Since the discovery of the surprisingly high activity of Au nanoparticles (NPs) in low temperature CO oxidation and acetylene hydrochlorination in the 1980s [1, 2], gold has no longer been regarded as an inert noble metal in catalytic reactions. With further "gold rush" research, it has been widely acknowledged that the size and support sensitivities are the intrinsic characteristics of gold nanocatalysts [3-11], while also the barriers for their practical application. Au NPs can exhibit excellent catalytic activity over a certain particle size range (2–5 nm) and when supported on "active" supports [6, 7]. Thus, stabilizing supported gold NPs in their "active" sizes and constructing a synergistic effect between the gold NPs and "active" support are necessary to exert these advantages. Recently, we have reported that MgGa₂O₄ spinel can be used to stabilize Au NPs within a particle size of 2–5 nm at temperatures above the melting point of bulk gold (1064 ℃) [12], which solves the long-term problem of gold nanocatalysts from the aspect of their thermal stability. Unfortunately, the MgGa₂O₄ support is an irreducible support, which limits the application of these supported Au NPs in redox reactions.

The strong dependence of gold nanocatalysts on the support is mainly due to the presence of a synergistic catalytic effect between the Au NPs and supports. For example, the water-gas shift reaction (WGSR: CO + H₂O = CO₂ + H₂), which is an important reaction for tuning the CO/H₂ ratio in syngas and upgrading H₂, prefers reducible supports or alkali metal ions, such as CeO₂ and Fe₂O₃ or Na⁺ and K⁺, respectively. This results in the intimate contact between the support and Au NPs in which CO activation occurs on the Au NPs and O‒H bond cleavage occurs on the reducible support or alkali metal ion [13-16]. Catalytic combustion reactions, such as CO oxidation and methane combustion, also prefer Au NPs supported on reducible supports because the Au NPs are efficient toward CO/CH₄ activation, but do not dissociate molecular O₂ [17-20]. In order to realize the activation of Au NPs on an inert support, reducible metal oxides have been widely used as additives. Tao et al. [21] have reviewed that reducible metal oxides as promoters can adjust the redox properties of the support and create surface defects/oxygen vacancies, which benefit the dissociation of certain reactants. Among them, CeO₂ is an excellent reducible support, which shows high oxygen mobility and storage capacity [22]. It has been reported that oxygen vacancies can be created in the ceria lattice at moderate temperatures and under reductive atmospheres, and these oxygen vacancies can provide active sites for H₂O or O₂ dissociation [23, 24]. Reina et al. [25] have designed highly efficient Au/Al₂O₃ WGSR catalysts by introducing CeO₂ as a promoter in the water activation step and MO_x (M = Fe, Cu, Zn) as dopants. Similarly, the introduction of redox supports can also improve the activity of Au NPs in catalytic combustion reactions. Gluhoi et al. [17] and Zhang et al. [26] have reported that CeO₂ can significantly improve the activity of Au/Al₂O₃ for CO oxidation, and also proved that CeO₂ acts as an oxygen supplier via the Mars and van Krevelen mechanism.

Herein, we present an example of modulating the activity of anti-sintering Au₲MgGa₂O₄ catalyst for use in the WGSR and catalytic combustion reactions by incorporating CeO₂ as a promoter. Characterizations using HAADF-STEM, X-ray diffraction (XRD), Energy dispersive X-ray spectroscopy (EDS) mapping, Hydrogen temperature-programmed reduction (H₂-TPR), and X-ray photoelectron spectroscopy (XPS) proved that after impregnating CeO₂ on the severely aged Au₲MgGa₂O₄ catalyst, the MgGa₂O₄ stabilizes the Au NPs intimate contact with CeO₂, which results in a strong synergistic effect. This stable matrix shows significantly improved activity in both the WGSR and catalytic combustion reactions. In-situ Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTs) was also used to clarify the possible promotion mechanism of CeO₂ using the WGSR as an example.

The magnesium gallate spinel support precursor was prepared via the homogeneous co-precipitation of stoichiometric magnesium and gallium nitrate with urea in methanol, as previously reported [12].

The Au₲MgGa₂O₄ catalyst was prepared by soaking 8 g of MgGa₂O₄ powder in 150 mL of an aqueous solution of chloroauric acid (HAuCl₄·4H₂O, Au content > 47.8%, Tianjin Fengchuan Chemical Reagent Technologies Co., Ltd) with nominal 2% weight loadings and stirred for 24 h. The suspension was then filtered and washed with concentrated ammonia (25%–28%, Tianjin Damao Chemical Reagent Factory). The filter cake was dried at 80 ℃, and then calcined in ambient air at 300 or 800 ℃ for 5 h at a heating rate of 5 ℃·min^–1. The samples were denoted as Au₲MgGa₂O₄-T-5h, where T is the thermal treatment temperature used. The as-obtained catalysts have Au weight loadings of 1.09 wt%, as-determined using ICP-AES analysis.

The CeO₂-modified Au₲MgGa₂O₄ catalyst was prepared via an incipient wetness impregnation process using the Au₲MgGa₂O₄-800℃-5h catalyst and an aqueous solution of cerium nitrate hexahydrate (Ce(NO₃)₃·6H₂O, > 99.0%, Sinopharm Chemical Reagent Co., Ltd). The impregnated sample was dried at 80 ℃ and then calcined at 400 ℃ for 5 h. The resulting catalyst was donated as CeO₂/[Au₲MgGa₂O₄-800℃-5h] and has a CeO₂ weight loading of 20.7 wt%, as-determined using ICP-AES analysis. A Au/CeO₂-RRCe-2 benchmark catalyst with a gold weight loading of 0.98 wt% was purchased from Haruta Gold Incorporated, which was calcined at 300 ℃ for 2 h and used as the reference.

The actual loadings of Au and CeO₂ were determined using inductively coupled plasma atomic emission spectrometry (ICP-AES) on an IRIS intrepid Ⅱ XSP instrument (Thermo Electron Corporation). Prior to the measurements, all the samples were dissolved using aqua regia.

XRD patterns were recorded on a PANalytical PW3040/60 X′ Pert PRO diffractometer equipped with a Cu K_α radiation source (λ = 0.15432 nm) operated at 40 kV and 40 mA.

High-resolution transmission electron microscopy (TEM) analysis was performed on a JEOL JEM-2100F operated at 200 keV with a specified point-to-point resolution of 0.19 nm and a lattice resolution of 0.10 nm. This instrument was equipped with a high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) detector and the resolution was 0.2 nm. EDS was recorded on an Oxford Instruments ISIS/INCA EDS system equipped with an Oxford Pentafet ultrathin window (UTW) detector in HAADF-STEM mode. To obtain the particle size distribution, we analyzed no less than 100 particles.

H₂-TPR experiment was performed on an automated absorption apparatus TP-5080 equipped with a thermal conductivity detector (TCD) (Tianjin Xianquan Co. Ltd). About 50 mg of the catalyst was loaded into a quartz reactor and pretreated at 400 ℃ in air at a flow rate of 30 mL·min^–1 for 30 min. After cooling to room temperature under a He atmosphere, the flow gas was switched to 5 vol% H₂/N₂ at a flow rate of 30 mL min^–1. Finally, the catalyst was heated to 900 ℃ at a ramping rate of 10 ℃·min^–1 and the TCD signals were recorded.

DRIFTs was recorded on a Bruker EQUINOX 55 spectrometer equipped with an MCT detector and operated at a resolution of 4 cm^‒1 for 32 scans. Before each experiment, the sample (~40 mg) was calcined in situ in air at 400 ℃ in a DRIFTs cell (HC-500, Pike technologies). Then under a flow of He, the temperature was cooled to 200 ℃ and held at this temperature for 30 min. Subsequently, a background spectrum was recorded for the sample, which was then subtracted automatically from the subsequent spectra. The corresponding gas for H₂O adsorption and dissociation (10 vol% H₂O/He) was introduced into the reaction cell and the spectra recorded as a function of time until saturation. The gas was then switched to He to purge the gaseous H₂O molecules and 2 vol% CO was introduced into the reaction cell to detect the reaction between CO and the adsorbed H₂O. The spectra were recorded as a function of time until no obvious changes were observed. The flow rate involved in these steps was 20 mL·min^–1.

XPS data were collected on a Thermofisher ESCALAB 250 spectrometer using a monochromatized Al K_α X-ray source (1486.6 eV) and applied power of 150 W. The C 1s peak (284.8 eV) was used as the reference.

All the reactions were carried out at atmospheric pressure in a fixed-bed micro-reactor. The catalyst powders (~50 mg) were diluted with 0.5 g quartz sand and placed in a U-shaped quartz reactor. A k-type thermocouple in a thin quartz tube was inserted into the catalyst bed to measure and control the temperature. The reactants and products were analyzed using an on-line gas chromatograph (Agilent 7890 B) equipped with a 5A molecular sieve and Parapak Q packed column, and a thermal conductivity detector. The feed gas used for the WGSR was 2 vol% CO + 10 vol% H₂O balanced with He. The feed gases used for CO oxidation and CH₄ combustion were 1 vol% CO + 20 vol% O₂ balanced with He and 1 vol% CH₄ + 20 vol% O₂ balanced with N₂, respectively. The flow rate used was 30 mL min^–1, which results in a space velocity of 36000 mL g_cat.^–1 h^–1. The conversion of CO was calculated using Conv._(CO) = C_CO2/(C_CO + C_CO2) × 100%, in which C_CO2 and C_CO are the concentration of CO₂ and CO in the effluent gas, respectively. The conversion of CH₄ was calculated using Conv._(CH4) = (C_{CH4, in} – C_{CH4, out})/C_{CH4, in} × 100%, in which C_{CH4, in} and C_{CH4, out} are the concentration of CH₄ in the feed and effluent gases, respectively.

HAADF-STEM and XRD were used to estimate the dispersion and thermal stability of the Au NPs before and after the addition of CeO₂ as a promoter. Fig. 1 shows the HAADF-STEM images obtained for the Au₲MgGa₂O₄-300℃-5h, Au₲MgGa₂O₄-800℃-5h, and CeO₂/[Au₲MgGa₂O₄-800℃-5h] samples. The fresh Au₲MgGa₂O₄-300℃-5h sample is made up of highly dispersed Au particles with high particle density (Fig. 1(A)). Most of the Au particles were below 2 nm with some in the range of 3–5 nm, resulting in a mean particle size (D_m) of 1.6 nm and a standard deviation of 0.6 nm. This indicates that the Au NPs could be readily dispersed on the MgGa₂O₄ support via the simple impregnation method. After the aging process, the resulting Au₲MgGa₂O₄-800℃-5h also displayed small Au NPs (~3.1 nm with a standard deviation of 1.0 nm) and a few larger Au particles in the range of 20–30 nm (Fig. 1(B) and inset). This indicates that most of the well-dispersed Au NPs were stabilized by the MgGa₂O₄ support, as previously reported [12]. After further modification with ~20% CeO₂, CeO₂/[Au₲MgGa₂O₄-800℃-5h] is still made up of lots of small Au NPs (~3.1 nm with a standard deviation of 1.6 nm) and well-dispersed CeO₂ NPs (~5 nm) on the MgGa₂O₄ surface (Fig. 1(C)). These results indicate that the addition of CeO₂ did not influence the thermal stability of the Au NPs. Furthermore, CeO₂ was well-dispersed on the MgGa₂O₄ support, which may form an interface with the Au NPs.

XRD was also carried out to estimate the dispersion and crystallinity of the Au and CeO₂ in Au₲MgGa₂O₄-300℃-5h, Au₲MgGa₂O₄-800℃-5h, and CeO₂/[Au₲MgGa₂O₄-800℃-5h] with the MgGa₂O₄ and CeO₂/MgGa₂O₄ supports used as the reference samples (Fig. 2). Accordingly, only the distinct peaks of MgGa₂O₄ spinel (PDF#00-010-0113) were observed in the fresh Au₲MgGa₂O₄-300℃-5h sample, indicating that Au was readily dispersed below the detection limit of XRD via the simple wetness impregnation method. After the aging process of the Au₲MgGa₂O₄-300℃-5h sample at 800 ℃ for 5 h, the distinct peaks of Au (PDF#00-004-0784) appeared and the corresponding particle sizes of the sintered Au NPs were estimated to be 19.9 nm according to the Scherrer equation, which correspond to the large Au NPs observed in the HAADF-STEM images (Fig. 1(B)). Meanwhile, the peaks of the MgGa₂O₄ support become much sharper and the crystal sizes of MgGa₂O₄ estimated by Scherrer equation increased from ~5 to ~12 nm, which indicates that sintering of the MgGa₂O₄ support occurred during the aging process at 800 ℃ for 5 h. The distinct peaks of Au still exist in the CeO₂/[Au₲MgGa₂O₄-800℃-5h] sample and the particle sizes of the sintered Au crystals remained at ~19 nm. Both the HAADF-STEM and XRD results illustrate that the Au NPs in the Au₲MgGa₂O₄-800℃-5h and CeO₂/[Au₲MgGa₂O₄-800℃-5h] samples exhibit bimodal size distributions with small-sized particles (~3 nm) and large-sized branches (~20 nm), which is in accordance with our previous report [12]. Meanwhile, the Au particle sizes were not influenced by the addition of the CeO₂ promoter. For the CeO₂/[Au₲MgGa₂O₄-800℃-5h] catalyst, the distinct peaks of CeO₂ (PDF#00-034-0394) appeared and the corresponding particle size of the CeO₂ crystals was estimated to be 6.0 nm, which is similar to that observed in CeO₂/MgGa₂O₄ (6.5 nm).

Due to the heavy Ce atoms, the contrast observed in the HAADF-STEM image shown in Fig. 1(C) is not good enough to distinguish CeO₂ from the MgGa₂O₄ support and Au NPs. Thus, EDS analysis was performed on the CeO₂/[Au₲MgGa₂O₄-800℃-5h] catalyst to confirm whether there was geometrically intimate contact between the Au NPs and CeO₂. As shown in Fig. 3(A) and (B), the signals of the Au NPs were surrounded by the signals of Ce, which indicate that the Au NPs are in contact with the CeO₂ promoter geometrically over the MgGa₂O₄ support. Fig. 3(C) shows the elemental analysis of the spots with (a) and without (b) visible Au particles. All of the Mg, Ga, Ce, and Au elements are present in both spots with varied intensities (Fig. 3(D)), which further confirms the close proximity between CeO₂ and the Au NPs. The geometrically intimate contact between the Au NPs and CeO₂ may induce a synergistic effect to enhance the catalytic performance, which will be discussed in the following section.

H₂-TPR was used to detect the reducibility of the CeO₂/[Au₲MgGa₂O₄-800℃-5h] catalyst with CeO₂/MgGa₂O₄ and Au₲MgGa₂O₄-800℃-5h as the reference samples. No obvious reduction peaks were detected over the Au₲MgGa₂O₄-800℃-5h catalyst below 500 ℃, which suggests that the Au NPs remain in their metallic state and is consistent with the fact that AuO_x species are not stable [27, 28]. For CeO₂/MgGa₂O₄, two major peaks (T_H = 423 ℃, T_L = 357 ℃) were observed in the detected region (30–900 ℃), which can be attributed to the different surface capping oxygen atoms (O²⁻ or O⁻ anions) in CeO₂ [29, 30]. It is noteworthy that the corresponding reduction peaks for CeO₂ at T_H and T_L in the CeO₂/[Au₲MgGa₂O₄-800℃-5h] catalyst were significantly decreased to 205 and 156 ℃, respectively. It was also observed that the reduction temperatures of CeO₂ in CeO₂/[Au₲MgGa₂O₄-800℃-5h] decreased upon increasing the Au loading (Fig. S1). The obvious decrease in the reduction temperatures observed for CeO₂ in CeO₂/[Au₲MgGa₂O₄-800℃-5h] may be attributed to the strong interactions formed between the Au NPs and CeO₂. A good synergistic effect between the Au NPs and CeO₂ is attractive for catalysis.

Furthermore, the electronic states of Au/(Ce)/O/Mg/Ga in the Au₲MgGa₂O₄-800℃-5h and CeO₂/[Au₲MgGa₂O₄-800℃-5h] samples were detected using XPS. As shown in Fig. 5(A), the Au 4f_7/2 XPS spectra of Au₲MgGa₂O₄-800℃-5h shows a binding energy peak at ~83.0 eV, which can be assigned to the metallic Au NPs. After adding the CeO₂ promoter, the signals for both Au 4f and Mg 2s became much weaker due to the coverage of CeO₂, and the binding energy for Au 4f_7/2 shifted to 84.5 eV (Au⁺) [31]. This indicates that there are strong interactions between Au and CeO₂, which in turn confirms that the Au NPs are in close contact with the CeO₂ promoter. In regard to the electronic states of Ce in the CeO₂/[Au₲MgGa₂O₄-800℃-5h] sample, there are four doublets observed in the Ce 3d XPS spectra (Fig. 5(B)). In accordance with the literature [32], the peaks labeled 'u' correspond to the 3d_3/2 spin-orbital states and those labeled 'v' correspond to the 3d_5/2 states. The u'''/v''', u′/v′ doublets are the primary photoemission from Ce(Ⅳ) and Ce(Ⅲ), and the u"/v", u/v doublets are shakedown features resulting from the transfer of one or two electrons from a filled O 2p orbital to an empty Ce 4f orbital. Thus, the appearance of both u'''/v''' and u′/v′ doublets indicates the coexistence of Ce(Ⅳ) and Ce(Ⅲ) in the CeO₂/[Au₲MgGa₂O₄-800℃-5h] sample. The existence of Ce(Ⅲ) can be used as an indicator for oxygen vacancies, which are widely acknowledged to be the activation sites for water or oxygen molecules in the WGSR and combustion reactions [33]. In addition, the O 1s signals of the Au₲MgGa₂O₄-800℃-5h and CeO₂/[Au₲MgGa₂O₄-800℃-5h] samples were also collected (Fig. 5(C)). For the Au₲MgGa₂O₄-800℃-5h sample, there are two kinds of O species with binding energy peaks at 530.6 and 532.2 eV, which can be assigned to the lattice O and surface under-coordinated O atoms in the spinel support [34]. In addition to the O species corresponding to MgGa₂O₄, two signals with binding energy peaks at 528.9 and 529.5 eV were observed in the CeO₂/[Au₲MgGa₂O₄-800℃-5h] sample, which can be ascribed to lattice and surface O atoms in the CeO₂ promoter. It is noteworthy that the binding energies of O1s for the CeO₂ promoter in the CeO₂/[Au₲MgGa₂O₄-800℃-5h] sample were similar to or lower than that of CeO₂ with abundant oxygen vacancies reported in the literature [35], which, when combined with the appearance of Ce(Ⅲ), confirms that oxygen vacancies exist in the CeO₂/[Au₲MgGa₂O₄-800℃-5h] sample. Mg and Ga were both detected to be Mg²⁺ and Ga³⁺ in the Au₲MgGa₂O₄-800℃-5h and CeO₂/[Au₲MgGa₂O₄-800℃-5h] samples, indicating that the addition of CeO₂ little effect on the electronic structure of the MgGa₂O₄ spinel support (Fig. 5(D) and (E)) [36, 37].

The catalytic performance of the as-synthesized catalysts were investigated using the WGSR and catalytic combustion (methane combustion and CO oxidation) reactions. A benchmark catalyst, Au/CeO₂-RRCe-2 with a Au loading of 0.98 wt% purchased from the Haruta Gold Incorporated, was used as a reference for comparison.

Fig. 6A presents the catalytic performance of the catalysts at various temperatures in the WGSR. Over the Au₲MgGa₂O₄-300℃-5h catalyst, the CO conversions increased from 3.2%, to 5.4%, 8.8%, and 17.9% upon increasing the temperatures from 250 to 350, 450, and 550 ℃, respectively, while Au₲MgGa₂O₄-800 ℃-5h was much less active. This can be ascribed to the sintering of MgGa₂O₄ because the XRD and HAADF-STEM results (Figs. 1 and 2) show that most of the Au NPs remain stable. These activities were inferior to those observed over the Au/CeO₂-RRCe-2 reference catalyst, which were 8.8%, 12.2%, 23.2%, and 63.9%, respectively. These are consistent with the reported results, which show that a reducible oxide support is favorable for the WGSR [13, 14]. Upon the addition of CeO₂ to the low activity Au₲MgGa₂O₄-800℃-5h catalyst, the catalytic activity of the resulting CeO₂/[Au₲MgGa₂O₄-800℃-5h] was significantly enhanced with CO conversions of 5.1%, 19.3%, 50.0%, and 69.6% at 250, 350, 450, and 550 ℃, respectively, demonstrating the benefit of the intimate contact between the stabilized Au NPs and CeO₂. Interestingly, when compared with the Au/CeO₂-RRCe-2 reference catalyst, the CeO₂/[Au₲MgGa₂O₄-800℃-5h] was also more active at higher temperatures. These observations are consistent with the support effect of Au nanocatalysts in the WGSR [13]. The excellent synergistic effect between the Au NPs and CeO₂ occurs due to the formation of CeO₂/Au perimeters via appropriately intimate contact with each other in geometry and the promotion of the reducibility of CeO₂, which were confirmed by the EDS analysis and H₂-TPR results shown in Figs. 3 and 4, respectively. Thus, the stabilized Au NPs on MgGa₂O₄ spinel support retain their catalytic functions and can be readily ignited once in contact with an appropriate oxide promoter for a specific reaction. It is noteworthy that the Au NPs in Au₲MgGa₂O₄ subjected to calcination at 800 ℃ with CeO₂ can display higher catalytic activity than that of the fresh benchmark Au/CeO₂-RRCe-2 catalyst, clearly showing the potential of Au₲MgGa₂O₄ as a thermally stable ensemble in the rational construction of highly active gold catalysts. In addition, we also investigated the long-term catalytic stability of CeO₂/[Au₲MgGa₂O₄-800℃-5h] in the WGSR at 450 ℃ (Fig. S2). During the 48 h test, CeO₂/[Au₲MgGa₂O₄-800℃-5h] showed a deactivation period of ~5 h and then a stable plateau at least for 40 h. The average CO conversion during the plateau observed for the CeO₂/[Au₲MgGa₂O₄-800℃-5h] catalyst was 34.5%. XRD, HAADF-STEM, and EDS-mapping were used to characterize the spent catalyst. Both the XRD and HAADF-STEM results (Figs. S3 and S4) show that the particle size of the Au NPs remained almost unchanged after long-term reaction, indicating that the Au NPs were rather stable over the MgGa₂O₄ spinel support. Furthermore, the Au NPs are still in contact with the CeO₂, as reflected in the EDS-mapping images (Fig. S4(B)). In regard to the CeO₂ promoter, the corresponding crystal size detected using XRD was slightly increased from 6.0 to 7.0 nm after the long-term reaction. Thus, the deactivation observed during the initial 5 h may be attributed to the slight increase in the crystal size of the CeO₂.

The catalysts were also investigated in the CO oxidation and CH₄ combustion reactions in which they showed various activities, but similar evolution trends (Fig. 6(B) and (C)). For CO oxidation (Fig. 6(B)), the fresh Au₲MgGa₂O₄-300℃-5h catalyst provides good catalytic activity in which the reaction temperature for 50% CO conversion (T₅₀) was ~110 ℃. For the aged catalyst, the Au₲MgGa₂O₄-800℃-5h becomes much less active, whose T₅₀ increases to ~230 ℃. However, upon introducing CeO₂ into the aged catalyst, the resulting CeO₂/[Au₲MgGa₂O₄-800℃-5h] catalyst presents enhanced activity and the T₅₀ decreases to ~130 ℃. Notably, the CeO₂/[Au₲MgGa₂O₄-800℃-5h] catalyst appears more active than the fresh Au₲MgGa₂O₄-300℃-5h catalyst at temperatures below 80 ℃. The Au/CeO₂-RRCe-2 benchmark catalyst is more active than all the Au₲MgGa₂O₄ catalysts with a T₅₀ of ~83 ℃, and represents a highly active Au nanocatalyst in this reaction. Fig. 5(C) shows the catalytic performance of the catalysts observed during the combustion of methane. Unlike catalyzing the CO oxidation reaction at low temperature, meaningful CH₄ conversions occur at temperatures over 400 ℃. The T₅₀ for the fresh Au₲MgGa₂O₄-300℃-5h catalyst was ~614 ℃ and that for Au₲MgGa₂O₄-800℃-5h increased to 648 ℃. The Au/CeO₂-RRCe-2 benchmark catalyst was also slightly more active than these Au₲MgGa₂O₄ catalysts with a T₅₀ of 598 ℃. However, the CeO₂/[Au₲MgGa₂O₄-800℃-5h] catalyst has a T₅₀ of 570 ℃ and was more active than Au₲MgGa₂O₄-300℃-5h and Au₲MgGa₂O₄-800℃-5h as well as Au/CeO₂-RRCe-2. It is obvious that CeO₂ can also significantly promote the activity of the Au₲MgGa₂O₄ ensemble for both the CO oxidation and CH₄ combustion reactions.

Here we take the WGSR as an example to illustrate the promotion mechanism of CeO₂ for H₂O splitting on the CeO₂/[Au₲MgGa₂O₄] catalyst. In order to detect the ability to adsorb and dissociate water, in-situ DRIFTs of the H₂O adsorption experiment was recorded on both the Au₲MgGa₂O₄-800℃-5h and CeO₂/[Au₲MgGa₂O₄-800℃-5h] catalysts under 10 vol% H₂O at 200 ℃. As shown in Fig. 7 (0 min), the typical bands for the O‒H stretching vibration are observed after the adsorption saturation of H₂O. For the CeO₂/[Au₲MgGa₂O₄-800℃-5h] sample, the O‒H stretching vibration of H₂O and the OH groups appeared between 3800 and 3000 cm^‒1. The bands appearing in this region can be summarized as follows: (1) The bands at 3750, 3670 cm^‒1, and 3494 cm^‒1 are related to the stretching vibrations of the monodentate, bidentate and tridentate ceria hydroxyl groups [38], and (2) the broad band at ca. 3400 cm^‒1 can be ascribed to physiosorbed H₂O molecules bound by weak hydrogen bonds with each other [39]. For the Au₲MgGa₂O₄-800℃-5h sample, only the band at ca. 3400 cm^‒1 due to the O-H stretching vibration of physiosorbed H₂O molecules was observed. These results indicate that CeO₂/[Au₲MgGa₂O₄-800℃-5h] can dissociate H₂O molecules to form ceria hydroxyl groups, while this does not happen over the Au₲MgGa₂O₄-800℃-5h catalyst.

In order to investigate the role of the dissociated H₂O molecules in the WGSR, we performed the reaction between CO and adsorbed H₂O over the catalysts studied. After purging with He, 2 vol% CO in He was introduced into the reaction cell and the spectrum was recorded as a function of time. Fig. 7 shows the DRIFTs signals observed for both CeO₂/[Au₲MgGa₂O₄-800℃-5h] and Au₲MgGa₂O₄-800℃-5h during the first 10 min of the reaction. For the CeO₂/[Au₲MgGa₂O₄-800℃-5h] catalyst, CO₂ bands appeared between 2280–2400 cm^-1 and their intensity sharply increased within the first 2 min and then decreased. Along with the changes observed for the CO₂ bands, the intensities of the bands corresponding to the ceria hydroxyl groups also obviously decreased during the first 2 min of the reaction and then remained stable, especially the tridentate ceria hydroxyl band observed at ca. 3494 cm^-1. These results indicate that CO reacts with the ceria hydroxyl groups to produce CO₂. For the Au₲MgGa₂O₄-800℃-5h catalyst, there were no obvious changes in the bands corresponding to the physiosorbed H₂O molecules and CO₂. In combination with the above results, it was obvious that the addition of CeO₂ promoted the dissociation of water and the dissociated hydroxyl groups then reacted with CO to form CO₂.

In view of the above analysis, the possible mechanism for the activity promotion of CeO₂ in anti-sintering Au₲MgGa₂O₄ is illustrated in Fig. 8. It has been reported that Au NPs are efficient for CO activation at low temperature, while not efficient at dissociating water molecules [13, 14]. The addition of CeO₂ makes up for the deficiency of Au₲MgGa₂O₄ toward water dissociation. The formation of Ce-OH observed in the in-situ DRIFTs results illustrates that the H₂O molecules adsorbed on the surface of CeO₂ are dissociated and the resulting hydroxyl groups then react with CO to form CO₂, which allows the reaction to proceed smoothly. According to the well acknowledged mechanism of the WGSR based on theoretical calculations [13, 40], the possible reaction pathway that occurs on the CeO₂/[Au₲MgGa₂O₄] catalyst is shown in Fig. 8. Firstly, the water molecule dissociates on the O vacancies of ceria to form ‒OH and ‒H, and CO is adsorbed on the Au sites located nearby. The adsorbed CO then reacts with the dissociated hydroxyl group to form carboxyl (‒HOCO), which decomposes to CO₂ and adsorbed •H. Two adsorbed •H then combine to produce H₂ and thus, the reaction cycle is completed. In addition, CeO₂ is an excellent reducible support, which has high oxygen mobility and storage capacity [22]. Meanwhile, it has been reported that oxygen vacancies can be created on the ceria lattice at moderate temperatures and under reductive atmospheres, and these oxygen vacancies can also provide active sites for O₂ dissociation [23, 25]. It has also been experimentally proven that the addition of CeO₂ supplies active oxygen for the oxidation reaction [17, 26, 41]. Thus, the promotion effect of CeO₂ in CeO₂/[Au₲MgGa₂O₄-800℃-5h] in the CO oxidation and CH₄ combustion reactions is attributed to the improvement in the activation of oxygen. On the whole, the CeO₂/[Au₲MgGa₂O₄] catalyst combines the advantages of MgGa₂O₄ and CeO₂ to stabilize the Au NPs and activate certain reactants, respectively, and can be used as a durable and active heterogeneous gold catalyst.

In this paper, we have reported that the catalytic activity of the anti-sintering Au NPs on MgGa₂O₄ can be further promoted by proximate contact with CeO₂. Characterizations using XRD, HAADF-STEM, EDS mapping, XPS, and H₂-TPR showed that the MgGa₂O₄ stabilized Au NPs were in intimate contact with the CeO₂ with strong synergistic effects in the CeO₂-modified Au/MgGa₂O₄ catalyst. Such a stable matrix exhibits significantly improved catalytic activity in the WGSR, catalytic combustion of methane, and CO oxidation reaction. Specifically, severely aged Au₲MgGa₂O₄ together with the CeO₂ promoter presents even higher activity than that of the fresh Au/CeO₂-RRCe-2 benchmark catalyst in the WGSR and reduces the "light off" temperature (T₅₀) for methane combustion and CO oxidation by 80 and 100 ℃, respectively when compared with that of bare Au₲MgGa₂O₄ catalyst. The great promotion effect of CeO₂ on the catalytic activity of anti-sintering Au₲MgGa₂O₄ was derived from its excellent ability for the dissociation of H₂O or O₂. This stepwise stabilization-activation strategy will be an effective way for rationally constructing stable and active heterogeneous gold catalysts.