VOCs are major contributors to ozone and photochemical smog, which lead to air pollution and human being's sickness [1]. Methanol is one of the VOCs and widely used as solvents, fuels for specialized vehicles or feedstock for manufacturing of other value-added chemicals [2, 3]. Furthermore, methanol is highly toxic and harmful to human health. Hence, it is essential to develop practical and cost-effective methods to control the emission of methanol. Up to now, the most efficient technology for VOC elimination is the catalytic combustion, which possesses a number of advantages, such as high efficiency, low light-off temperature, simple equipment, and no secondary pollution [4, 5]. The key point is to choose effective catalysts with high performance.

Supported noble metals (especially Pd) have been considered as the most popular catalysts used for VOCs combustion due to their high performance at low temperatures [6, 7]. Nevertheless, the presence of moisture can be the adverse factor influencing the activity of a supported noble metal catalyst [8]. Noble metal alloy catalysts were reported to possess activities superior to those of their single-component counterparts, and the former also displayed better thermal stability and poison-resistant ability than the latter. For example, Wei and coworkers reported that the AuPd/3DOM-TiO₂ [9], PtRu/TiO₂ [10], and Pd_xCo_3–xO₄/3DOM Ce_0.2Zr_0.8O₂ [11] catalysts were active for the photocatalytic reduction of CO₂ and soot combustion. Calzada et al. [1] found that the Ru-Au/TiO₂ catalyst performed better than the supported Ru or Au counterpart at low temperatures (< 50 ℃) in the total oxidation of methanol. Wang et al. [12] reported that the 0.96AuPd_1.92/Co₃O₄ sample showed a higher catalytic activity for toluene and o‐xylene oxidation (the temperature required for achieving 90% conversion of toluene and o‐xylene was 180 and 187 ℃, respectively) than the supported Au or Pd catalyst at a space velocity (SV) of 40000 mL (g^–1 h^–1). The authors assigned the good catalytic activity of 0.96AuPd_1.92/Co₃O₄ to the high concentration of adsorbed oxygen species and strong interaction between Au-Pd nanoparticles (NPs) and Co₃O₄. Xie et al. [13] synthesized a series of the supported Au-Pd catalysts and found that these porous transition-metal oxide-supported Au-Pd NPs showed high activity and good (hydro)thermal stability for toluene or methane combustion. Guo et al. [14] observed better performance for methanol combustion of the TiO₂-supported Pd-Pt NPs than the supported Pd NPs, a result due to a more amount of adsorbed oxygen species on the former than on the latter. Xie et al. [15] pointed out that the three-dimensionally ordered macroporous (3DOM) Co₃O₄-supported Au-Pd-xCoO catalysts performed well in CH₄ combustion.

The choice of a transition-metal oxide support is important in formulating an effective catalyst because it could provide oxygen vacancies that would be favorable for the adsorption and activation of oxygen and/or VOCs molecules [16, 17]. Transition-metal oxides were widely used in catalytic oxidation of VOCs because of their good redox ability [18-21]. Among transition-metal oxides, cobalt oxide was used as support for the removal of VOCs and methane [22-24]. In the previous studies, mesoporous cobalt oxides with high surface areas and ordered pore structures showed better catalytic performance than their nonporous counterparts. For example, Wu et al. [25] reported that 2.94Au_0.5Pd/meso-Co₃O₄ performed better than 2.90Au_0.5Pd/bulk-Co₃O₄ for methane combustion, which was associated with its porous structure, high adsorbed oxygen species concentration, good low-temperature reducibility, and strong interaction between Au-Pd alloy NPs and meso-Co₃O₄. Liu et al. [26] observed excellent performance for benzene, toluene, and o-xylene oxidation of the 6.5Au/meso-Co₃O₄ sample with high surface areas.

Previously, our group prepared a series of mesopoorous transition-metal oxides (e.g., meso-Co₃O₄ [25, 26], meso-MnO_x [27, 28], and meso-CrO_x [29, 30]) via the KIT-6-templating route, and found that most of these porous materials exhibited good performance for VOC combustion. Although a number of bimetal alloy catalysts were reported in the literature, there have been no works on the ternary metal alloy catalysts used for VOC combustion. In the present work, we choose Pd as the major active component, add cheaper noble metals (e.g., Ag and Au) to formulate the ternary metal (Ag-Au-Pd) alloy NPs, load them on the surface of meso-Co₃O₄, and investigate catalytic properties of the as-obtained materials for the combustion of methanol. It is found that the 0.68 wt% Ag_0.75Au_1.14Pd/meso-Co₃O₄ sample showed the best catalytic activity for the addressed reaction.

Three-dimensionally mesoporous silica (KIT-6) was synthesized according to the procedures described by Kleitz et al. [31]. We use KIT-6 as the hard template and manganese nitrate as the metal source to fabricate the three-dimensionally ordered mesoporous Co₃O₄ (meso-Co₃O₄). In a typical synthesis, 1.0 g of Co(NO₃)₂·6H₂O was dissolved in 10 mL of ethanol, and then 0.5 g of KIT-6 was added to the Co-containing ethanol solution. The mixture was stirred at 65 ℃ until the solution was dried. The above mixture was placed in a crucible and calcined in a muffle furnace at a ramp of 1 ℃/min from room temperature (RT) to 600 ℃ and kept at this temperature for 6 h. The obtained powders were leached twice with a hot NaOH aqueous solution (2.0 mol/L), followed by washing with deionized water and ethanol two times and drying at 60 ℃ for 24 h, thus obtaining the meso-Co₃O₄ support.

The meso-Co₃O₄-supported Ag_xAu_yPd (Ag/Au/Pd molar ratio = x : y : 1), Ag, Au, and Pd samples were prepared using the polyvinyl alcohol (PVA)-protected sodium borohydride reduction strategy [28]. The typical preparation procedure is as follows. 2.25 mL of PVA (MW = 10000 g/mol) was added to (ⅰ) a mixture of 0.79, 0.34 or 0.40 mL of AgNO₃ aqueous solution (14.7 mmol/L), 0.56, 0.96 or 0.56 mL of HAuCl₄ aqueous solution (10.3 mmol/mL), and 0.21, 0.18 or 0.41 mL of PdCl₂ aqueous solution (28.0 mmol/mL) with the metal/PVA mass ratio = 1.0 : 1.5 and the theoretical Ag:Au:Pd molar ratio was 2:1:1, 1:2:1, and 1:1:2, respectively, or (ⅱ) 0.40 mL of AgNO₃ aqueous solution (14.7 mmol/L), or (ⅲ) 0.56 mL of HAuCl₄ aqueous solution (10.3 mmol/mL), or (ⅳ) 0.41 mL of PdCl₂ aqueous solution (28.0 mmol/mL) at RT under vigorous stirring for 15 min. After rapidly injecting the NaBH₄ aqueous solution (0.1 mol/L, and metal/NaBH₄ molar ratio = 1.0:5.0), a dark-brown sol was generated. 0.3 g of the meso-Co₃O₄ support (the nominal Ag_xAu_yPd loading in each of the supported samples was 1.00 wt%; and the nominal Ag, Au, and Pd loadings in the supported single noble metal samples were 0.42, 0.38, and 0.35 wt%, respectively) was added to the above dark-brown sol under stirring until complete adsorption (the solution was totally decolorized) was achieved. The obtained mixture was filtered, washed with deionized water, dried at 80 ℃ for 12 h, and calcined at a ramp rate of 1 ℃/min from RT to 500 ℃ and maintained this temperature for 2 h. According to the results of the inductively coupled plasma-atomic emission spectroscopic (ICP-AES) characterization, the 0.68 wt% Ag_0.75Au_1.14Pd/meso-Co₃O₄, 0.84 wt% Ag_0.54Au_2.29Pd/meso- Co₃O₄, 0.93 wt% Ag_0.51Au_0.65Pd/meso-Co₃O₄, 0.28 wt% Ag/meso-Co₃O₄, 0.35 wt% Au/meso-Co₃O₄, and 0.33 wt% Pd/meso-Co₃O₄ samples were obtained and denoted as 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄, 0.84Ag_0.54Au_2.29Pd/meso- Co₃O₄, 0.93Ag_0.51Au_0.65Pd/meso-Co₃O₄, 0.28Ag/meso-Co₃O₄, 0.35Au/meso-Co₃O₄, and 0.33Pd/meso-Co₃O₄, respectively.

The ICP-AES technique was used to detect the actual metal contents of Ag, Au, and Pd in the samples on a Thermo Electron IRIS Intrepid ER/S spectrometer. Before analysis, the sample was dissolved in a mixture of concentrated HCl and HNO₃ with a volumetric ratio of 3:1. X-ray diffraction (XRD) patterns of the samples were recorded on a Bruker D8 Advance diffractometer using the Cu K_α radiation and nickel filter (λ = 0.15406 nm). The transmission electron microscopic (TEM) images were obtained on a JEOL JEM-2010 instrument. High angle annular dark field and scanning transmission electron microscopic (HAADF-STEM) techniques were used to record the HAADF-STEM images and EDX element mapping of the typical sample on the equipment FEI G2 80-200/Chemi-STEM Cs-corrected TEM with a probe corrector operated at accelerating voltage 200 kV. Before measurements, a small amount of the sample was ultrasonicated in ethanol for 15 min, and the resulting slurry was then deposited on the copper-coated grids. BET (Brunauer-Emmett-Teller) surface areas and pore-size distributions of the samples were measured via N₂ adsorption at -196 ℃ on a Micromeritics ASAP 2020 analyzer, with the samples being degassed at 250 ℃ for 2.5 h under vacuum before measurement.

X-ray photoelectron spectroscopy (XPS, Thermo Fisher Scientific ESCALAB 250 Xi) was used to determine the binding energies (BEs) of Co 2p, O 1s, Ag 3d, Au 4f, Pd 3d, and C 1s of surface species using Al K_α (hv = 1486.6 eV) as excitation source. In order to remove the adsorbed water and carbonate species on the surface, the samples were pretreated in O₂ (flow rate = 20 mL/min) at 450 ℃ for 1 h and then cooled to RT, followed by transferring the pretreated samples into the spectrometer in a transparent Glove Bag (Instruments for Research and Industry, USA) filled with helium. The pretreated samples were degassed in the preparation chamber (10^-5 Torr) for 0.5 h and then introduced into the analysis chamber (3 × 10^-9 Torr) for XPS spectrum recording. The C 1s signal at BE = 284.6 eV was taken as reference for BE calibration.

Hydrogen temperature-programmed reduction (H₂-TPR) experiments were carried out on a chemical adsorption analyzer (Autochem Ⅱ 2920, Micromeritics). Before TPR measurement, ca. 0.03 g of the sample (40-60 mesh) was loaded to a quartz fixed-bed U-shaped microreactor (i.d. = 4 mm) and pretreated in an air flow of 30 mL/min at 250 ℃ for 1 h. After being cooled at the same atmosphere to RT, the pretreated sample was exposed to a flow (40 mL/min) of 10% H₂-90% Ar (v/v) mixture and heated from RT to 450 ℃ at a ramp rate of 10 ℃/min. The alteration in H₂ concentration of the effluent was monitored online by the chemical adsorption analyzer. The reduction peak was calibrated against that of the complete reduction of a known standard powdered CuO (Aldrich, 99.995%).

Catalytic activities of the samples were evaluated in a continuous flow fixed-bed quartz microreactor (i.d. = 4.0 mm). 25 mg of the sample (40-60 mesh) was diluted with 0.15 g of quartz sands (40-60 mesh) to minimize the effect of hot spots. A thermocouple was placed in the middle of the catalyst bed, and methanol conversion was measured after 1 h of the reaction that reached a steady state. Before the test, each sample was treated in an oxygen flow of 20 mL/min at 250 ℃ for 1 h. After being cooled to a given temperature, the reactant gas mixture was passed through the catalyst bed. The total flow rate of reactant gas mixture (0.1 vol% methanol + O₂ + N₂ (balance)) was 33.4 mL/min, giving a methanol/O₂ molar ratio of 1/200 and a space velocity (SV) of ca. 80000 mL (g^–1 h^–1). The 0.1 vol% methanol was generated by passing a N₂ flow through a pure methanol-containing bottle that was chilled in an ice-water isothermal bath (0 ℃). In the case of water vapor introduction, 3.0 vol% H₂O was introduced by passing the feed stream through a water saturator at 33 ℃. In the case of CO₂ addition, 5.0 vol% CO₂ was introduced from a CO₂ cylinder with N₂ as the balanced gas. Reactants and products were analyzed online on a gas chromatograph (GC-2010, Shimadzu) equipped with a flame ionization detector (FID), using a stabilwax@-DA column (30 m in length) for methanol separation. Catalytic activities of the samples were evaluated using the temperatures (T_50% and T_90%) required for achieving methanol conversions of 50% and 90%, respectively. Methanol conversion was defined as (c_inlet − c_outlet)/c_inlet × 100%, where c_inlet and c_outlet are toluene concentrations of the inlet and outlet feed stream, respectively. Only CO₂ was detected in the outlet. The balance of carbon throughout the investigation was estimated to be 98.5% ± 1.5%. Reaction rate per gram of noble metal was calculated according to the following formula: r = c_{R, in} × X/m_{noble metal}, where c_{R, in} (mol/s) is the initial reactant concentration, X is the methanol conversion rate, and m_{noble metal} (g) is the mass of the catalyst.

Fig. 1 shows the wide-angle XRD patterns of the as-prepared samples. All of the diffraction peaks can be well indexed, as indicated in Fig. 1g. By referring to the XRD pattern (JCPDS PDF# 42-1467) of the standard Co₃O₄ sample, one can realize that the Co₃O₄ support was cubic in crystal structure [32, 33], and the appearance of a signal at 2θ ≈ 1° in the small-angle XRD pattern (not shown here) suggests generation of an ordered mesoporous structure in the Co₃O₄ support, which was synthesized using the same procedures described in our previous works [15, 26]. There were no apparent differences in XRD patterns of the supported noble metal samples, indicating that the loading of noble metal NPs did not change the structure of the support and the noble metal NPs were highly dispersed on the surface of meso-Co₃O₄.

TEM images shown in Fig. 2 demonstrate that a 3D ordered mesoporous structure was formed in the samples, and the mesopore diameter of meso-Co₃O₄ was ca. 6 nm. Ag, Au, Pd, and Ag_xAu_yPd NPs were uniformly dispersed on the surface of meso-Co₃O₄. The HAADF-STEM images of the 0.68Ag_0.75Au_1.14Pd/ meso-Co₃O₄ sample is shown in Fig. 3. It can be obviously seen that the NPs in the selected region were composed of Ag, Au, and Pd, indicating formation of the Ag-Au-Pd alloy in Ag_0.75Au_1.14Pd NPs. After statistically analyzing 100 noble metal NPs in the TEM images, we measured their average sizes and the results are shown in Fig. 4. As summarized in Table 1, the average sizes of Ag, Au, and Pd NPs were 4.1, 4.5, and 2.8 nm, respectively, while those of Ag_xAu_yPd NPs were 2.9-3.0 nm (Fig. 4).

N₂ adsorption-desorption isotherms and pore-size distributions of the samples are shown in Fig. 5A and B, respectively. All of the samples exhibited a type Ⅳ isotherm with a H1 hysteresis loop in the relative pressure (p/p₀) range of 0.4-1.0, demonstrating the generation of ordered mesopores. There was a peak in the pore-size range of 5.7-6.0 nm of each sample (Fig. 5B), further confirming the presence of mesopores in these samples. The above results were in good accordance with those of the small-angle XRD and TEM investigations. Surface areas and pore volumes of the samples are summarized in Table 1. The surface area, average pore size, and pore volume of meso-Co₃O₄ were 125 m²/g, 6.1 nm, and 0.14 cm³/g, respectively. After loading noble metal NPs on the support, however, there were no significant decreases in surface area of the supported noble metal samples. The supported noble metal samples possessed a surface area of 115-120 m²/g, an average pore size of 5.7-6.0 nm, and a pore volume of 0.15-0.16 cm³/g (Table 1).

XPS technique was applied to investigate the surface element compositions, metal oxidation states, and adsorbed oxygen species of the samples. Fig. 6 shows the Co 2p_3/2, O 1s, Ag 3d_5/2, Au 4f, and Pd 3d XPS spectra of the samples. By using the curve-fitting method, we can decompose the Co 2p_3/2 into three components (Fig. 6A): the ones at binding energy (BE) = 779.8 and 781.5 eV were assigned to the surface Co³⁺ and Co²⁺ species, respectively, whereas the one at BE = 788.9 eV was attributed to the satellite signal of the surface Co²⁺ species [34, 35]. The asymmetrical O 1s XPS signals could be decomposed into three components at BE = 529.9, 531.3, and 532.7 eV (Fig. 6B), ascribable to the surface lattice oxygen (O_latt), adsorbed oxygen (O_ads, e.g., O₂^-, O₂^2- or O^-), and adsorbed molecular water or carbonate species [36], respectively. It is well known that there is a close link between the ability to activate oxygen and the activity of a catalyst. The electrophilic O₂^-, O₂^2- or O^- species can play an important role in the oxidation of organics. The supported Ag_xAu_yPd NPs possessed much higher amounts of the O_ads species than the supported Ag, Au or Pd NPs, making the former outperform the latter for methanol combustion, as confirmed by their activity data. The broad and asymmetrical Ag 3d_5/2 XPS signal of each sample could be deconvoluted into two components at BE = 367.5 and 368.0 eV, which was ascribed to the surface Ag⁺ and Ag⁰ species [37, 38], respectively. The Au XPS signal of each sample could be decomposed into four components (Fig. 6D): the ones at BE = 84.0 and 87.7 eV were attributed to the surface Au⁰ species, whereas the ones at BE = 85.1 and 88.7 eV were assigned to the surface Au^δ+ species [39]. As shown in Fig. 6E, the Pd 3d XPS signal was decomposed into two sets of components: the ones at BE = 335.5 and 340.7 eV were assignable to the surface Pd⁰ species, while the ones at BE = 337.2 and 342.4 eV were attributable to the surface Pd²⁺ species [40, 41]. A higher Pd²⁺/Pd⁰ molar ratio could also favor the improvement in catalytic activity because Pd²⁺ was the active site for activation of VOCs. The surface element compositions of the samples are summarized in Table 2. After loading Ag_xAu_yPd NPs on meso-Co₃O₄, the surface Co³⁺/Co²⁺ molar ratio decreased from 0.80–0.90 to 0.60–0.69 (Table 2), indicating that the surface Co²⁺ concentration increased. Such a decrease in Co³⁺/Co²⁺ molar ratio as well as formation of the Au^δ+, Ag⁺, and Pd²⁺ species was due to the enhanced strong interaction between noble metal NPs and support. Obviously, the 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ sample possessed the lowest surface Co³⁺/Co²⁺ molar ratio. In other words, there was the highest oxygen vacancy concentration on the surface of 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄. A more amount of oxygen vacancies can give rise to a more amount of the O_ads species, consequently resulting in a better performance of a catalyst. Therefore, it is understandable that the 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ sample showed the best activity for the oxidation of methanol.

H₂-TPR profiles of the samples are shown in Fig. 7A. According to the literature, the two reduction peaks at 300 and 320-410 ℃ of the Co₃O₄ sample were ascribable to reduction of Co³⁺ to Co²⁺ and of Co²⁺ to Co⁰ [42, 43], respectively. After loading Ag and Au, the newly appeared peaks at 200 and 265 ℃ were attributable to the reduction of these noble metal oxides. When Pd or Ag_xAu_yPd was loaded on the surface of meso-Co₃O₄, the reduction peaks were significantly shifted to low temperatures (the first peak at 115 ℃ was mainly due to reduction of PdO_x, and the reduction of Co³⁺ and Co²⁺ was shifted to a lower temperature, which indicates the presence of a strong interaction between noble metal alloy NPs and support). Such a phenomenon was a result due to reduction of meso-Co₃O₄ promoted by the hydrogen chemically adsorbed on Pd or Ag_xAu_yPd NPs. The results demonstrate that loading of Pd or Ag_xAu_yPd NPs considerably improved the low-temperature reducibility of the sample. By quantitatively analyzing the reduction peaks in the H₂-TPR profiles, one can obtain the H₂ consumption of the samples, as summarized in Table 2. The H₂ consumptions of the samples were in the range of 19.6-21.0 mmol/g_cat (slightly higher than the theoretical H₂ consumption of Co₃O₄), which was due to the reduction of the oxidized noble metal NPs and the removal of the adsorbed oxygen species.

The initial H₂ consumption rate was calculated according to the H₂ consumption per gram of catalyst per second, which corresponded to the initial 25% of the first reduction peak where no phase transformation of the sample occurred [44, 45]. The results are shown in Fig. 7B. The initial H₂ consumption rate increased in the sequence of 0.28Ag/meso-Co₃O₄ < 0.35Au/meso-Co₃O₄ < 0.33Pd/meso-Co₃O₄ < 0.84Ag_0.54Au_2.29Pd/meso-Co₃O₄ < 0.93Ag_0.51Au_0.65Pd/meso- Co₃O₄ < 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄. That is to say, the 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ sample exhibited the best low-temperature reducibility. The trend in low-temperature reducibility was in agreement with the order in catalytic activity of the samples (shown below).

No significant methanol conversions (< 2%) at temperature < 220 ℃ and SV = 80000 mL g^–1 h^–1 were detected in the blank experiment (only quartz sands were loaded in the microreactor), indicating that no considerable gas-phase reaction existed in the catalytic system. Fig. 8A shows the catalytic activities of the as-prepared samples at SV = 80000 mL g^–1 h^–1. Obviously, the supported Ag_xAu_yPd samples outperformed the supported Ag, Au or Pd sample. Among the supported Ag_xAu_yPd samples, catalytic activity decreased in the order of 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ > 0.93Ag_0.51Au_0.65Pd/meso- Co₃O₄ > 0.84Ag_0.54Au_2.29Pd/meso-Co₃O₄, with the 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ sample exhibiting the highest activity (T_50% = 100 ℃ and T_90% = 112 ℃ at SV = 80000 mL g^–1 h^–1. Fig. 8B shows the methanol reaction rate normalized per gram of noble metal versus temperature of the samples. It can be seen that the 0.28Ag/meso-Co₃O₄ sample almost showed the same reaction rate as the 0.35Au/meso-Co₃O₄ sample, while the supported Pd sample performed better than the supported Ag or Au sample. The reaction rate decreased in the order of 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ > 0.33Pd/meso-Co₃O₄ > 0.93Ag_0.51Au_0.65Pd/meso-Co₃O₄ ≈ 0.84Ag_0.54Au_2.29Pd/meso- Co₃O₄ > 0.35Au/meso-Co₃O₄ ≈ 0.28Ag/meso-Co₃O₄, with the 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ sample possessing the highest catalytic efficiency. The turnover frequencies (TOF_{noble metal}) normalized per gram of noble metal at 100 ℃ (× 10^‒3 s^‒1) were calculated and summarized in Table 2. The 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ sample possessed the highest TOF_{noble metal} as well as the highest methanol conversion and reaction rate, which might be a result due to the good dispersion of Ag_0.75Au_1.14Pd alloy NPs, high adsorbed oxygen species concentration, good low-temperature reducibility, and strong interaction between Ag_0.75Au_1.14Pd alloy NPs and meso-Co₃O₄. The above results demonstrate that the alloying of Pd with proper amounts of Ag and Au favored the enhancement in methanol combustion activity of the sample.

There have been several works on the oxidation of methanol over various catalysts in the literature. Our 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ sample showed a methanol reaction rate at 100 ℃ of 0.99 μmol g_cat^–1 s^–1, which was higher than that (0.01 μmol g_cat^–1 s^–1) over meso-Cr₂O₃ [30], that (0.03 μmol g_cat^–1 s^–1) over wormhole-like Fe₂O₃ [46], that (0.11 μmol g_cat^–1 s^–1) over meso-Co₃O₄ [47], that (0.60 μmol g_cat^–1 s^–1) over 12 wt% MnO_x/3DOM LaMnO₃ [48], that (0.45 mmol g_cat^–1 s^–1) over 2 wt% Ag/CeO₂ [49], and that (0.88 μmol g_cat^–1 s^–1) over 0.1 wt% Pd/5.0 wt% Ag/35CZA [50], but lower than that (3.50 μmol g_cat^–1 s^–1) over 0.8% Pd-Pt/CeO₂-Al₂O₃-TiO₂ [14].

Considering a weak mobility of lattice oxygen in meso-Co₃O₄ at low temperatures, methanol combustion over Ag_xAu_yPd/meso-Co₃O₄ would hardly follow the Mars van Krevelen-type mechanism under the present reaction conditions. The interface mechanism could work well for methanol combustion over the reducible oxide-supported noble metal catalysts. Gas-phase oxygen was adsorbed on the surface oxygen vacancies of meso-Co₃O₄ to produce the active oxygen species, and methanol was adsorbed on the noble metal sites of the catalysts to form methoxy species that were easily oxidized to the formate species by the active oxygen species [49]; then, the generated formate species decomposed to CO and hydroxyl species; and finally, the formed CO was completely oxidized to CO₂.

It is well known that catalytic activity is intimately associated with the transition-metal oxidation state, O_ads species concentration, and reducibility of a sample. The O_ads/O_latt molar ratio versus surface Co³⁺/Co²⁺ molar ratio and methanol reaction rate at 100 ℃ versus O_ads/O_latt molar ratio or initial H₂ consumption rate of the samples are shown in Fig. 9. It can be observed that (ⅰ) the O_ads/O_latt molar ratio increased with a drop in surface Co³⁺/Co²⁺ molar ratio (i.e., an increase in surface oxygen vacancy density that was induced by the strong interaction between noble metal NPs and meso-Co₃O₄) (Fig. 9A), (ⅱ) methanol reaction rate increased with increasing the O_ads/O_latt molar ratio (Fig. 9B), and (ⅲ) methanol reaction rate increased with an increase in initial H₂ consumption rate (i.e., low-temperature reducibility) (Fig. 9C). Among all of the samples, 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ possessed the highest O_ads species concentration (the lowest Co³⁺/Co²⁺ molar ratio gave rise to the highest oxygen vacancy density) and the best low-temperature reducibility, hence showing the best catalytic performance for methanol combustion.

Because water vapor exerted a negative effect on catalytic activity for the oxidation of VOCs [51], it is necessary to examine the effect of water vapor on catalytic activity of the typical sample. By introducing 3.0 vol% water vapor to the reaction system, we measured methanol conversions at 110 ℃ and SV = 80000 mL g^–1 h^–1 over the 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ sample, as shown in Fig. 10A. It is well known that water is easily adsorbed on the surface of noble metal (especially Pd) NPs. Obviously, methanol conversion dropped by ca. 6% after addition of 3.0 vol% water vapor. When water vapor was cut off, methanol conversion was restored to its original value in the absence of water vapor, indicating that the partial deactivation induced by water vapor addition was reversible. Such a partial deactivation was due to the competitive adsorption of H₂O and methanol as well as O₂, leading to a decrease in catalytic activity [52, 53].

As one of the products of methanol combustion, CO₂ might influence the activity of a catalyst. To examine the effect of CO₂ on activity, we conducted methanol combustion at 110 ℃ and SV = 80000 mL g^–1 h^–1 over 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ in the presence of 5.0 vol% CO₂, and the results are shown in Fig. 10B. It is apparent that methanol conversion decreased by ca. 7% after 15 h of on-stream reaction. When CO₂ was cut off, methanol conversion was recovered to its original value in the absence of carbon dioxide. CO₂ molecules were adsorbed on the active sites of the sample to form the CO₃^2‒ or HCO₃^‒ species, thus influencing the activity. After CO₂ was cut off, however, these carbonate species were possibly desorbed from the active sites of the sample, hence recovering the occupied active sites.

The cubically crystallized meso-Co₃O₄ and its supported Ag_xAu_yPd, Ag, Au, and Pd catalysts were prepared using the KIT-6-templating and PVA-protected NaBH₄ reduction methods, respectively. The as-obtained materials displayed a three-dimensionally ordered mesoporous structure and a high surface area (115-125 m²/g), and the noble metal NPs with an average size of 2.8-4.5 nm were highly dispersed on the surface of meso-Co₃O₄. The supported Ag_xAu_yPd samples outperformed the Ag, Au or Pd sample, with the 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ sample showing the highest catalytic activity (T_50% = 100 ℃ and T_90% = 112 ℃ at SV = 80000 mL g^–1 h^–1) for methanol combustion. The partial deactivation due to H₂O or CO₂ addition of the 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ sample was reversible. It is concluded that the good catalytic performance of 0.68Ag_0.75Au_1.14Pd/meso-Co₃O₄ was related to the highly dispersed Ag_0.75Au_1.14Pd alloy NPs, high O_ads species concentration, good low-temperature reducibility, and strong interaction between Ag_0.75Au_1.14Pd alloy NPs and meso-Co₃O₄.