, and with the vibration frequency that reflects the strength of the O-O bond [70]. Particularly, the authors remarked the results obtained with the Fe-doped solids because the adsorption of O2 involved the electronic transfer from the oxide to the molecule, which was activated as a superoxide O2- specie, while Fe2+ was reoxidized to Fe3+. Therefore combining the ability for the O2 adsorption and for reducing the activation energy for the creation of oxygen vacancies in the TiO2 lattice, an enhanced oxidation activity would be expected for the Fe-doped TiO2. Finally the authors also suggested that strategy of promoting the oxidation activity by doping with gold the titanium oxide has been employed for the same purpose with ceria. Despite that discussed above, there is not an exclusive mechanism for explaining the CO oxidation reaction. For instance some researchers suggested that the CO oxidation proceeds by a Mars-van Krevelen mechanism [71], and it is widely accepted that factors such as gold particle size, synthesis methods, pretreatment conditions and the nature of support influence the catalytic activity but the CO oxidation are still subject of debate. In particular the role of the support and oxygen supply is controversial as was remarked by Corma et al. [72]. 4. Conclusions The modification of CeO2 with different Zr At.% (10, 25 and 50) was successfully achieved and the oxygen vacancies population and the reducibility increase with the loading Zr. In all cases a solid solution formation was detected. However, in the case of the 50 Zr At.% system the segregation of different oxides, probably metastable tetragonal symmetries with different Ce/Zr atomic ratios, was also detected. Comparing our results with those presented in literature, it can be pointed out that the evolution of the possible structures not only depends on the chemical composition of the oxide but also in the synthesis procedure applied for obtaining the materials (being relevant the precursors of Zr4+ and Ce3+ cations, and the calcination program). 25  No major structural or textural modifications were detected in the solids after the gold deposition, although an electronic interaction between gold nanoparticles and the oxygen vacancies of the supports was observed. Such interaction became stronger as the loading of Zr increased in the solid, leading to an enhancement of the gold dispersion. Gold catalysts did not present a considerable higher population of reducible Ce4+ species than the corresponding supports. However, the electronic interaction established between gold and the support surface resulted in the decreasing of the energetic barrier for achieving the reduction of the surface of the solids, which is pointing out the synergy between the oxygen vacancies and the clusters of such noble metal. Finally, the catalytic activity evaluation in the CO oxidation exhibited wide superior catalytic performance of the gold catalysts compared to the supports. Gold catalysts presented higher CO conversion at temperatures below 150 ºC as the amount of Zr increases. This behaviour may be due to a modified dispersion of gold nanoparticles as the amount of oxygen vacancies population was improved, which is directly related to the Zr loading in the support. The modified dispersion of such noble metal clusters allows the enhancement of the electronic transfer between the catalyst and the reaction environment, and an easier activation of the O2 molecules. 5. Acknowledgments The authors thank the financial support from the Ministerio de Economía y Competitividad Español (ENE2012-374301-C03-01 and ENE2013-47880-C3-2- R) co-financed by FEDER funds from the European Union and from Junta de Andalucía (TEP-8196). In addition the authors thank to Dr. Juan Almagro (ACERINOX S.A. Spain) for his help with high-resolution FESEM experiments. 6. References [1] R. Di Monte, J. Kašpar, Catalysis Today, 100 (2005) 27-35. [2] H.-S. Roh, K.-W. Jun, S.-E. Park, Applied Catalysis A: General, 251 (2003) 275- 283. [3] B. Wen, M.Y. He, C. Costello, Energy Fuels, 16 (2002) 1048-1053. 32  Figure 2   0 1020304050 5.25 5.30 5.35 5.40 5.45 Cell parameter (Å) Zr At.% Y = -0.00255x + 5.40665 r2= 0.99907 33  Figure 3   32 33 34 35 36 46 47 48 49 50 51 c -CeO2( t' o t'' ?) CeO2 (220) (200) Relative intensity (a.u.) 2 CeZr50 ( t' o t'' ?) 34  Figure 4   20 25 30 35 40 45 50 55 60 65 C.Z. = 6.8 nm C.Z. = 6.7 nm C.Z. = 9.1 nm (311) (220) (200) Au/CeZr50 Au/CeZr25 Au/CeZr10 Relative intensity (a.u) 2 Au/CeO2 (111) c -CeO2 C.Z.= 14.8 nm 35  Figure 5   250 300 350 400 450 500 550 600 650 700 750 Ov 600 cm-1 F2g 460 cm-1 CeZr50 CeZr25 CeZr10 Normalized intensity (a.u.) Raman shift (cm-1) CeO2 260 cm-1 0 4x10-2 8x10-2 1x10-1 2x10-1 2x10-1 CeZr50 CeZr25 CeZr10 Ov/F2g ratio (a.u) Solids CeO2 36  Figure 6   200 250 300 350 400 450 500 550 600 650 Au/CeZr50 Au/CeZr25 Normalized intensity (u.a.) Raman shift (cm-1) Au/CeO2 Au/CeZr10 F2g (440 cm-1) 37  Figure 7  38  Figure 8  39  Figure 9   1000 2000 3000 4000 5000 6000 7000 735 °C 884 °C 468 °C 503 °C 495 °C 754 °C Time (s) TCD signal (mV) 470 °C 0.1 mV 100 200 300 400 500 600 700 800 900 CeZr50 CeZr25 CeZr10 Temperature (°C) CeO2 Heating program 40  Figure 10   1000 2000 3000 4000 5000 6000 Au/CeZr50 Au/CeZr25 Au/CeZr10 Time (s) TCD signal (mV) Au/CeO2 0.1 mV 100 200 300 400 500 600 700 800 900 190 ºC 189 ºC 117 ºC 158 ºC 184 ºC 329 ºC 873 ºC 562 ºC 664 ºC 549 ºC 744 ºC Temperature (ºC) 781 ºC Heating program 41  Figure 11 