The flow scheme of the preparation of multi-catalysts was shown in Figure 1. The structured γ-Al₂O₃/Al monolith was prepared by the anodization method. The Al mesh (1060, 99.6%) was pretreated in a 10 wt% sodium hydroxide solution for 4 min and a 10 wt% nitric acid solution for 2 min, followed by a flush with deionized water. Then the pretreated mesh was anodized in a 0.4 mol/L oxalic acid solution for 10 h with a current density of 25 A/m² at 20˚C. The resulting mesh with porous alumina layers was calcined in air at 350˚C for 1 h to remove the residual oxalic acid. Afterward, it was subjected to hot water treatment (HWT) in deionized water at 80˚C for 1 h. Finally, the obtained meshes were dried and calcined in air at 500˚C for 4 h.

With the anodic alumina substrate, a series of Cu/X/γ-Al₂O₃/Al (X = Ni, Zn, La or Fe) catalysts were prepared through impregnation method. The monolithic γ-Al₂O₃/Al substrates were impregnated in an aqueous solution of Cu (II) nitrate or X nitrate (X = Ni, Zn, La or Fe) under ambient conditions. The prepared catalyst was then dried naturally, and calcined in air at 500˚C for 4 h.

The surface morphology of catalysts was measured by a scanning electron microscope (JSM-6360LV, JEOL).

The specific surface area and pore structure of catalysts was examined using adsorption method by a physisorption analyzer (ASAP 2020-M, Micromeritics). The specific surface area was calculated by the BET method, and the BJH method was used to determine the pore volume and average pore diameter.

The metal loading was analyzed by an Inductively Coupled Plasma-Atomic Emission Spectrometry (ICP-AES, 725ES, Agilent) and is reported here based on the quantity of the surface alumina layers.

An X-ray diffraction (D/max 2500 VB/PC, Rigaku) was applied to characterize the crystal structure of catalysts, and the average grain size of metal species was calculated through Scherrer’s equation. The tested powder was scratched from the surface of the catalysts.

H₂ temperature-programmed reduction (H₂-TPR) analysis was performed in a Micromeritics ChemiSorb 2720 apparatus. Each sample including 100 mg catalyst was outgassed at 160˚C for 40 min under He flow (25 ml/min), then cooled to room temperature. Afterward, the temperature was raised to 920˚C with a ramping rate of 10˚C/min under a 10 vol% H₂/Ar flow atmosphere.

The dispersion of Cu and exposed copper surface area (S_Cu) were measured by selective N₂O chemisorption method in a chemisorption analyzer (ChemiSorb 2720, Micromeritics). First, catalysts were reduced in a 10 vol% H₂/Ar stream at 400˚C for 2 h. Subsequently, the samples were purged with He at 400˚C for 20 min to remove hydrogen species adsorbed on the surface, and subsequently cooled to 50˚C. A flow of 20% N₂O/N₂ (30 mL/min) was used to oxidize surface copper atoms to Cu₂O at 50˚C for 0.5 h. The reactor was flushed with He to remove the oxidant. Finally, another TPR experiment was performed in 10% H₂/Ar at a flow rate of 30 mL/min. The copper surface density is 1.46 × 1019 copper atoms per square meter. The copper dispersion (DCu) was defined as the ratio of copper atoms on the surface of the catalysts to the total amount of copper atoms in the catalyst.

X-ray photoelectron spectroscopy (XPS) was obtained with a Thermo ESCALAB 250 spectrometer with an Al Kα radiation. The binding energies scale (BEs) of the spectrometer was calibrated using the carbonaceous C 1s line at 284.6 eV.

Two different reactors were applied to quantify the catalytic performances of the catalysts under atmospheric pressure. The mesh-type catalysts were cut into small pieces (2 mm*2 mm), mixed with Raschig rings (20 - 40 mesh), then packed into a fixed-bed reactor (I.D. 12 mm). The catalyst charge was 3 g. And no H₂ pre-reduction treatment was conducted prior to the evaluation of catalysts. Figure 2 shows the diagram of the testing system for SRD. The water was injected with a constant flux pump and evaporated at 150˚C, then was mixed with DME and fed into the reforming chamber (fixed-bed reactor or microreactor). The compositions of reformate gases were determined by an on-line gas chromatograph equipped with a flame ionization detector (FID) and a thermal conductivity detector (TCD). The steam in the feed and reformate was trapped by a condenser before the gas analysis. A PORAPAK-Q column was used to separate the DME and CH₄ gas, and a TDX-01 column was used to separate the

CO, CO₂, and N₂. DME conversion and selectivity of products are defined as follows:

DMEconversion = F D M E , i n − F D M E , o u t F D M E , o u t × 100 % (4)

Selectivityofproducts = F i , o u t ∑ F i , o u t × 100 % (5)

where F_DME,in and F_DME,out are the influent and effluent molar flow rates of DME, respectively, F_i,out are the molar flow rates of gaseous products (H₂, CO, CO₂, CH₄).

The chemical composition, surface area and pore properties of different catalysts were summarized in Table 1. It can be seen that the supporting material, γ-Al₂O₃/Al catalyst presented the highest BET surface area and pore volume. The addition of active metals decreased the specific surface and pore volume. This could be attributed to the deposition of the active components in pore and the block of micropore in the substrate. Besides, the Cu dispersion of different catalysts decreased in the order of: Cu/Ni/Fe/γ-Al₂O₃/Al > Cu/Ni/γ-Al₂O₃/Al > Cu/Ni/Zn/γ-Al₂O₃/Al > Cu/Ni/La/γ-Al₂O₃/Al, indicating that the Cu dispersion was improved by the incorporation of Fe promoter, while decreased by the addition of Zn or La.

In order to study the impacts of promoters on the crystal phases, the diffraction patterns for the Cu/Ni/γ-Al₂O₃/Al catalysts loaded with different metals were presented in Figure 3(a). It can be seen that all four types of catalysts had strong peaks of CuO, which appeared at 2θ = 35.6˚ and 38.5˚. Obviously, the addition of La and Zn promoters made the Cu diffraction peak sharper and stronger to varying degrees than Cu/Ni/γ-Al₂O₃/Al catalyst while the Cu/Ni/Fe/γ-Al₂O₃/Al catalyst exhibited the weakest peak of CuO. This further confirmed that the addition of Fe promoted the dispersion of Cu and reduced the crystallite size of Cu and Fe. The formation of smaller crystallite size of Cu provided more active sites, leading to more active catalyst at low temperature. There was no peak assigned to NiO, indicating that NiO existed in a high dispersion or a microcrystalline state on the gamma alumina substrate [20] .

H₂-TPR was used to analyze the reduction properties of catalysts loaded with different metals. From Figure 3(b), the two reduction peaks located at 425˚C and 750˚C for Fe/γ-Al₂O₃ were ascribed to the reduction of Fe₂O₃ to Fe₃O₄ and Fe₃O₄ to FeO or Fe, respectively [21] [22] . An interesting observation from this study was that the addition of Fe to Cu/Ni/γ-Al₂O₃/Al catalyst decreased the reduction temperature of hematite to magnetite considerably, to 259˚C. In general, the reduction temperature of CuO to Cu was around 300˚C [23] . For

Cu/Ni/Fe/γ-Al₂O₃/Al catalyst, the reduction peak of CuO decreased to 225˚C from 251˚C over the Cu/Ni/γ-Al₂O₃/Al catalyst. On the contrary, the addition of Zn and La to Cu/Ni/γ-Al₂O₃ made the reduction temperature increase to 287˚C and 324˚C, respectively. The results were consistent with Table 1 and XRD analysis, which indicated that compared with the Cu/Ni/γ-Al₂O₃ catalyst, a higher metallic dispersion and a decrease of the crystallite size lead to an improvement of reducibility for Cu/Ni/Fe/γ-Al₂O₃ catalyst.

Figure 4 shows the influences of promoters and temperatures on the values of the DME conversion and the effect of CO in situ removal in SRD system. It can be seen that the DME conversion increased while CO selectivity decreased over all the catalysts with the temperature increasing, because SRD is endothermic. The Cu/Ni/La/γ-Al₂O₃/Al catalyst had the lowest DME conversion and H₂ selectivity. It was due to the generation of CuLaO₃ co-oxide that occupied and

decreased the active sites of Cu, and increased the reduction temperature of CuO. The Cu/Ni/Zn/γ-Al₂O₃/Al catalyst had a lower DME conversion than the Cu/Ni/γ-Al₂O₃/Al catalyst due to a lower dispersion of Cu (as shown in Table 1). As reported, Zn and La could promote WGSR (Equation (6)) and decrease the amount of CO [24] [25] [26] [27] . Nonetheless, the best temperature range is 250˚C - 300˚C for Cu/Zn/Al oxides [24] , and 200˚C - 250˚C for La₂O₃/CuO/CeO₂ catalyst [26] . As such, the addition of Zn and La could not exhibit a better performance at the SRD reaction temperature at 350˚C - 450˚C. It also can be seen that for the Cu/Ni/Fe/γ-Al₂O₃/Al catalyst performed the highest H₂ selectivity and best effect of CO removal. This was mainly attributed to that Fe-based catalysts have a high catalytic performance on WGSR (Equation (6)) above 330˚C [28] [29] . It was obvious that for Cu/Ni/γ-Al₂O₃/Al catalyst, the H₂/CO₂ molar ratio of ca. 6 was larger than that produced by the SRM process, indicating the step of methanol decomposition (Equation (7)) took place to produce H₂, and CO₂ methanation (Equation (9)) to consume CO₂. The fact that the selectivity of CH₄ increased with the decrease of CO selectivity indicated the occurrence of CO methanation (Equation (10)).

CO + H 2 O ↔ CO 2 + H 2 (6)

CH 3 OH → CO + 2 H 2 (7)

CH 3 OCH 3 → CH 4 + CO + H 2 (8)

CO 2 + 4 H 2 ↔ CH 4 + 2 H 2 O (9)

CO + 3 H 2 ↔ CH 4 + H 2 O (10)

The catalytic performance of different packing structures of Fe/γ-Al₂O₃/Al and Cu/Ni/γ-Al₂O₃/Al was evaluated to investigate the synergetic effect between Cu/Ni/γ-Al₂O₃ and Fe on the Cu/Ni/Fe/γ-Al₂O₃/Al catalyst. The three packing structures included supported Cu/Ni/Fe/γ-Al₂O₃/Al catalyst, layered Cu/Ni/γ-Al₂O₃/Al (former) and Fe/γ-Al₂O₃/Al (later) and mechanical mixture of the two kinds of catalysts. For convenience, “supported Cu/Ni/Fe”, “layered Cu/Ni+Fe” and “mixed Cu/Ni*Fe” represents above three packing structures, respectively. And “supported Cu/Ni” represents Cu/Ni/γ-Al₂O₃/Al. The results were showed in the Figure 5. It was found that DME conversion was 100% over the four catalysts above 375˚C. However, there was a large difference in the selectivity of H₂ and CO on different catalyst. The order of CO selectivity is: Cu/Ni > layered Cu/Ni+Fe > mixed Cu/Ni*Fe > supported Cu/Ni/Fe, while the H₂ selectivity had a contrary order. The minimum CO removal over layered Fe+Cu/Ni catalyst is mainly due to that WGSR over Fe/γ-Al₂O₃/Al can’t proceed effectively after SRD over Cu/Ni/γ-Al₂O₃/Al, which results from the non-contact between the active components of Cu/Ni/γ-Al₂O₃ and Fe at the unit level. On the contrary, CO can be reformed to CO₂ and H₂ instantly over the supported Cu/Ni/Fe catalyst. This is probably because of the good synergetic effect of the active components of Cu/Ni/γ-Al₂O₃ and Fe at the atomic level.

To verify the change that occur in the oxidation of active components during SRD, XPS measurements were further performed on the Cu/Ni/Fe/γ-Al₂O₃ catalyst before and after above activity evaluation. “A-catalyst” and “B-catalyst” are used to represent the fresh and spent catalyst, respectively. The photoelectron peaks of the Cu 2p were presented in Figure 6(a). It was found that the A-catalyst exhibited Cu⁺/Cu⁰ and Cu²⁺ at near 932.4 and 934.4 eV, accompanied

by the characteristic Cu²⁺ shakeup satellite peaks (938 - 945 eV) [30] [31] [32] . B-catalyst exhibited the presence of Cu⁺/Cu⁰ at near 932.4 eV, along with a very slight amount of Cu²⁺ species (satellite) at ca. 943 eV. Cu 2p3/2 XPS cannot be used to differentiate the peak at 932.4 eV between Cu⁺ and Cu⁰ species because they are separated by only 0.1 eV and neither would exhibit satellite peaks [33] . Auger Cu LMM spectra were used to verify the presence of Cu⁺ at BE ∼570 eV [33] [34] . Both the XPS and Cu LMM Auger spectra the Cu⁺/Cu⁰ and Cu²⁺ species co-existed in the surface of A-catalyst. The surface compositions of the catalysts were calculated by measuring the areas under the curve, corresponding results were shown in Table 2. After evaluation, a significant increase from

34.5% to 91.1% in the amount corresponding to Cu⁺/Cu⁰ species was found. This observation confirms that the CuO species were reduced to metallic Cu⁺ and Cu⁰ species such as Cu₂O and Cu, which further speculated the Cu⁺ and Cu⁰ species all are the active phase for the SRM process [35] [36] .

The photoelectron peaks of the Fe 2p were presented in Figure 6(b). The peaks of Fe 2p spectra were due to magnetite with contributions from both Fe²⁺ and Fe³⁺ ions. Specifically, Fe²⁺ ions can be assigned to peaks around 710 ~ 711 eV with a satellite around 723 ~ 724 eV, while the peaks around 712 ~ 713 eV with a satellite at 725 ~ 726 eV are assigned to Fe³⁺ ions [37] [38] . In was magnetite, 2/3 of the sites are occupied by Fe³⁺ ions and 1/3 of the sites by Fe²⁺ ions leading to a Fe²⁺/Fe³⁺ ratio in the bulk [37] . The results from Table 2 showed that the Fe²⁺/Fe³⁺ fraction increased from 0.45 on the surface of the A-catalyst to 1.12 over B-catalyst, indicating that Fe₂O₃ and Fe₃O₄ species co-existed in the surface of A-catalyst, and Fe₃O₄ species existed in the surface of B-catalyst. The results further confirmed Fe₃O₄ supported active sites for WGS.

From the above XPS and catalytic performance results of Cu/Ni/Fe/γ-Al₂O₃ catalysts, the synergetic mechanism of the active components (γ-Al₂O₃, Cu or Cu₂O, Fe₃O₄) for SRD and CO in suit removal can be obtained in Figure 7. In the initial stage (Figure 7(a)), DME hydrolysis reaction took place on the γ-Al₂O₃/Al substrate to produce MeOH. Both DME and MeOH were decomposed to CO₂ and H₂ on Cu₂O species. Meanwhile, the produced CO and reactant H₂O were shifted to CO₂ and H₂ on Fe₃O₄ species with the WGSR. The produced H₂ initiated the reduction of the CuO and Fe₂O₃ to Cu₂O/Cu and Fe₃O₄ in turn. And the processes were recycled all the time until the CuO and Fe₂O₃ species on the catalyst could be completely reduced by hydrogen from SRD reaction. In the stabilization phase (Figure 7(b)), SRD was catalyzed on the γ-Al₂O₃/Al substrate and the reduced Cu₂O/Cu species, and then CO was directly transformed to H₂ and CO₂ on Fe₃O₄ specie.

Since Cu/Ni/Fe/γ-Al₂O₃/Al catalyst had excellent catalytic performance, high H₂ selectivity, and low CO selectivity, it was selected to evaluate the thermal stability and the results were shown in Figure 8. It can be seen that Cu/Ni/Fe/γ-Al₂O₃/Al catalyst exhibited a good stability and no deactivation was observed during a 400 h test at 400˚C. And during the whole test process, DME conversion remained at 100%. The selectivity of H₂, CO₂ and CH₄ maintained at ca. 71.2%, 21.5% and 4.3%, while the CO selectivity remained at ca. 3.0%, which met the requirement of inlet CO concentration (3% - 5%) for HT-WGSR. In addition, the addition of

Fe to Cu/Ni/γ-Al₂O₃/Al catalysts not only promoted the in situ CO removal, but also reduced the crystallite size of Cu, thus improved the thermal stability of Cu/Ni/γ-Al₂O₃/Al catalyst.