Excessive carbon dioxide (CO₂) emissions caused by human activities, for example, the consumption of fossil fuels, deforestation, and forest degradation, have been considered the major culprit of climate change and ocean acidification[1–3]. In the past decade, shale gas has set off a global energy revolution. Especially in the United States, shale gas has become the most important source of natural gas. With CO₂ as a soft oxidant, ethane (C₂H₆), the second most abundant component (approximately 10% of the content) of shale gas, can be converted into important raw materials [4]. Generally, the reaction between C₂H₆ and CO₂ occurs via two distinct pathways: ① dry reforming of ethane (DRE) into syngas through cleavage of the C–C bond (C₂H₆ + 2CO₂ → 4CO + 3H₂); and ② oxidative dehydrogenation of ethane (ODHE) into ethylene (C₂H₄) by blocking cleavage of the C–C bond (C₂H₆ + CO₂ → C₂H₄ + CO + H₂O)[5–7]. Syngas, a mixture of hydrogen (H₂) and carbon monoxide (CO), is an important feedstock for fuels and chemicals and is conventionally produced by steam reforming or partial oxidation of natural gas, liquefied gas, naphtha, and so on[8–10] or dry reforming of methane (DRM)[11–14]. However, the above processes are highly endothermic, with high energy consumption, and most DRM catalysts suffer deactivation due to coke formation and active site sintering at high operation temperatures above 1000 K [6,15–17]. A reaction temperature at least 100 K lower for DRE enables the production of syngas and the reduction of catalyst deactivation under milder conditions[5–6].

Ni-based catalysts, especially supported Ni catalysts, are widely used in DRM because of their high catalytic activity [12,18–22]. However, Ni-based catalysts suffer from deactivation due to poor coke resistance and particle sintering. Thus, alloying Ni with transition metals (Co, Ru, Pd, Pt) [21,23–25], developing advanced supports[26–28], and using alkali cations as promoters[29–31] have been investigated to overcome the drawbacks of traditional Ni-based catalysts. Because of the broad application prospects in producing chemicals and fuels at operation temperatures below 900 K, DRE has drawn much attention, and a series of catalysts have been developed, including trimetallic perovskites[32,33], supported Pt-based bimetallic catalysts [5,6,17], and supported Ni composite catalysts[34,35]. The strong metal–support interaction (SMSI) effect has been proven to have a significant impact on the catalytic performance of Ni-based catalysts for DRM and DRE. Ceria (CeO₂) supports with more oxygen vacancies show a stronger SMSI effect, which not only improves the dispersion of Ni species but also enhances the bonding between Ni species and the CeO₂ supports[36,37]. Liu et al. [38] reported the SMSI effect between small Ni nanoparticles (NPs) and partially reduced CeO₂. CO₂ adsorbs and dissociates at oxygen vacancies to generate CO and active oxygen. The synergy between Ni and active oxygen reduces the activation barrier of CH₄ bond dissociation and generates CH_x (x = 2, 3) species on the surface of the Ni/CeO₂ catalyst, making the dissociation temperature of CH₄ as low as 700 K. Lustemberg et al. [39] further proved the important role of Ce³⁺ in the dissociation of C–H bonds. Smaller Ni particles on the CeO₂ support experience larger electronic perturbations, resulting in a more significant binding energy and a lower activation barrier for the first C–H bond cleavage. Recently, Xie et al. [40] investigated the effects of oxide supports for DRE over both reducible and irreducible oxide-supported Pt–Ni bimetallic catalysts. The DRE performance over the PtNi/CeO₂ catalyst was greatly improved because the reducible CeO₂ effectively activated CO₂ and promoted the dissociation of C₂H₆ through a bifunctional Mars–van Krevelen redox mechanism. 

Moreover, the dry reforming process is always accomplished by the reverse water–gas shift (RWGS) reaction with a low activation barrier [13,17,40,41]. It has been proven that the introduction of Al into CeO₂ promotes the formation of oxygen vacancies, which inhibits the RWGS reaction because of the powerful activation and dissociation capacity of CO₂[13,42]. FeO_x has also been reported to be a promoter of supported Ni catalysts for both the RWGS reaction [43] and DRM [44] due to the enhancement of Ni dispersion and the formation of Ni-rich NiFe alloys. Recently, Yan et al. [45] reported the adjustment of active interfacial sites by changing the composition of FeNi active components, which was shown to greatly influence the selectivity toward ODHE or DRE. Herein, we consider an FeNi/CeO₂ catalyst with high ethylene selectivity as the initial system and adjust the distribution of surface active components by introducing Al into the CeO₂ support to obtain highperformance catalysts for DRE. 

In this article, we synthesized a series of FeNi/Al–Ce–O catalysts with an enhanced SMSI effect via a facile sol–gel and impregnation method and investigated the catalytic properties for DRE over supported FeNi catalysts under steady-state reaction conditions. The enhanced SMSI effect between the surface active components and Al–Ce–O supports and its influences were confirmed via X-ray photoelectron spectroscopy (XPS), H₂-temperature programmed reduction (TPR), and energy dispersive X-ray spectroscopy (EDS) element mapping. This work also proposes a possible reaction mechanism and the corresponding adsorbed intermediates over supported FeNi catalysts via in situ Fourier transform infrared (FTIR) spectra under the reaction conditions. We established a relationship between catalytic properties and surface structure over supported FeNi catalysts, which will greatly contribute to the research on related catalytic systems.

2.1.1. Materials

Pluronic^® F127 (EO₁₀₆PO₇₀EO₁₀₆, M_w = 12 600) was purchased from Sigma-Aldrich Chemical Inc. (USA). Ce(NO₃)₃·6H₂O (analytical reagent (AR), 99.0%), Al(NO₃)₃·9H₂O (AR, 99.0%), Ni(NO₃)₃·6H₂O (AR, 98.0%), quartz sand (25–50 mesh, AR, 95.0%), and absolute ethanol (AR, 99.7%) were purchased from Sinopharm Chemical Reagent Co., Ltd. (China). Fe(NO₃)₃·9H₂O (AR, 98.5%) was purchased from Nanjing Chemical Reagent Co., Ltd. (China). All the chemicals were used as received. 

2.1.2. Synthesis of Al–Ce–O supports

Al–Ce–O supports were synthesized via a facile sol–gel process combined with evaporation-induced self-assembly (EISA) in ethanol using F127 as the template, which provides a large specific surface area and superior catalytic performance [13,46–49]. In a typical synthesis, 1.6 g of F127 was dissolved in 40 mL of ethanol at room temperature (RT). A total of 10.0 mmol of metal precursors (Ce(NO₃)₃·6H₂O and Al(NO₃)₃·9H₂O) with an Al molar ratio between 10% and 90% (10% increment of each sample) were added into the above solution with vigorous stirring. The mixture was covered with polyethylene (PE) film and stirred for at least 5 h at RT. The homogeneous sol was then transferred into an oven and underwent solvent evaporation. After aging at 313 and 333 K for 24 h successively, the gel product was dried in an oven at 373 K for another 24 h. Calcination was performed by slowly increasing the temperature from RT to 923 K (2 K·min^–1 ramping rate) and then heating at 923 K for 4 h in air. The CeO₂ and Al₂O₃ supports were synthesized via the same procedure except the mixed metal precursors were replaced with 10.0 mmol of Ce(NO₃)₃·6H₂O (4.34 g) or Al(NO₃)₃·9H₂O (3.75 g), respectively. The yellow or white products were then ground into a powder.

2.1.3. Synthesis of supported FeNi bimetallic catalysts

Supported FeNi bimetallic catalysts were synthesized via an incipient wetness impregnation method over as-synthesized CeO₂, Al₂O₃, and Al–Ce–O supports[6,40,45]. In a typical synthesis, a bimetallic co-impregnation procedure was used to maximize the interaction between the two metals. Precursor solutions were prepared by dissolving 101 mg of Fe(NO₃)₃·9H₂O and 83 mg of Ni(NO₃)₂·6H₂O in a specific amount of deionized water sufficient to fill the pores of 0.981 g of corresponding metal oxide support, the pore volume of which was determined by means of nitrogen adsorption measurements. The solution was added dropwise to the support with thorough stirring. The loadings of bimetallic active components were 1.15 wt% for Fe and 0.40 wt% for Ni to obtain a 3:1 molar ratio of Fe/Ni. The catalyst was then dried at 353 K for 12 h and calcined at 723 K for 4 h with a ramping rate of 2 K·min^–1 from RT to 723 K. Since the FeNi/CeO₂ catalyst with an Fe/Ni molar ratio of 3:1 shows high ethylene selectivity due to its special Ni–FeO_x interfacial sites [45], such active metal components would better reflect the enhanced SMSI effect of Al–Ce–O supports on the catalytic properties. 

Powder X-ray diffraction (XRD) analysis was performed on a Bruker D8 Advance SS diffractometer (Bruker Corporation, USA) operated at 40 kV and 40 mA with a slit of 0.5° at a 2θ scanning speed of 2°·min^–1 under a Cu-Kα source (0.15432 nm). Nitrogen adsorption–desorption isotherms were measured at 77 K using a Micromeritics ASAP 2010 analyzer (Micromeritics Instrument Corporation, USA). The Brunauer–Emmett–Teller (BET) specific surface area was measured after degassing the samples at 373 and 623 K for 3 h successively under vacuum. The elemental composition of the supported FeNi catalysts was determined by inductively coupled plasma-atomic emission spectrometry (ICP-AES; Varian Vista AX, Varian Inc., USA). ²⁷Al magic-angle spinning nuclear magnetic resonance (MAS NMR) spectroscopy was conducted with a Varian VNMRS-400WB nuclear magnetic resonance instrument (Varian Inc., USA) with a frequency of 104.18 MHz, a spinning speed of 10 000 Hz, and a relaxation delay of 4 s. Chemical shift values are reported with respect to KAl(SO₄)₂·12H₂O as the standard. XPS was performed on a Thermo ESCALAB 250 spectrometer (Thermo Fisher Scientific Inc., USA) with a monochromatic Al-Kα X-ray source (1486.6 eV, 1 eV = 1.602176 × 10^–19 J) and an analyzer pass energy of 20 eV. The C 1s line at 284.6 eV was used to calibrate the binding energies (BEs) of the measured elements.

H₂-TPR experiments were conducted on a Micromeritics AutoChem II 2920 chemisorption analyzer (Micromeritics Instrument Corporation, USA). In a typical experiment, 50 mg of the assynthesized catalyst was put into a U-shaped quartz tube and pretreated in a 50 mL·min^–1 He flow at RT. Then, the TPR experiment was performed under a 10 vol% H₂/Ar mixture at a space velocity of 50 mL·min^–1 with a ramping rate of 10 K·min^–1 from RT to 1173 K. Pulse CO chemisorption experiments were also performed on a Micromeritics AutoChem II 2920 chemisorption analyzer to measure the CO uptake value of the catalyst. Approximately 150 mg of the catalyst was first reduced at 873 K in a 10 vol% H₂/Ar flow for 30 min. Then, the reduced catalyst was purged in a He flow until the temperature was decreased to 313 K. The loop gas of 10% CO/He (590 μL) was pulsed with a He stream over the catalyst until the peak area of CO became constant. The CO uptake values of the catalysts could provide an approach to estimate the turnover frequency (TOF), as reported previously[17,33,45]. 

Transmission electron microscopy (TEM), high-resolution transmission electron microscopy (HRTEM), high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM), and EDS element mapping were conducted on an FEI Talos F200X TEM (Thermo Fisher Scientific Inc., USA) with a probe aberration corrector operating at 300 kV. The TEM samples were prepared by drying a drop of the sample dispersion in ethanol on carboncoated copper grids.

In situ FTIR spectra were obtained on a Bruker Vertex 70 V FTIR spectrometer (Bruker Corporation, USA) with a stainless steel highvacuum transmission infrared cell. The samples were pressed on a tungsten mesh support and heated to 623–673 K at a ramping rate of 10 K·min^–1 under vacuum for 2 h to remove the surface adsorbed water. The background spectra were collected after the tungsten mesh support was cooled to RT. In a typical in situ infrared (IR) experiment under the reaction conditions, 1.0 mbar (1 mbar = 100 Pa) of C₂H₆ and 1.0 mbar of CO₂ were introduced into the cell, and IR spectra of each sample were collected in the temperature range of 373–873 K. In situ CO adsorption IR spectra of the FeNi/Ce–Al_0.5 catalyst were collected under a CO pressure of 5.0 mbar from 373 to 573 K and 1.0 mbar from 673 to 873 K.

The catalytic performance of supported FeNi bimetallic catalysts was evaluated in a continuous-flow fixed-bed quartz tabular reactor (7.5 mm inner diameter) under atmospheric pressure, utilizing a mixture of 100 mg of catalyst (20–40 mesh) and 100 mg of quartz sand loading between two quartz wool plugs at the center of the reactor. The catalyst was pretreated in situ with 40 mL·min^–1 H₂ at 673 K for 1 h and then heated to 873 K at a ramping rate of 5 K·min^–1 with a constant total flow rate of C₂H₆, CO₂, and N₂ of 40 mL·min^–1 . The volume ratio of the C₂H₆, CO₂, and N₂ mixed gas was 1:1:2. The temperature of the catalyst bed was held at 873 K for 8 h, and the outlet stream was analyzed online using a gas chromatograph (Agilent 6820B, Agilent Technologies, Inc., USA) with a thermal conductivity detector (TCD). Water within the outlet stream was removed by a condenser. N₂ was used as an internal standard to account for the volume effects due to the high temperature during the reaction. Blank experiments without supported FeNi catalysts were conducted at 873 K to evaluate the contribution of the gas-phase reaction and system, and the results showed a negligible effect on DRE performance. In this article, the steady-state conversion (X), TOF, C₂H₆-based selectivity (S), and yield (Y) of species i were defined using the following equations:

where F_in and F_out are the inlet and outlet flow rates of the reactant (mol·min^–1 ), U_CO is the CO uptake value (mol·g^–1 ), m_cat is the mass of the catalyst (g), and N_C is the carbon atom number of the products.

Powder XRD patterns of the as-synthesized catalysts with standard Joint Committee on Powder Diffraction Standards (JCPDS) cards are shown in Fig. 1. The diffraction peaks of FeNi/Ce–Al_x (x ≤ 70%) and FeNi/CeO₂ confirm the fluorite cubic structure of CeO₂ (, JCPDS 75-0120), while the pattern of FeNi/Al₂O₃ is consistent with the structure of η-Al₂O₃ (, JCPDS 77-0396). FeNi/ Ce–Al_0.9 shows significant phase separation into CeO₂, η-Al₂O₃, and Al(OH)₃ (P1(1), JCPDS 24–0006). No peaks attributed to Fe and Ni are observed in these patterns. As shown in Table 1, the average crystallite size of the catalysts calculated by the Scherrer equation [50] significantly decreases with increasing Al content, which indicates that Al improves the sintering resistance of the CeO₂ supports [47]. The size distributions of supported FeNi catalysts were further estimated by the size statistics of NPs from TEM images. The TEM and HRTEM images of supported FeNi catalysts are shown in Figs. S1 and S2 in Appendix A. The size distributions of the FeNi/CeO₂ and FeNi/Ce–Al_x (10% ≤ x ≤ 50%) catalysts in Table 1 are close to the corresponding average crystallite sizes. As shown in Table 1 and Fig. S3 in Appendix A, the BET specific surface area of supported FeNi catalysts is positively related to the Al content. The average crystallite size of the FeNi/Ce–Al_x (70% ≤ x ≤ 90%) catalysts becomes even smaller upon formation of a mesoporous structure, as shown in Fig. S2, which reveals the relationship between the size distribution and BET specific surface area of supported FeNi catalysts. In Figs. S1 and S2, lattice spacings of 3.12, 2.71, 1.91, and 2.38 Å correspond to the (111), (200), and (220) facets of CeO₂ and (311) facet of η-Al₂O₃, respectively. The lattice spacing of FeNi/Ce–Al_0.9 is indistinguishable owing to the poor crystallinity and severe phase separation. 

Fig. 1. XRD patterns of as-synthesized supported FeNi catalysts. The standard XRD patterns of CeO₂ (, JCPDS 75-0120), η-Al₂O₃ (, JCPDS 77-0396), and Al(OH)₃ (P1(1), JCPDS 24-0006) are shown in the panel as a reference. a.u.: arbitrary unit.

Table 1 Average crystallite size (XRD size), crystal microstrain, lattice parameter, TEM size distribution data, and BET surface area (S_BET) of supported FeNi catalysts.

^a Calculated (440) diffraction peaks for FeNi/Al₂O₃ and FeNi/Ce–Al_0.9 and (111) diffraction peaks for other supported FeNi catalysts.

^b Estimated via the single line method for analysis of XRD line broadening using a pseudo-Voigt profile function.

To further discuss the microstructure of the catalysts, the microstrains and lattice parameters were calculated, and the results are listed in Table 1. The microstrain in the lattice (lattice strain) of the samples was estimated via the single line method for analysis of XRD line broadening using a pseudo-Voigt profile function[51,52]. As shown in Fig. 1, for the FeNi/CeO₂ and FeNi/ Ce–Al_x (10% ≤ x ≤  70%) catalysts with fluorite cubic structures, the microstrain in the crystal lattice of the oxide supports increases as a function of Al content to maintain the original crystal structure. The lattice distortion is relieved by a decrease in the microstrain and phase transition after the introduction of a high content of Al. The lattice parameters determined via Bragg’s law from the (111) diffraction peak of CeO₂ (, JCPDS 75-0120) for FeNi/CeO₂ and FeNi/Ce–Al_x (10% ≤ x ≤ 70%) as well as the (440) peak of η-Al₂O₃ (, JCPDS 77-0396) for FeNi/Ce-Al_0.9 and FeNi/Al₂O₃ are also listed in Table 1. Fig. S4 in Appendix A shows the ²⁷Al MAS NMR spectra of FeNi/Ce–Al_x (10% ≤ x ≤ 50%) and of FeNi/Al₂O₃. The peaks at approximately 8 and 66 ppm are assigned to octahedrally (Al_oct) and tetrahedrally (Al_tet) coordinated Al³⁺, while the Al species with a chemical shift of 38 ppm is a CeO₂ lattice occupied by Al³⁺ [53]. The increased peak intensity at 38 ppm for FeNi/Ce–Al_0.1 and FeNi/Ce–Al_0.3 is a result of Al³⁺ ions present in the CeO₂ lattice. The similarity of the spectra of FeNi/Ce– Al_0.5 and FeNi/Al₂O₃ indicates a stable octahedral coordination of Al³⁺ species with a high content of Al in FeNi/Ce–Al_x (50% ≤ x ≤ 90%). Regardless of the phase transition, the change in the lattice parameter of the supports is mainly related to two factors: ① the formation of oxygen vacancies by replacing Ce⁴⁺ with Al³⁺, which leads to crystal lattice shrinkage, and ② the transition from Ce⁴⁺ to Ce³⁺ with a corresponding reduction in the ionic radius, which is essential to balance the electric charge of the unit cell. The change in the lattice parameter is thought to be a result of the synergistic effect of the two factors: the formation of surface oxygen vacancies and Ce³⁺ species. 

The steady-state catalytic performance of the FeNi/Al–Ce–O catalysts at 873 K is shown in Figs. 2 and 3, where the FeNi/CeO₂ and FeNi/Al₂O₃ catalyst data were also plotted as a reference. The experimental data indicate that the composition of oxide supports plays an important role in the catalytic properties for DRE over supported FeNi catalysts. As shown in Figs. 2(a) and (b), C₂H₆ and CO₂ conversion is positively correlated with the Al content (0 ≤ x ≤ 50%), whereas completely opposite trends are observed in Figs. 3(a) and (b) when the Al content is above 50%. In Fig. S5 in Appendix A, the CO selectivity from C₂H₆ also significantly increases with increasing Al content (0 ≤ x ≤ 30%), while the ethylene selectivity decreases correspondingly. As the Al content changes between 30% and 90%, the CO selectivity of the FeNi/ Al–Ce–O catalysts remains stable at 96%–98%. In Figs. 2(c) and 3(c), the CO yield from C₂H₆ over the FeNi/Al–Ce–O catalysts follows the same trend of first increasing and then decreasing with increasing Al content. The FeNi/Ce–Al_0.5 catalyst provides the best DRE performance, with the highest C₂H₆ and CO₂ conversions and CO selectivity and yield. The introduction of Al into CeO₂ possibly enhances the interaction between the surface active components and the Al–Ce–O support, which further affects the catalytic properties over supported FeNi catalysts. 

Fig. 2. (a) C₂H₆ conversion, (b) CO₂ conversion, (c) CO yield, and (d) H₂/CO molar ratio over FeNi/Ce–Al_x (10% ≤ x ≤ 50%) and FeNi/CeO₂ catalysts under steady-state reaction conditions at 873 K and 1 atm (1 atm = 1.01325 × 10⁵ Pa) with a flow rate of 40 mL·min^–1 (C₂H₆:CO₂:N₂ = 1:1:2) and gas hourly space velocity (GHSV) of 24 000 mL·(h·g)^–1 .

Fig. 3. (a) C₂H₆ conversion, (b) CO₂ conversion, (c) CO yield, and (d) H₂/CO molar ratio over FeNi/Ce–Al_x (50% ≤ x ≤ 90%) and FeNi/Al₂O₃ catalysts under steady-state reaction conditions at 873 K and 1 atm with a flow rate of 40 mL min1 (C₂H₆:CO₂:N₂ = 1:1:2) and GHSV of 24 000 mL·(h·g)^–1 .

The average catalytic performance data of the supported FeNi catalysts between 420 and 480 min are summarized in Table 2. After several hours of steady-state reaction, the C₂H₆ and CO₂ conversion, CO selectivity, and CO yield over the supported FeNi catalysts maintain the same relative order. The FeNi/Ce–Al_0.5 catalyst exhibits the best DRE performance with the highest C₂H₆ conversion (11.7%), CO₂ conversion (33.1%), and CO yield (11.5%). The TOF values based on CO uptake also indicate the outstanding catalytic activity of the FeNi/Ce–Al_0.5 catalyst for both C₂H₆ (47.1 min^–1 ) and CO₂ (133.1 min^–1 ). As a comparison, the catalytic performance data of recently reported DRE catalysts are listed in Table S1 in Appendix A. The FeNi/Ce–Al_0.5 catalyst shows high TOFs and CO selectivity similar to other high-performance DRE catalysts, whereas the conversions are possibly restricted by the low loading of bimetallic active components. Catalysts with Al contents above 50% show lower TOF values for both C₂H₆ and CO₂ in this reaction, which demonstrates that the enhancement of DRE performance over the FeNi/Ce–Al_0.5 catalyst should be attributed not only to the increasing Al content but also to the interaction between surface active components and the Al–Ce–O support. H₂/CO molar ratios lower than 0.75 due to the side reaction of the RWGS are also shown in Table 2 [33]. The highest H₂/CO ratio of the FeNi/Ce–Al_0.5 catalyst indicates that the SMSI effect enhanced by the introduction of Al partially inhibits the RWGS process in this reaction, which is also related to the surface oxygen vacancy over the FeNi/Al–Ce–O catalysts.

Table 2 Summary of CO uptake and catalytic performance over supported FeNi catalysts under steady-state reaction conditions averaged between 420 and 480 min at 873 K.

The elementary composition and chemical valence on the surface of hydrogen-reduced catalysts were detected via XPS. Typical Ce 3d and Al 2p core level spectra of the FeNi/CeO₂, FeNi/Al–Ce–O, and FeNi/Al₂O₃ catalysts are shown in Appendix A Fig. S6. During typical data processing, the complex spectra of the samples in the Ce 3d region were deconvoluted into ten components via the generally accepted approach of extracting the ratio of Ce³⁺ and Ce⁴⁺[54,55]. The ten peaks contained five spin-orbital split pairs of Ce 3d_5/2 (v_i : v₀, v, v' , v'', and v''') and Ce 3d_3/2 (u_i : u₀, u, u' , u'', and u'''), of which the area intensities, the full widths at halfmaximum (FWHM), and the position distances were fixed as constants during the deconvolution. Herein, the peak positions are marked in Fig. S6(a); the relative contents of Ce³⁺ to the total Ce content (c_Ce)in the samples were calculated via the following equation, and the results are listed in Table 3: 

where is the content of Ce³⁺, and I is the area intensity of the given component. As shown in Table 3, the surface relative content of Ce³⁺ increases significantly with Al content. Shyu et al. [56] reported that the area under the u''' peak in the total Ce 3d region could be used to describe the relative content of Ce⁴⁺ in the samples. In Table 3, it is observed that the area under the u''' peak shows a negative relationship with the Al content, which also confirms the correlation above. Herein, the theoretical effective surface area (TESA, S_eff) of supported FeNi catalysts is defined by the following equation:

where S_BET is the BET specific surface area of the catalyst, is the relative surface content of Ce³⁺, and P_Ce is the total surface content of Ce of the catalyst. According to the data in Table 3, as the Al content increases, the TESA shows the same trend as the conversions and TOFs of C₂H₆ and CO₂, indicating that the reactivity of C₂H₆ with CO₂ is closely related to the content of surface Ce³⁺ species over supported FeNi catalysts. Moreover, the dispersion of surface active components should be another important factor to be discussed later. In Fig. S6, the binding energy of the Ce 3d core level of the FeNi/Al–Ce–O catalysts decreases slightly with increasing Al content compared with that of FeNi/CeO₂. In addition, the peaks of the Al 2p core level of the FeNi/Al–Ce–O catalysts shift to lower binding energies than that of FeNi/Al₂O₃ with increasing Ce content, which demonstrates electron transfer between Ce or Al and adjacent atoms. 

Fig. S7 in Appendix A shows the O 1s and Fe 2p XPS spectra of the reduced FeNi catalysts. In Fig. S7(a), the binding energies at approximately 529.2 and 531.7 eV are ascribed to the lattice oxygen of Ce-based oxides (O_I) and the adsorbed oxygen or hydroxyl groups (O_II) on the surface, respectively[41,57,58]. The O 1s core level binding energy of Al₂O₃ is located at 530.9 eV [57]. The O_II/O_I ratios in Table 3 increase gradually with increasing Al content, which indicates an increase in surface oxygen vacancies over the FeNi/Al–Ce–O catalysts[13,41]. In Fig. S7(b), the Fe 2p core level binding energies of the reduced FeNi catalysts at approximately 710.9 and 724.0 eV are ascribed to Fe₂O₃, which means that the surface Fe species should be highly oxidized during the reaction [57]. Raman spectroscopy was also conducted to determine the surface oxygen vacancy of the catalysts. Fig. S8 in Appendix A shows the Raman spectra of the supported FeNi catalysts excited by a 532 nm laser. The strong band at approximately 462 cm^–1 is ascribed to the F_2g vibration mode of the Ce–8O vibrational unit of the fluorite structure, while the weak bands at approximately 254 and 596 cm^–1 are attributed to the second-order transverse acoustic (2TA) mode and the defect-induced (D) mode of oxygen vacancies, respectively. The relative intensity ratio I_D/I_F2g reflects the content of oxygen vacancies[59,60]. As seen in Fig. S8, the intensity ratio of I_D/I_F2g increases slightly with Al content, which also indicates that the introduction of Al improves the content of surface oxygen vacancies and the inhibition of the RWGS reaction over the FeNi/Al–Ce–O catalysts. According to the results above, the introduction of Al into the CeO₂ support leads to a higher density of surface Ce³⁺ species and oxygen vacancies, which further improves the catalytic performance for DRE over FeNi/Al–Ce–O catalysts.

Table 3 Relative surface content of Ce³⁺, area under the u''' peak, total surface content of Ce, TESA (S_eff), and O_II/O_I ratio of O 1s core level over supported FeNi catalysts.

The H₂-TPR profiles of the as-synthesized FeNi catalysts are shown in Fig. 4. To assign the peaks in the pattern, profiles of the as-synthesized Ce–Al_0.5 supported monometallic catalysts and pure CeO₂ support are also shown in Fig. S9 in Appendix A as a comparison. As seen in Fig. S9, CeO₂ reduction can be divided into two stages. The first short and wide peak located between 600 and 800 K is assigned to the reduction of surface active oxygen of CeO₂, which leads to the formation of surface oxygen vacancies and nonstoichiometric CeO_x, and the second peak above 800 K is attributed to bulk CeO₂ reduction[61–63]. In Fig. 4, after adding surface active components to the support, a strong peak at approximately 550 K appears for the FeNi/CeO₂ catalyst. This peak can be attributed to the reduction of both surface active components and the CeO₂ support by surface hydrogen spillover [64]. Compared with the patterns of monometallic catalysts in Fig. S9, for the Ni/Ce–Al_0.5 catalyst, the peak below 673 K shows a lower reduction cutoff temperature than that of Fe/Ce–Al_0.5, which indicates that surface Ni species are much easier to reduce than Fe on Ce-based composite oxides. For the FeNi/Al–Ce–O catalysts, compared with the pattern for FeNi/CeO₂ in Fig. 4, the peak broadening and cutoff temperature rise also indicate that the introduction of Al significantly enhances the SMSI effect between surface active components and oxide supports.

Fig. 4. H₂-TPR profiles of as-synthesized FeNi catalysts between 298 and 1173 K.

To identify the dispersion of surface active components on the composite support, EDS elemental mapping measurements were conducted on three representative samples: FeNi/Ce–Al_0.1, FeNi/ Ce–Al_0.5, and FeNi/Ce–Al_0.9. The element mapping images in this article are representatively chosen from many different regions of the samples. As shown in the EDS mapping images of Ce and Al in Figs. 5–7, Ce and Al are well distributed over the FeNi/ Al–Ce–O catalysts. Nevertheless, the elemental distributions of Fe and Ni are quite different. Small bimetallic FeNi NPs are observed on the surface of FeNi/Ce–Al_0.1, as confirmed by the EDS mapping images shown in Fig. 5. Bimetallic FeNi NPs with similar structures have been proven to have high selectivity for ethylene [45]. As demonstrated by the gradual FeNi distribution changes shown in Figs. 6 and 7, as the Al content increases, the surface Fe and Ni species become well dispersed on the Al–Ce–O supports. The introduction of Al greatly increases the interaction between the surface active components and the composite support. The surface Fe and Ni species are dispersed randomly and independently throughout the support because of the enhanced SMSI effect, leading to peak broadening and an increase in the reaction cutoff temperature of H₂-TPR over the supported FeNi catalysts, as shown in Fig. 4.

Fig. 5. (a) HAADF-STEM image. EDS mapping images: (b) overlay, (c) Ce, (d) Al, (e) Fe, and (f) Ni of the reduced FeNi/Ce–Al_0.1 catalyst with a scale bar of 10 nm.

Fig. 6. (a) HAADF-STEM image. EDS mapping images: (b) overlay, (c) Ce, (d) Al, (e) Fe, and (f) Ni of the reduced FeNi/Ce–Al_0.5 catalyst with a scale bar of 10 nm.

Fig. 7. (a) HAADF-STEM image. EDS mapping images: (b) overlay, (c) Ce, (d) Al, (e) Fe, and (f) Ni of the reduced FeNi/Ce–Al_0.9 catalyst with a scale bar of 20 nm.

To further investigate the surface active species during the DRE reaction, in situ IR spectroscopy studies were carried out at temperatures ranging from 373 to 873 K and a total pressure of 2.0 mbar (C₂H₆:CO₂ = 1:1). In situ IR spectra of the FeNi/Ce–Al_0.1, FeNi/Ce–Al_0.5, and FeNi/Ce–Al_0.9 catalysts are shown in Figs. S10–S12 in Appendix A. All the spectra were normalized by subtracting the corresponding IR spectrum under vacuum at RT. A typical IR spectrum is divided into three different characteristic vibrational regions that will be discussed individually.

The in situ IR spectra in the region of 3900–3500 cm^–1 in Fig. 8 provide information on surface hydroxyl and carbonate species. The wide band at 3770–3790 cm^–1 and the strong band at approximately 3706 cm^–1 are ascribed to the monocoordinated OH groups (Type I OH) of Al and Ce, respectively[65–68]. The band at approximately 3732 cm^–1 is mainly ascribed to the Type II-A OH species of Al (hydroxyl groups bibridged across Al–Al ion pairs) with a possible contribution of terminal OH groups bound to surface Ce⁴⁺ cations [65]. The band located at approximately 3625 cm^–1 is thought to be a combination of two bands: ① the Type II-B OH species of Ce with adjacent oxygen vacancies (O–Ce–OH–Ce–□) at 3630 cm^–1 ; and ② the surface bicarbonate (HCO₃^– ) species at 3619 cm^–1 , as confirmed by the delay of the band at approximately 3706 cm^–1[67,68]. The presence of Type II-B OH species and the absence of Type II-A OH species of Ce indicate that the surface of the FeNi/Al–Ce–O catalysts is highly active with oxygen vacancies under the reaction atmosphere. The band at approximately 3598 cm^–1 can also be separated into two bands: the tribridged OH species (Type III OH) of Ce at approximately 3600 cm^–1 and surface protonated carboxylate species (–COOH) at approximately 3593 cm^–1[67–70]. The formation of surface carbonate species will be discussed in detail below. As seen in Figs. 2(d) and 3(d), as the reaction temperature increases, the decreased intensity of the OH band over the FeNi/Ce–Al_0.5 catalyst implies a reduction in H₂O production, which further leads to lower RWGS activity and a higher H₂/CO ratio than those of other supported FeNi catalysts. 

Fig. 8. In situ IR spectra of (a) FeNi/Ce–Al_0.1, (b) FeNi/Ce–Al_0.5, and (c) FeNi/Ce–Al_0.9 in the region of 3900–3500 cm^–1 at temperatures ranging from 373 to 873 K and a total pressure of 2.0 mbar (C₂H₆:CO₂ = 1:1).

The bands in the region of 3150–2750 cm^–1 correspond to the CH stretching bands of adsorbed species. The strong wide bands at approximately 3005 and 2931 cm^–1 in Fig. 9 can be ascribed to the antisymmetrical () and symmetric () CH stretching vibrations of a series of methyl species in the gas phase [71]. The CH vibration bands in this region indicate the presence of adsorbed ethyl ( at 2970 cm^–1 , at 2931 cm^–1 , and  at 2880 cm^–1 )[72,73] and ethanol ( at 2977 cm^–1 , at 2933 cm^–1 , and at 2878 cm^–1 )[73–75]. The sharp band at 2953 cm^–1 is ascribed to the CH vibration band of bridged formate species, while the other sharp band at 2895 cm^–1 is attributed to  the CH stretching band of bidentate formate[67,76,77]. Little difference in the spectra of the three catalysts is observed in the temperature range from 373 to 873 K in the region of 3150–2750 cm^–1 , which means that the FeNi/Al–Ce–O catalysts have the same kind of surface CH-containing species during the reaction, independent of the Al content and reaction temperature. 

Fig. 9. In situ IR spectra of (a) FeNi/Ce–Al_0.1, (b) FeNi/Ce–Al_0.5, and (c) FeNi/Ce–Al_0.9 in the region of 3150–2750 cm^–1 at temperatures ranging from 373 to 873 K and a total pressure of 2.0 mbar (C₂H₆:CO₂ = 1:1).

Fig. S13 in Appendix A shows the in situ IR spectra in the region of 1800–1000 cm^–1 , where the peaks are mainly ascribed to the carbonate-like (OCO) species adsorbed on the samples [67,68,77– 79]. The complex band assignments of different carbonate, carboxylate, and formate species adsorbed on the supported FeNi catalysts are summarized in Table S2 in Appendix A. The peaks attributed to the corresponding carbonate-like species are marked in the original spectra of FeNi/Al–Ce–O in Fig. 10. The band distribution in the IR spectra over the FeNi/Ce–Al_0.1 catalyst in Fig. 10(a) is quite similar to those of the other two samples in Figs. 10(b) and (c), except for the bands at 1430–1425, 1236–1217, and 1057– 1050 cm^–1 . The CO adsorption IR spectra of the FeNi/Ce–Al_0.5 catalysts are shown in Fig. 10(d) for comparison. The missing bands at 1430–1425 cm^–1 in both Figs. 10(a) and (d) are attributed to the intermediate adsorbed species of C₂H₆. Since it has been reported that the bands at 1580 cm^–1 (), 1429 cm^–1 (), 1306 cm^–1 (), and 1026 cm^–1 ( ) are the characteristic peaks of acetate species, an oxidation product of C₂H₆ on CeO₂ at 355 K [74], the increased IR spectral intensity for the FeNi/Ce–Al_0.5 and FeNi/Ce–Al_0.9 catalysts at 1430–1425 cm^–1 can be attributed to the formation of surface acetate species. The bands at 1660– 1640 cm^–1 and approximately 1230 cm^–1 arising from the olefinic C=C and CH stretching vibrations imply the formation and adsorption of ethylene on the FeNi/Ce–Al_0.1 catalyst, rather than surface acetate species [80]. Thus, it can be inferred that the different product selectivity of the FeNi/Al–Ce–O catalysts results from the changes in the surface adsorbed species. Moreover, the bands at 1057–1050 cm^–1 in Fig. 10(a) below 673 K are attributed to the CO stretching vibration of bidentate ethoxide and methoxide species[74,75]. Since the formation of the ^*C₂H_yO intermediate has proven to be essential for C–C bond cleavage and syngas production [7], the reduced band intensity at high reaction temperatures indicates that the weak adsorption of surface bidentate ethoxide species on the FeNi/Ce–Al_0.1 catalyst is beneficial to the formation of ethylene.

Fig. 10. Original in situ IR spectra of (a) FeNi/Ce–Al_0.1, (b) FeNi/Ce–Al_0.5, and (c) FeNi/Ce–Al_0.9 in the region of 1800–1000 cm^–1 at temperatures ranging from 373 to 873 K and a total pressure of 2.0 mbar (C₂H₆:CO₂ = 1:1). (d) In situ IR spectra of CO adsorption over the FeNi/Ce–Al_0.5 catalyst in the region of 1800–1000 cm^–1 with 5.0 mbar CO from 373 to 573 K and 1.0 mbar CO from 673 to 873 K. Vac: vacuum.

On the basis of the discussion above, a typical catalytic cycle of CO₂ over FeNi/Al–Ce–O catalysts involves the following process, as shown in Fig. 11. CO₂ in the gas phase first adsorbs onto the surface hydroxyl species or oxygen vacancies and generates adsorbed carbonate-like species, such as bicarbonate and carboxylate. In addition, C₂H₆ adsorbs on metallic or oxidized FeNi active sites and dissociates into ethyl or ethoxy groups and a hydrogen atom. As a result of surface hydrogen spillover, adsorbed bicarbonate or carboxylate species are reduced to carboxylic acid or formate species, which further decompose into surface hydroxyl and carbonyl species through a possible formyl transition intermediate [77]. Surface hydroxyl species or oxygen vacancies regenerate after the release of CO to the gas phase. Nevertheless, the oxidation of C₂H₆ involves two different paths determined by the dispersion of FeNi active components. The impregnation of Fe and Ni precursors on pure CeO₂ tends to generate bimetallic FeNi NPs, which prevents the excessive oxidation of adsorbed ethyl or ethoxy species and improves the selectivity of ethylene [45]. The introduction of Al into the lattice of CeO₂ greatly improves not only the content of surface Ce³⁺ species and oxygen vacancies but also the dispersion of surface active components through the enhanced SMSI effect. The strong interaction between FeNi and the Al–Ce–O support stabilizes the adsorbed ethoxy moiety and its further oxidation products, which are essential for C–C bond cleavage and syngas generation.

Fig. 11. Typical catalytic cycle of CO₂ over FeNi/Al–Ce–O catalysts.

FeNi/Al–Ce–O catalysts synthesized via a facile sol–gel and impregnation method exhibit a composition-induced SMSI effect for DRE. The Al content in the Al–Ce–O supports significantly influences the metal–support interface structure of the catalysts and further determines the catalytic properties during the reaction. As the Al content increases, the C₂H₆ and CO₂ conversion, CO selectivity and yield, and TOF first increase and then decrease according to the same trend as the TESA. The FeNi/Ce–Al_0.5 catalyst exhibits the best DRE performance with the highest C₂H₆ conversion (11.7%), CO₂ conversion (33.1%), and CO yield (11.5%). The increased surface oxygen vacancy partially inhibits the RWGS reaction over FeNi/Ce–Al_0.5 catalysts, which leads to a higher H₂/CO ratio than that of other FeNi/Al–Ce–O catalysts. The selectivity over the supported FeNi catalysts is determined by the dispersion of the surface active components. As the Al content in the Al–Ce–O supports increases, the dispersion of surface active components is promoted by the enhanced SMSI effect over the supported FeNi catalysts. The enhanced SMSI effect stabilizes the adsorbed ^*C₂H_yO intermediate and produces excessive oxidation products, leading to C–C bond cleavage and syngas generation. In summary, the introduction of Al into the CeO₂ support not only increases the content of surface Ce³⁺ and oxygen vacancies but also promotes the dispersion of surface active components, which further adjusts the catalytic properties for DRE over supported FeNi catalysts.

The authors gratefully acknowledge the support from the National Key Research and Development Program of China (2017YFB0702800), the China Petrochemical Corporation (Sinopec Group), and the National Natural Science Foundation of China (91434102 and U1663221).

Tao Zhang, Zhi-Cheng Liu, Ying-Chun Ye, Yu Wang, He-Qin Yang, Huan-Xin Gao, and Wei-Min Yang declare that they have no conflicts of interest or financial conflicts to disclose.

Supplementary data to this article can be found online at https://doi.org/10.1016/j.eng.2021.11.027.