Effect of Nickel Oxide Doping to Ceria-Supported Gold Catalyst for CO Oxidation and Water-Gas Shift Reactions

Ceria-supported gold catalyst has drew much research interest owing to its high reactivity on CO oxidation and water-gas shift (WGS) reactions. However, till now, there were relatively limited studies on the effect of secondary metal/metal oxide component into gold-ceria system to enhance its catalytic performance. In this work, we synthetized the ceria supported gold-nickel samples via a deposition-precipitation method with the base of NaHCO3 to adjust final pH value of 8~9. We found that the addition of nickel oxide drove off the gold species from the stock solution during synthesis, and thus resulted in a dramatical decrease on doped Au concentration. No crystallized phases of gold and nickel were observed on the surface of ceria nanorods in both X-ray diffraction (XRD) and transmission electron microscopy (TEM). The valence of nickel was maintained as Ni2+ for all the measured samples by X-ray photoelectron spectroscopy (XPS), while gold was oxidized with the increased nickel amount after analysis of X-ray absorption near edge spectroscopy (XANES). The corresponding catalytic tests showed that with the introduction of nickel oxide, the activity of gold-ceria catalyst was promoted for the WGS reaction, but inhibited for the CO oxidation reaction.

Although single metal gold catalyst has been studied profoundly during the last two decades, the research on the bimetallic gold catalyst was still limited [23][24][25], especially for the effect of low-price transition metal or metal oxide doping [24,26,27].Few practical preparation routes have been achieved in the fine control on the formation of bimetallic catalyst at the subnanometer scale [26,27].For example, Scott et al. reported the synthesis and characterization of TiO 2 -supported Pd-Au bimetallic nanoparticle catalysts prepared using dendrimer-encapsulated nanoparticles [23] and Eveline et al. prepared Pt-Au bimetallic cluster supported on the SiO 2 or TiO 2 by wet impregnation to investigated the electronic structure of platinum and gold differed from that in the monometallic clusters [25].However, the size of bimetallic cluster or nanoparticle were larger than 2 nm in the above work.On the other hand, it is extremely difficult to precisely explore the electronic structure and geometric structure of low-concentration active metals.In recent years, combination between aberration-corrected transmission electronic microscopy (TEM) and X-ray adsorption fine structure (XAFS) techniques provide a comprehensive characteristic approach to identify the structural information on small-size bimetallic species anchored on the surface of oxide matrix [11,[28][29][30].Previously, Qiao et al. took the lead in proposing the single atom catalyst that FeO x supported Pt single-atom and characterized by aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and XAFS, revealed that Pt isolated single atoms structure [28], Guo et al. compared different Au species (atom, cluster, particle) and confirmed the metal gold is the active species during CO oxidation with the help of HAADF-STEM and XAFS techniques [21].These investigations confirmed the possibility of obtaining reliable and accurate structural information around gold active species.
Owing to the easily reversible transformation between Ce 3+ and Ce 4+ , nanosized cerium oxide (CeO 2 ) always applied to the reducible oxide support to deposit different metals or oxide [1,22].For example, our group prepared conformably dispersed subnanometer copper [31] and iron [32] oxide clusters on CeO 2 .Our previous findings also indicated that ceria-supported gold catalyst is very active for both CO oxidation [21] and WGS [7] reactions.In addition, some research groups found the synergetic effect of gold and nickel [33,34].For example, Shirakawa et al. reported the gold atoms improve the release of oxygen from nickel atoms of NiO within cluster [33].Thus, in this work, we tried to combine the effect of nickel on gold activity and the structure-activity relationship on WGS and CO oxidation reactions.Herein, in this work, we report the synthesis of nanorod-like ceria-supported gold-nickel catalyst via a deposition-precipitation method with bicarbonate sodium as precipitating agent, as well as the multiple characterizations such as X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), TEM and XAFS towards the comprehensive structural properties of these nickel-doped gold-ceria samples.Furthermore, we elucidate the effect of nickel oxide in catalysis by catalytic tests for both CO oxidation and WGS reactions, in which the introduction of nickel displays totally different performance (positive/negative effect) on the above reactions.

Effect of Nickel Oxide during Deposition of Gold
In this work, the concentration of gold or nickel was designed to be fixed at 1 or 5 at%, respectively.During the catalyst preparation, we investigated three different precipitating agents: ammonium carbonate ((NH 4 ) 2 CO 3 ), sodium carbonate (Na 2 CO 3 ) and sodium hydrogen carbonate (NaHCO 3 ), and found that: (1) (NH 4 ) 2 CO 3 is very suitable for gold deposition onto pure ceria [7], but nickel ion was complexed with molecular ammonia, which is prejudicial to the deposition of nickel; (2) When Na 2 CO 3 is added, the fast growth of nickel phase made the NiO crystallized easily; (3) If NaHCO 3 is used as precipitant, no large-size gold or nickel phase appeared.Therefore, we finally selected sodium hydrogen carbonate as the base to simultaneously precipitate both gold and nickel ions.
Table 1 shows that the experimental loading of gold concentrations (Au), determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES), are 0.21 and 0.13 at% for ceria-supported pure gold (AuCe) and gold-nickel (AuNiCe) samples, respectively.Clearly, there was ca.40% loss of gold species during the applied deposition-precipitation process.Here, we hypothesized that the as-formed Ni(OH) x (H 2 O) y nucleation in stock solution can cover part of ceria surface and drive the Au(OH) x (H 2 O) y nuclei off from the oxide support.Since the purpose of this work is focused on the formation of small-size gold and nickel atoms or clusters, we accepted this gold loss, which could be compensated by the normalization of catalytic reactivity.

Structural and Textural Properties for Ceria-Supported Gold-Nickel Catalysts
The nitrogen adsorption/desorption experiment was used to investigate the textural properties for fresh samples.Table 1 shows that BET specific surface areas of AuCe and AuNiCe identical (99 m 2 /g).The related adsorption/desorption isotherms and pore diameter distribution histograms are shown in Figure 1a,b, respectively.All the adsorption/desorption curves are defined as the type IV isotherm.The hysteresis loop appear in the P/P 0 range of 0.8-1.0associated with AuCe or AuNiCe is type H3, which identifies the materials consisted of aggregated particles [35].Figure 1b shows that AuNiCe appears slightly larger average pore diameter (20.5 nm)/BJH pore volumes (0.53 cm 3 /g) than those for AuCe (18.3 nm, 0.44 cm 3 /g), indicating the addition of nickel may add extra surface layers of ceria nanorods.Nevertheless, all the above data demonstrate that the introduction of Ni actually did not bring major changes for the textural properties of our ceria-supported gold catalyst.
ceria-supported pure gold (AuCe) and gold-nickel (AuNiCe) samples, respectively.Clearly, there was ca.40% loss of gold species during the applied deposition-precipitation process.Here, we hypothesized that the as-formed Ni(OH)x(H2O)y nucleation in stock solution can cover part of ceria surface and drive the Au(OH)x(H2O)y nuclei off from the oxide support.Since the purpose of this work is focused on the formation of small-size gold and nickel atoms or clusters, we accepted this gold loss, which could be compensated by the normalization of catalytic reactivity.

Structural and Textural Properties for Ceria-Supported Gold-Nickel Catalysts
The nitrogen adsorption/desorption experiment was used to investigate the textural properties for fresh samples.Table 1 shows that BET specific surface areas of AuCe and AuNiCe identical (99 m 2 /g).The related adsorption/desorption isotherms and pore diameter distribution histograms are shown in Figure 1a,b, respectively.All the adsorption/desorption curves are defined as the type IV isotherm.The hysteresis loop appear in the P/P 0 range of 0.8-1.0associated with AuCe or AuNiCe is type H3, which identifies the materials consisted of aggregated particles [35].Figure 1b shows that AuNiCe appears slightly larger average pore diameter (20.5 nm)/BJH pore volumes (0.53 cm 3 /g) than those for AuCe (18.3 nm, 0.44 cm 3 /g), indicating the addition of nickel may add extra surface layers of ceria nanorods.Nevertheless, all the above data demonstrate that the introduction of Ni actually did not bring major changes for the textural properties of our ceria-supported gold catalyst.TEM and high resolution transmission microscopy (HRTEM) were used to verify the morphology of ceria supported gold-nickel samples.As shown in Figure 2, CeO2 support is with 20-150 nm length and uniform diameter, and their average widths (8.1-8.7 nm) are nearly the same between AuCe and AuNiCe (see Table 1).Figure 3a,c confirm the fcc structure for CeO 2 in both of the investigated samples, and no lattice fringes of Au or Ni/NiO were detected.Figure 3b,d verify the homogenous dispersions of either Au or Ni on the surface of CeO 2 for both pure and Ni-doped Au/CeO 2 catalysts at the nanometer scale by the aids of scan transmission electron microscopy-energy dispersive spectrometer (STEM-EDS) mappings.Thus, on the basis of microdomain characterizations, we can draw a conclusion that the introduced gold and nickel ions were uniformly converted to deposited species on the surface of CeO 2 nanorods during our synthesis.composed by nanorods with 20-150 nm length and uniform diameter, and their average widths (8.1-8.7 nm) are nearly the same between AuCe and AuNiCe (see Table 1).Figure 3a,c confirm the fcc structure for CeO2 in both of the investigated samples, and no lattice fringes of Au or Ni/NiO were detected.Figure 3b,d verify the homogenous dispersions of either Au or Ni on the surface of CeO2 for both pure and Ni-doped Au/CeO2 catalysts at the nanometer scale by the aids of scan transmission electron microscopy-energy dispersive spectrometer (STEM-EDS) mappings.Thus, on the basis of microdomain characterizations, we can draw a conclusion that the introduced gold and nickel ions were uniformly converted to deposited species on the surface of CeO2 nanorods during our synthesis.

Catalytic Performance of Ceria-Supported Gold-Nickel Catalysts
The catalytic performance of the as-synthetized ceria supported gold-nickel samples was evaluated by both CO oxidation and WGS reactions.Figure 4a,b display the limited activity of gold-free samples (NiCe) for these two reaction, confirming the importance of Au to activate the CO conversion.Figure 4a shows that the introduction of nickel impaired the reactivity of gold-ceria catalyst.Correspondently, Table 2 lists that the temperature of 50% conversion (T50) for CO oxidation is raised from 50 °C (AuCe) to 66 °C (AuNiCe), and the reaction rate normalized by gold amount at 40 °C decreases from 1030 (AuCe) to 631 μmol gAu −1 s −1 (AuNiCe).The higher reaction rate obtained over gold-ceria samples is comparable to that of Au supported on ceria nanorods synthesized by deposition-precipitation with NaOH as precipitating agent under the same reaction conditions (1167 μmol gAu −1 s −1 ) [15].On the other hand, Figure 4b shows the activity in WGS reaction as a function of reaction temperature.Clearly, the addition of nickel significantly promoted the reactivity of gold-ceria catalyst.It is important to highlight that the reaction rate normalized by gold concentration at 250 °C increases from AuCe (781 μmol gAu −1 s −1 ) to AuNiCe (2542 μmol gAu −1 s −1 , see Table 2).Therefore, we can draw a conclusion that the nickel doping indeed has positive effect for the WGS reaction, while negative effect for the CO oxidation reaction in this work.composed by nanorods with 20-150 nm length and uniform diameter, and their average widths (8.1-8.7 nm) are nearly the same between AuCe and AuNiCe (see Table 1).Figure 3a,c confirm the fcc structure for CeO2 in both of the investigated samples, and no lattice fringes of Au or Ni/NiO were detected.Figure 3b,d verify the homogenous dispersions of either Au or Ni on the surface of CeO2 for both pure and Ni-doped Au/CeO2 catalysts at the nanometer scale by the aids of scan transmission electron microscopy-energy dispersive spectrometer (STEM-EDS) mappings.Thus, on the basis of microdomain characterizations, we can draw a conclusion that the introduced gold and nickel ions were uniformly converted to deposited species on the surface of CeO2 nanorods during our synthesis.

Catalytic Performance of Ceria-Supported Gold-Nickel Catalysts
The catalytic performance of the as-synthetized ceria supported gold-nickel samples was evaluated by both CO oxidation and WGS reactions.Figure 4a,b display the limited activity of gold-free samples (NiCe) for these two reaction, confirming the importance of Au to activate the CO conversion.Figure 4a shows that the introduction of nickel impaired the reactivity of gold-ceria catalyst.Correspondently, Table 2 lists that the temperature of 50% conversion (T50) for CO oxidation is raised from 50 °C (AuCe) to 66 °C (AuNiCe), and the reaction rate normalized by gold amount at 40 °C decreases from 1030 (AuCe) to 631 μmol gAu −1 s −1 (AuNiCe).The higher reaction rate obtained over gold-ceria samples is comparable to that of Au supported on ceria nanorods synthesized by deposition-precipitation with NaOH as precipitating agent under the same reaction conditions (1167 μmol gAu −1 s −1 ) [15].On the other hand, Figure 4b shows the activity in WGS reaction as a function of reaction temperature.Clearly, the addition of nickel significantly promoted the reactivity of gold-ceria catalyst.It is important to highlight that the reaction rate normalized by gold concentration at 250 °C increases from AuCe (781 μmol gAu −1 s −1 ) to AuNiCe (2542 μmol gAu −1 s −1 , see Table 2).Therefore, we can draw a conclusion that the nickel doping indeed has positive effect for the WGS reaction, while negative effect for the CO oxidation reaction in this work.

Catalytic Performance of Ceria-Supported Gold-Nickel Catalysts
The catalytic performance of the as-synthetized ceria supported gold-nickel samples was evaluated by both CO oxidation and WGS reactions.Figure 4a,b display the limited activity of gold-free samples (NiCe) for these two reaction, confirming the importance of Au to activate the CO conversion.Figure 4a shows that the introduction of nickel impaired the reactivity of gold-ceria catalyst.Correspondently, Table 2 lists that the temperature of 50% conversion (T 50 ) for CO oxidation is raised from 50 • C (AuCe) to 66 • C (AuNiCe), and the reaction rate normalized by gold amount at 40 • C decreases from 1030 (AuCe) to 631 µmol g Au −1 s −1 (AuNiCe).The higher reaction rate obtained over gold-ceria samples is comparable to that of Au supported on ceria nanorods synthesized by deposition-precipitation with NaOH as precipitating agent under the same reaction conditions (1167 µmol g Au −1 s −1 ) [15].On the other hand, Figure 4b shows the activity in WGS reaction as a function of reaction temperature.Clearly, the addition of nickel significantly promoted the reactivity of gold-ceria catalyst.It is important to highlight that the reaction rate normalized by gold concentration at 250 • C increases from AuCe (781 µmol g Au −1 s −1 ) to AuNiCe (2542 µmol g Au −1 s −1 , see Table 2).
Therefore, we can draw a conclusion that the nickel doping indeed has positive effect for the WGS reaction, while negative effect for the CO oxidation reaction in this work.

Structural and Textural Properties for Used Ceria-Supported Gold-Nickel Catalysts
To better understand the origin of doping effect by nickel over the studied gold-ceria catalyst, we further investigated the related samples by the aids of more characterization techniques.Figure 5 shows that fresh AuCe and AuNiCe are of fcc CeO2 phase (JCPDS card no.: 34-394) with lattice constants a around 5.41 Å and the average grain size of 9.4-9.7 nm (see Table 1).Beyond that, there is no other component including Au or Ni/NiO can be identified in the patterns, which is in good agreement with the TEM results.It is probably due to the low loading of Au and/or small-size atoms or clusters of deposited Au or Ni species over the ceria surface.

Structural and Textural Properties for Used Ceria-Supported Gold-Nickel Catalysts
To better understand the origin of doping effect by nickel over the studied gold-ceria catalyst, we further investigated the related samples by the aids of more characterization techniques.Figure 5 shows that fresh AuCe and AuNiCe are of fcc CeO 2 phase (JCPDS card no.: 34-394) with lattice constants a around 5.41 Å and the average grain size of 9.4-9.7 nm (see Table 1).Beyond that, there is no other component including Au or Ni/NiO can be identified in the patterns, which is in good agreement with the TEM results.It is probably due to the low loading of Au and/or small-size atoms or clusters of deposited Au or Ni species over the ceria surface.

Structural and Textural Properties for Used Ceria-Supported Gold-Nickel Catalysts
To better understand the origin of doping effect by nickel over the studied gold-ceria catalyst, we further investigated the related samples by the aids of more characterization techniques.Figure 5 shows that fresh AuCe and AuNiCe are of fcc CeO2 phase (JCPDS card no.: 34-394) with lattice constants a around 5.41 Å and the average grain size of 9.4-9.7 nm (see Table 1).Beyond that, there is no other component including Au or Ni/NiO can be identified in the patterns, which is in good agreement with the TEM results.It is probably due to the low loading of Au and/or small-size atoms or clusters of deposited Au or Ni species over the ceria surface.Crystal structures of ceria supported gold-nickel samples after both CO oxidation and WGS reactions were also determined by XRD. Figure 5 exhibit that the all the measured catalysts maintained the pure fcc ceria phase as that of fresh catalyst.Still, no patterns of Au or Ni/NiO were detected and the lattice constants (a) of used catalysts are similar to those of fresh (see Table 1), revealing the relatively stable gold and nickel species during these two reactions [36,37] and/or low contents of gold and nickel.
The HRTEM images of used Au/CeO 2 and AuNi/CeO 2 samples in Figure 6a,c,e,g determine the presence of fcc CeO 2 phase in both AuCe and AuNiCe after either CO oxidation or WGS reactions.The corresponding STEM-EDS mapping results in Figure 6d,f,h verify the well dispersion of gold after reaction for the tested AuNi/CeO 2 catalysts, except that ~5 nm Au nanoparticles appeared for AuCe after CO oxidation (Figure 6b), indicating the agglomeration of small-size gold species under CO + O 2 working conditions without the addition of nickel.
Catalysts 2018, 8, x FOR PEER REVIEW 6 of 13 Crystal structures of ceria supported gold-nickel samples after both CO oxidation and WGS reactions were also determined by XRD. Figure 5 exhibit that the all the measured catalysts maintained the pure fcc ceria phase as that of fresh catalyst.Still, no patterns of Au or Ni/NiO were detected and the lattice constants (a) of used catalysts are similar to those of fresh (see Table 1), revealing the relatively stable gold and nickel species during these two reactions [36,37] and/or low contents of gold and nickel.
The HRTEM images of used Au/CeO2 and AuNi/CeO2 samples in Figure 6a,c,e,g determine the presence of fcc CeO2 phase in both AuCe and AuNiCe after either CO oxidation or WGS reactions.The corresponding STEM-EDS mapping results in Figure 6d,f,h verify the well dispersion of gold after reaction for the tested AuNi/CeO2 catalysts, except that ~5 nm Au nanoparticles appeared for AuCe after CO oxidation (Figure 6b), indicating the agglomeration of small-size gold species under CO + O2 working conditions without the addition of nickel.

Electronic Structure of Gold and Nickel for Ceria-Supported Catalysts
X-ray absorption near edge structure (XANES) spectroscopy was carried out to determine the electronic structure of Au in this work.Seen in Figure 7a,b, the oxidation of gold was totally different between AuCe and AuNiCe.We have run linear combination fits, and obtained the average valence states of gold to be +0.3 and +1.7 for fresh AuCe (Figure 7c) and AuNiCe (Figure 7d), respectively.Clearly, with the addition of secondary metal (Ni), the oxidized gold species become dominant on the ceria surface.Figure 7 also depicts the totally different XANES profiles of Au L3 edge for used catalysts from those of fresh samples.The intensity of white-line of AuCe sample slightly oxidized after CO oxidation reaction, but obviously decreased after WGS reaction (Figure 7a).For the AuNiCe sample, the intensity of white-line drastically decreased after both CO oxidation and WGS reactions (Figure 7b).It is worthwhile mentioning that the Au still obtained oxidized after CO oxidation, but almost equal to pure Au foil (Au 0 ) after WGS reaction [38,39], whatever nickel is doped or not.It reveals that the reaction conditions of CO + H2O significantly reduce more fraction of gold species supported on the ceria surface than those of CO + O2.

Electronic Structure of Gold and Nickel for Ceria-Supported Catalysts
X-ray absorption near edge structure (XANES) spectroscopy was carried out to determine the electronic structure of Au in this work.Seen in Figure 7a,b, the oxidation of gold was totally different between AuCe and AuNiCe.We have run linear combination fits, and obtained the average valence states of gold to be +0.3 and +1.7 for fresh AuCe (Figure 7c) and AuNiCe (Figure 7d), respectively.Clearly, with the addition of secondary metal (Ni), the oxidized gold species become dominant on the ceria surface.Figure 7 also depicts the totally different XANES profiles of Au L3 edge for used catalysts from those of fresh samples.The intensity of white-line of AuCe sample slightly oxidized after CO oxidation reaction, but obviously decreased after WGS reaction (Figure 7a).For the AuNiCe sample, the intensity of white-line drastically decreased after both CO oxidation and WGS reactions (Figure 7b).It is worthwhile mentioning that the Au still obtained oxidized after CO oxidation, but almost equal to pure Au foil (Au 0 ) after WGS reaction [38,39], whatever nickel is doped or not.It reveals that the reaction conditions of CO + H 2 O significantly reduce more fraction of gold species supported on the ceria surface than those of CO + O 2 .Owing to the influence of intensive fluorescence of Ce, the Ni edge cannot be measured by XAFS.Here, we applied X-ray photoelectron spectroscopy (XPS) to verify the electronic structure of nickel.Figure 8a exhibits that the unique Ni 2+ species for the nickel-containing samples, whether before or after the catalytic tests.Table 1 displays that the surface concentration of nickel (fresh: 4.5 at%; used in CO oxidation: 3.8 at%; used in WGS: 4.3 at%) is almost identical the bulk concentration of nickel (3.9 at%), manifesting no nickel segregation on the surface of ceria nanorods.Besides, Figure 8a exhibits that the valence of nickel for all the used samples was remained as Ni 2+ after either CO oxidation or WGS reaction.Thus, the nickel species did not change their electronic structure during reaction.In another word, the state of gold is more important to the catalytic reactivity of our AuNi/CeO2 samples.
The temperature-programmed reduction by hydrogen (H2-TPR) was performed to investigate the reducibility and metal-support interaction for the ceria-supported gold-nickel catalysts.Figure 8b exhibits that no observable reduction was found for pure ceria nanorods, and the peaks at 184 and 257 °C for NiCe can be assigned to the reduction of surface oxygen species of CeO2 and Ni δ+ species interacting with the CeO2 [36,40].A broad reduction bands at 118 °C for the AuCe, which can be attributed to the Au δ+ species interacting with CeO2 [7], Remarkably, the reduction peaks for the AuCe was maintained but shifted to higher temperature after the addition of nickel (Figure 8b and Table 2), possibly due to the stronger interaction between nickel-modified gold and ceria support [36].Similarly, the reduction temperatures for nickel-containing species were also much lower in AuNiCe than the gold-free NiCe sample, revealing the interaction between gold and nickel oxide to form Au-O-Ni structure.The H2 consumption values of all the tested samples are listed in Table 2.It must be mentioned that the real consumption of H2 in experiment is significantly higher than those theoretical (only considering Au δ+ → Au 0 and Ni 2+ → Ni 0 ).This could be explained by the partial reduction of ceria support (Ce 4+ → Ce 3+ ) starting around 100 °C, which is Owing to the influence of intensive fluorescence of Ce, the Ni edge cannot be measured by XAFS.Here, we applied X-ray photoelectron spectroscopy (XPS) to verify the electronic structure of nickel.Figure 8a exhibits that the unique Ni 2+ species for the nickel-containing samples, whether before or after the catalytic tests.Table 1 displays that the surface concentration of nickel (fresh: 4.5 at%; used in CO oxidation: 3.8 at%; used in WGS: 4.3 at%) is almost identical the bulk concentration of nickel (3.9 at%), manifesting no nickel segregation on the surface of ceria nanorods.Besides, Figure 8a exhibits that the valence of nickel for all the used samples was remained as Ni 2+ after either CO oxidation or WGS reaction.Thus, the nickel species did not change their electronic structure during reaction.In another word, the state of gold is more important to the catalytic reactivity of our AuNi/CeO 2 samples.
The temperature-programmed reduction by hydrogen (H 2 -TPR) was performed to investigate the reducibility and metal-support interaction for the ceria-supported gold-nickel catalysts.Figure 8b exhibits that no observable reduction was found for pure ceria nanorods, and the peaks at 184 and 257 • C for NiCe can be assigned to the reduction of surface oxygen species of CeO 2 and Ni δ+ species interacting with the CeO 2 [36,40].A broad reduction bands at 118 • C for the AuCe, which can be attributed to the Au δ+ species interacting with CeO 2 [7], Remarkably, the reduction peaks for the AuCe was maintained but shifted to higher temperature after the addition of nickel (Figure 8b and Table 2), possibly due to the stronger interaction between nickel-modified gold and ceria support [36].Similarly, the reduction temperatures for nickel-containing species were also much lower in AuNiCe than the gold-free NiCe sample, revealing the interaction between gold and nickel oxide to form Au-O-Ni structure.The H 2 consumption values of all the tested samples are listed in Table 2.It must be mentioned that the real consumption of H 2 in experiment is significantly higher than those theoretical (only considering Au δ+ → Au 0 and Ni 2+ → Ni 0 ).This could be explained by the partial reduction of ceria support (Ce 4+ → Ce 3+ ) starting around 100 • C, which is activated by either gold or nickel oxide species.On the basis of the above H 2 -TPR results, we can identify the interactions of both metal to support and gold to nickel in our study.
Catalysts 2018, 8, x FOR PEER REVIEW 8 of 13 activated by either gold or nickel oxide species.On the basis of the above H2-TPR results, we can identify the interactions of both metal to support and gold to nickel in our study.

Effect of Nickel Oxide to Gold-Ceria Catalyst for CO Oxidation Reaction
It is generally accepted that Ni deposited on the CeO2 has supremely low even hardly any activity for CO oxidation reaction between 20 and 120 °C [37,40,41], and we also observed the same results during the catalytic tests (NiCe, Figure 4a).Hence, the reactivity of investigated catalysts in this work should be correlated to the structural and textural properties of gold itself.Previously, Deng et.al reported the metallic Au 0 species favor the CO oxidation reaction in Au-Ce-O and Au-Fe-O systems, which was demonstrated by the comparison between different states (oxidation pretreatment and reduction pretreatment) for the catalytic measurements [14].In our study, we found the distinct structural changes of gold when nickel was introduced, and thus the two samples (AuCe and AuNiCe) show obvious differences in activity for CO oxidation process.Combining the XANES results on gold, we confirmed that the fraction of metallic Au decreased, as well as the activity of samples decreased with the addition of nickel, which is in a good agreement with the argument that metallic gold is active species for CO oxidation [21].Here, we emphasized the importance of comparison on normalized reaction rate (r) by gold amount to exclude the variation on the Au concentration in each sample (Table 2).In one word, metallic gold is indeed active for the low-temperature CO oxidation reaction.

Effect of Nickel Oxide to Gold-Ceria Catalyst for Water-Gas Shift Reaction
On the other hand, for the investigated WGS reaction, we found a totally contrast phenomenon towards the CO oxidation reaction.First, NiCe has catalytic reactivity for this reaction, but very limited below 300 °C (Figure 4b).We have determined a clear conclusion on the order of WGS activity: AuNiCe > AuCe > NiCe (Figure 4b), especially for the low-temperature (up to 300 °C) reaction.Here, according to our XANES data (Figure 7b), the ionic gold species, which was promoted by the addition of nickel component, benefit the catalytic performance of ceria supported catalysts for the WGS reaction.Analogously, Fu et al. removed the gold particles with NaCN aqueous solution leaching and remained very fine clusters or atomically dispersed gold which were identified as ionic Au, and such gold species were proved to be very important to WGS reaction [22].Later on, considerable researchers have been devoted to the cationic gold are the active species [7,22].In this work, we designed another approach and prepared two samples with totally different oxidation state of Au by the addition of nickel, and compared the activity of ceria-supported gold catalysts with after the normalization by gold amount.Finally, we disclosed that the normalized reaction rate increased with the decreased fraction of ionic gold in Au-Ni-Ce-O system (Table 2).

Effect of Nickel Oxide to Gold-Ceria Catalyst for CO Oxidation Reaction
It is generally accepted that Ni deposited on the CeO 2 has supremely low even hardly any activity for CO oxidation reaction between 20 and 120 • C [37,40,41], and we also observed the same results during the catalytic tests (NiCe, Figure 4a).Hence, the reactivity of investigated catalysts in this work should be correlated to the structural and textural properties of gold itself.Previously, Deng et.al reported the metallic Au 0 species favor the CO oxidation reaction in Au-Ce-O and Au-Fe-O systems, which was demonstrated by the comparison between different states (oxidation pretreatment and reduction pretreatment) for the catalytic measurements [14].In our study, we found the distinct structural changes of gold when nickel was introduced, and thus the two samples (AuCe and AuNiCe) show obvious differences in activity for CO oxidation process.Combining the XANES results on gold, we confirmed that the fraction of metallic Au decreased, as well as the activity of samples decreased with the addition of nickel, which is in a good agreement with the argument that metallic gold is active species for CO oxidation [21].Here, we emphasized the importance of comparison on normalized reaction rate (r) by gold amount to exclude the variation on the Au concentration in each sample (Table 2).In one word, metallic gold is indeed active for the low-temperature CO oxidation reaction.

Effect of Nickel Oxide to Gold-Ceria Catalyst for Water-Gas Shift Reaction
On the other hand, for the investigated WGS reaction, we found a totally contrast phenomenon towards the CO oxidation reaction.First, NiCe has catalytic reactivity for this reaction, but very limited below 300 • C (Figure 4b).We have determined a clear conclusion on the order of WGS activity: AuNiCe > AuCe > NiCe (Figure 4b), especially for the low-temperature (up to 300 • C) reaction.Here, according to our XANES data (Figure 7b), the ionic gold species, which was promoted by the addition of nickel component, benefit the catalytic performance of ceria supported catalysts for the WGS reaction.Analogously, Fu et al. removed the gold particles with NaCN aqueous solution leaching and remained very fine clusters or atomically dispersed gold which were identified as ionic Au, and such gold species were proved to be very important to WGS reaction [22].Later on, considerable researchers have been devoted to the cationic gold are the active species [7,22].In this work, we designed another approach and prepared two samples with totally different oxidation state of Au by the addition of nickel, and compared the activity of ceria-supported gold catalysts with after the normalization by gold amount.Finally, we disclosed that the normalized reaction rate increased with the decreased fraction of ionic gold in Au-Ni-Ce-O system (Table 2).However, the XANES results on used samples indicate the reduction of ionic Au δ+ species during the WGS reaction, especially for AuNiCe (Figure 7b).It indicates that the ionic gold species were converted to dominantly metallic Au during WGS.However, the initial small-size Au δ+ species interacting with either nickel oxide (Au-O-Ni) or ceria support (Au-O-Ce) effectively kept the active gold atoms/clusters survived after the catalytic measurement, and thus held the high reactivity on WGS, which is well consistent with our previous findings demonstrated by in-situ XAFS results [42].
In this work, there was no special treatment between the end of catalytic tests and the structural characterization of XANES on gold.Since the as-formed metallic Au 0 component from ionic Au δ+ species under WGS reaction conditions is not easy to be reoxidized by air in room temperature (actually 400 • C was required revealed by in-situ XAFS and cyclic H 2 -TPR) previously [42], we believe the used catalysts after these two reactions (considering the reducing potential in CO oxidation is much lower than that in WGS) should largely maintained their structural evolution completed during the catalytic tests.

Raw Materials
All the reagents used in this work were purchased from Sinopharm Chemical Reagent Co., all of which were analytical grade and without purified any more.

Preparation of Ceria Nanorods
The CeO 2 nanorods synthesized by hydrothermal approach [13].Ce(NO 3 ) 3 •6H 2 O (0.05 mol/L) was added into NaOH (6 mol/L) solution (70 mL) with vigorous stirring about 15 min.The suspension was moved into Teflon bottle (inner volume: 100 mL) and then tightly seal in stainless-steel autoclave.Hydrothermal treatment operated at 100 • C for 24 h.Then, the precipitates were washed by Millipore (>18 MΩ) water (ca.1000 mL) four times and alcohol (ca.250 mL) once.The as-washed ceria nanorods were dried in vacuum under 70 • C for 12 h.

Preparation of Gold-Nickel Cluster on Ceria
Ceria supported gold-nickel catalyst were synthesized according to deposition-precipitation (DP) method.The atomic percentage of gold was designed to be 1 at%, the nickel was adjusted to be 5 at.%.The NaHCO 3 (8.4g) and calcined ceria nanorods (1 g) were stirred in 180 mL Millipore (>18 MΩ) water to form a buffering solution with pH value of ~9.HAuCl 4 •4H 2 O and Ni(NO 3 ) 2 •6H 2 O were dissolved in 20 mL Millipore water and then slowly dropped into the above stock solution, and still aged at room temperature for 1 h.The precipitates were separated by centrifugation and washed by Millipore water for three times.The as-washed powders were dried in vacuum at 65 • C for 12 h and further calcined in air at 400 • C for 4 h (heating rate: 2 • C/min).The prepared ceria-supported gold-nickel clusters were denoted as AuCe and AuNiCe for single metal and bimetallic samples, respectively.

Characterizations
The experimental concentration of gold and nickel were detected by inductively coupled plasma atomic emission spectroscopy (ICP-AES) with IRIS Intrepid II XSP instrument (Thermo Electron Co., Waltham, MA, USA).
The nitrogen adsorption-desorption characterizations were analyzed on an ASAP2020-HD88 analyzer (Micromeritics Co., Ltd., Atlanta, GA, USA) at 77 K.The as-calcined powders were degassed at 250 • C under vacuum (<100 mm Hg) over 4 h before the introduction of N 2 .The pore size (r p ) distribution of each sample was calculated on the basis of desorption branch of the tested isotherms with the BJH method.The BET specific surface areas (S BET ) were counted from data between 0.05 and 0.20 in the relative pressure.
The powder X-ray diffraction (XRD) patterns were experimented on Bruker D8 Advance diffractometer (Bruker Co., Billerica, MA, USA) (40 kV, 40 mA) using Cu Kα 1 radiation (λ = 1.5406Å), with a scanning rate of 4 • /min.The 2θ angles were calibrated with a µm-scale alumina (α-Al 2 O 3 ) standard.The diffraction patterns of samples were gathered from 10 to 90 • with 0.02 • /step.The ground sample was planished on a quartz sample holder before each test.Cell dimensions of cubic CeO 2 were obtained from the corresponding powder data with the "LAPOD" software of least-squares refinement by Cohen's method [43,44].
The transmission electron microscopy (TEM), high-resolution transmission electron microscopy (HRTEM) images and scanning transmission electron microscopy-energy dispersive spectrometer (STEM-EDS) elemental mapping results were obtained from a FEI Tecnai G2 F20 microscope (FEI Co., Hillsboro, TX, USA) executing at 200 kV, equipped with a Genesis XM2 accessory (EDAX Inc., Mahwah, NJ, USA).Samples were ultrasonic dispersed in absolute alcohol about 5 min, and then placed a drop of liquid supernatant on an ultra-thin (3-5 nm) carbon film-coated Cu grid.The sample grid was dried spontaneously under dark conditions before inserted into the sample holder.
The X-ray photoelectron spectroscopy (XPS) measurements were performed on an Axis Ultra XPS spectrometer (Kratos Co., Manchester, UK) of Al Kα radiation (225 W).The binding energies were calibrated by C 1s line at 284.8 eV.

X ray Absorption Fine Structure
The X-ray absorption fine structure (XAFS) spectroscopy at the Au L3 (E 0 = 11919 eV) edge was operated at 3.5 GeV in "top-up" mode with a constant current of 260 mA at the BL14W1 beamline [45] of the Shanghai Synchrotron Radiation Facility (SSRF).The XAFS data were collected with a 7-element Ge solid state detector in fluorescence mode.The X-ray energy was calibrated with the absorption edge of pure Au foil.Before the XAFS measurement, we ground the investigated sample into fine powders (smaller than 200 mesh), and then pressed the pure sample without any dilution to a solid pellet.
The profiles of X-ray absorption near edge structure (XANES) part were pretreated and fitted with Athena software.The "normalized absorption" was defined as experimental absorption coefficients as a function of energies µ(E) were processed by background subtraction and normalization procedures.The chemical valence of Au was determined with the linear combination fit [46] by comparison to the corresponding references of Au foil and HAuCl 4 based on the normalized XANES profiles in Athena software.

Catalytic Tests
The reducibility of samples were probed with temperature-programmed reduction by hydrogen (H 2 -TPR) operated in a Builder PCA-1200 instrument (BJBUILDER Optics Co., Ltd., Beijing, China), and the consumption of hydrogenation determined with a thermal conductivity detector (TCD) (BJBUILDER Optics Co., Ltd., Beijing, China).The 30 mg laminar and sieved samples (40-60 mesh) were initially oxidized in pure O 2 at 300 • C for 30 min and cooled naturally to room temperature before reduction.The reduction process was performed with a gas mixture of 5 vol.%H 2 /Ar (Jinan Deyang Co., Jinan, China, 99.997% purity) from room temperature to 500 • C (flowing rate: 30 mL/min; heating rate: 5 • C/min).
The catalytic test for CO oxidation reaction over ceria-supported gold-nickel samples were estimated in a plug flow reactor, in which the thermocouple is placed on the top of the samples and connected with a PID controller (YUDIAN Co., Xiamen, China).The variations of CO and CO 2 in outlets were monitored by a non-dispersive Gasboard 3500 IR spectroscopy (Wuhan Sifang Co., Wuhan, China).The 50 mg of sieved (40-60 mesh) samples were pretreated in air at 300 • C for 30 min before test and cooled down to room temperature in pure N 2 .Then, a gas mixture of 1 vol.%CO/20 vol.% O 2 /N 2 (99.997% purity, Jinan Deyang Co., Jinan, China) passed through the catalysts at a flow rate of 67 mL/min, corresponding to a space velocity of 80,000 mL•h −1 g cat −1 .The "light off" profiles were obtained with the changes of CO conversions from 20 to 150 • C with a ramping rate of 3 • C/min.The whole measurement of CO oxidation reaction took ca.43 min.The CO conversion was calculated with the following equations: CO conversion (%) = (CO in − CO out )/CO in × 100.In the kinetic experiments, the reaction rates were calculated on the basis of data points of less than 20% CO conversions to exclude the mass transfer limitation.The catalytic test of water gas shift (WGS) reaction activities were evaluated in the plug flow reactor, and the thermocouple was placed on the top of the catalysts and connected with a PID controller.All outlet gases were monitored by a Gasboard 3500 IR spectroscopy (Wuhan Sifang Co., Wuhan, China).50 mg of sieved (40-60 mesh) sample was pretreated with 5% H 2 /Ar gas mixture at 300 • C for 30 min.The reactant gas consisting of 2 vol.%CO/N 2 (99.997% purity, Jinan Deyang Co., Jinan, China) was fed to mix with water vapor (CO:H 2 O = 1:5) at a flow rate of 70 cm 3 /min with a total GHSV of 84,000 cm 3 g −1 h −1 .The reactant temperature began from 150 • C to 400 • C under the heating rate 5 • C/min and the catalytic data were recorded at 50 • C per step.The whole measurement of WGS reaction took 7-8 h.The CO conversion of WGS is calculated according to the following equation: CO conversion (%) = (CO in − CO out )/CO in × 100.In the kinetic experiments, the reaction rates were calculated on the basis of data points of less than 20% CO conversions to exclude the mass transfer limitation.

Conclusions
In this work, gold-ceria catalysts doped with nickel were synthesized by the depositionprecipitation method.The use of NaHCO 3 as precipitating agent drove part of the gold species off from the ceria surface.The corresponding catalytic tests demonstrated that the addition of nickel shows positive effect to the WGS reaction, while negative effect to the CO oxidation reaction.On the basis of multiple characterization means including XRD, XPS, XANES and TEM/HRTEM, we verified that the ionic gold may benefit WGS process and the metallic gold may promote the reactivity of gold-ceria catalyst for the CO oxidation reaction.Furthermore, the main contribution of nickel oxide is to tune the electronic structure of active gold species.
a Determined by ICP-AES; b Determined by XPS for fresh, used after CO oxidation and used after WGS, respectively; c From N 2 adsorption/ desorption; d From XRD; e After CO oxidation reaction; f After WGS reaction; g From TEM; h Width; i Length.Numbers in brackets are errors.

Sample Au (at%) a Ni (at%) a SBET (m 2 /g) c Vp (cm 3 /g) c a (Å) d DXRD (nm) d DTEM (nm) g
Determined by ICP-AES; b Determined by XPS for fresh, used after CO oxidation and used after WGS, respectively; c From N2 adsorption/ desorption; d From XRD; e After CO oxidation reaction; f After WGS reaction; g From TEM; h Width; i Length.Numbers in brackets are errors. a

Table 2 .
H 2 reduction temperatures (T R ); H 2 theoretical and real consumption values; temperatures of 50% conversion (T 50 ) and reaction rates (r) of CO oxidation and WGS reactions for AuNi/CeO 2 samples.