In view of the potential impact of CO2 on climate change, its atmospheric concentration needs to be controlled and stabilized. Carbon capture and utilization (CCU) technologies, including the catalytic valorization of CO2, can highly contribute to achieving this goal  . Among the different processes, CO2 methanation, i.e. its hydrogenation to form CH4, stands as a very promising technology, which also offers a solution for off-peak renewable energy storage  . Though thermodynamically feasible even at ambient temperature, CO2 methanation is considerably hindered by its extremely slow reaction kinetics. The use of active catalysts is imposed, moreover, undesirable byproducts such as CO start to be produced at temperatures higher than 350˚C. Unfortunately, most of the commercially existing catalytic systems start to be active at this temperature.
Noble metal-based catalysts (Ru, Rh, and Pd) are well known to be active in CO2 methanation   . Ni-based catalysts display somehow lower but still acceptable activity being substantially cheaper than noble-metal based ones   . Moreover, Ni-containing catalysts suffer from deactivation by: 1) sulfur poisoning; 2) carbon deposition; and 3) Ni-phase sintering  . Besides, it was reported that the stability of the nickel on the Al2O3 carrier is much higher than on the other carriers  , and the Al2O3 has a strong interaction with NiO, which may promote the formation of NiAl2O4 spinel phase  .
Since Levy et al. reported that tungsten carbides displayed similar activity as Pt in neo-pentane isomerization  . These materials have been subject of growing interest, since they can be employed in many other catalytic reactions  . Indeed, the activity of molybdenum and tungsten carbides (Mo2C and WC) in dry reforming of methane, partial oxidation and stream reforming of methane to synthesis gas was found to be higher than Pt and Pd-based catalysts, though still lower than the activity measured for Ru and Rh catalysts  . Shi et al. reported high catalytic activity and stability for a Ni-Mo2C catalyst in dry reforming of methane  . The activity of Mo2C and WC was linked to their facility to activate the extremely stable CO2 molecule. They can be also used as hydrogenation catalysts and, in fact, Huo and co-workers recently reported interesting activity, selectivity and stability in CO methanation for Co-supported on Mo carbide  . Mo2C- and/or WC-based catalysts can be therefore promising materials, able to boost the CO2 methanation reaction. However, to the best of our knowledge, there are no studies considering the use of Mo2C-supported catalysts for this particular application.
In our previous work, we found carbide Ni-Mo/Al2O3 catalyst (Ni-Mo2C/ Al2O3) was a better catalyst for dry reforming of methane than that of reduced Ni-Mo/Al2O3 catalyst  . It was investigated the influence of preparation condition (the ratio of H2/CH4) for the carburization process in detail in the previous work  .
The present work considers the preparation and characterization of Ni-Mo2C/ Al2O3 catalysts for CO2 methanation. These catalysts aim to combine the well-known high catalytic activity of Ni, together with the promoting features of Mo2C. The activity of the carbide-containing catalysts was compared to that of bimetallic Ni-Mo catalysts, proving an important promotion effect of the presence of Mo2C.
2.1. Catalysts Preparation
2.1.1. Mo/Al2O3 and Ni-Mo/Al2O3 Catalysts
Mo/Al2O3 and Ni-Mo/Al2O3 catalysts were prepared through excess solvent impregnation. Mo and Ni precursors were used, (NH4)6Mo7O24・4H2O (Sigma-Aldrich) and Ni(NO3)2・6H2O (Sigma-Aldrich), corresponding to nominal loadings of 10 wt.% of each metal. The support, γ-Al2O3, was obtained through air calcination of a commercially available boehmite (Disperal, Sasol), at 500˚C for 4 h. After 2 h impregnation, the excess solvent (deionized water) was removed in a rotary evaporator at 60˚C. Then the catalyst was dried in an oven at 110˚C overnight, and finally calcined in synthetic air at 550˚C for 4 h.
2.1.2. Mo2C/Al2O3 and Ni-Mo2C/Al2O3 Catalysts
A temperature-programmed method in CH4/H2 atmosphere ( = 10 mL/min and = 40 mL/min) was followed, in order to obtain the Mo2C/Al2O3 and Ni-Mo2C/Al2O3 catalysts   . Both Mo2C/Al2O3 and Ni-Mo2C/Al2O3 catalysts was obtained by the carburization of the Mo/Al2O3 and Ni-Mo/Al2O3 catalysts, respectively. Temperature was raised from room temperature to 300˚C at a rate of 5˚C/min, then from 300˚C to 700˚C at a rate of 1˚C/min, and subsequently kept at 700˚C for 2 h. The gas flow was then switched from CH4/H2 to Ar for cooling down (overnight).
2.2. Catalytic Activity Experiments
The CO2 methanation activity test were carried out in a tubular quartz reactor at atmospheric pressure using a H2/CO2/Ar = 12/3/5 reactant mixture (total flow 100 ml/min). The gas hourly space velocity (GHSV) was 20,000 h−1. Both Mo/Al2O3 and Ni-Mo/Al2O3 catalysts were either carburized or reduced prior to activity runs.
For the reduced catalyst, prior the activity tests, the Mo/Al2O3 and Ni-Mo/ Al2O3 catalysts were reduced in-situ at 900˚C in 5%H2 in Ar for 1 h, and then cooled down to 250˚C. The catalytic activity experiments were carried out from 250˚C to 500˚C. Steady-state conversions were reached at each temperature (30 min isothermal step). For the carbide catalyst, the methanation experiments were carried out after the in-situ carburization of the sample and its cooling in Ar to 250˚C. The reactants and products were analyzed by a micro gas chromatograph (Varian CPi 4900), equipped with a TCD detector.
The conversions of CO2 and the selectivity of CH4 during the methanation reaction were calculated using the following equations, respectively:
in which is the conversion of CO2 (%), is the selectivity for CH4 (%).
2.3. Physico-Chemical Characterization
CO2 temperature programmed desorption (CO2-TPD) was performed in a BELCAT-M apparatus (BEL Japan). The reduced and carburized catalysts after CO2 methanation were first degassed at 500˚C for 2 h, then cooled to 80˚C. 10% CO2-He was fed for 1 h in order to saturate the catalyst’s surface. After flushing He for 15 min, the materials were heated up from 80˚C to 800˚C under He, at the rate of 10˚C/min, while the evolution of CO2 followed with the aid of a TCD detector. H2 temperature programmed reduction (H2-TPR) carried out in the same device as the CO2-TPD, for both the calcined and the carburized catalysts. The materials were first pretreated at 100˚C for 2 h, then reduced from 100˚C to 900˚C at a rate of 7.5˚C/min in 5% H2 in Ar flow. X-ray photoelectron spectroscopy (XPS) experiment was performed on an AXIS Ultra DLD (KRATOS) spectrometer.
3. Results and Discussion
The XPS spectra corresponding to Ni-Mo2C/Al2O3 carbide catalysts are presented in Figure 1.
Though the presence of carbon-containing compounds, and thus the C 1s peak area, may be dependent on sample/device contamination, the Mo/C ratios are somehow smaller for the two carburized catalyst, in comparison to the calcined ones. This points already to a higher carbon content in the former, as a consequence of effective carburization. The results of the deconvolution of the C 1s, Mo 3d and Ni 2p orbitals can be also found in Table 1. The experimental peaks were decomposed into mixed Gaussian-Lorentzian contributions. The deconvolution of the C 1s orbital was performed as described elsewhere   , considering different species: carbides (238.1) polymeric C-species (284.5 eV), oxidized carbon (286.8 eV), graphite (285.5 eV) and carbonyls/quinones (288.8 eV). The deconvolution points to higher amount of carbide species in the case of the catalysts submitted to the carburization treatment. The Mo 3d orbital
Figure 1. XPS spectrum of Ni 2p, Mo 3d binding Energies for Ni-Mo2C/Al2O3 carburized catalysts.
Table 1. (a) Deconvolution of the C 1s, orbital for the different catalysts before and after carburization (BE in eV in italics); (b) Deconvolution of the Mo 3d and Ni 2p orbitals for the different catalysts before and after carburization (BE in eV in italics).
was deconvoluted into two linked doublets Mo 3d5/2 and Mo 3d3/2, using an intensity ratio of 2/3 and a splitting of 3.2 eV. Five different contributions to the 3d5/2 orbital were considered, respectively corresponding to Mo0, Mo2+, Mo4+, Mo5+ and Mo6+ species. Concretely, the presence of Mo2+ species, also denoted as Moδ+, has been clearly linked to Mo involved in a Mo-C bond, whereas relatively high content of Mo4+ and Mo5+ was assigned to the formation of oxycarbides   . In agreement to the deconvolution of the C 1s orbital, the presence of Mo-carburized species increases after the carburization treatment.
However, higher amount of oxycarbides seems to be as well produced as a consequence of this treatment. This can be due either to the partial oxidation of molybdenum species during passivation upon or to their incomplete carburization   . In any case, carburization seems to be more effective when Ni is present in the formulation of the catalyst   . The Ni 2p orbital was deconvoluted into two doublets Ni 2p3/2 and Ni 2p1/2. The Ni 2p3/2 core-level photoemission spectra presents two states: Ni0 and Ni δ+, respectively appearing at 852.6 and 856.5 eV, together with a shake-up satellite at 861.7 eV. A main peak at 873.8 eV and its shake-up satellite at 880.1 eV conform Ni 2p3/2 contribution. The spin-orbit splitting value between Ni 2p3/2 and Ni 2p1/2 was found to be around 17.5 eV, pointing to the presence of NiAl2O4  . A certain amount of reduced Ni0 species are formed upon the carburization treatment. Let us note here that the X-ray diffraction patterns acquired for this series of catalysts (not shown) did not evidence the presence of Mo2C species, presumably due to its relative low concentration and high dispersion   .
Table 2 shows the results of the integration of the CO2-TPD profiles for this series of catalysts, upon either reduction or carburization, whereas the CO2-TPD profiles acquired for the carburized catalysts Mo2C/Al2O3 and Ni-Mo2C/Al2O3 are detailed in Figure 1. Total basicity extremely increases for the catalysts submitted to the carburization treatment. While both the reduced Mo and the Ni-Mo containing catalysts show very small ability to absorb CO2, the carburized are able to absorb more than 40 times more CO2 than the reduced ones.
Figure 2 showed that both CO2-TPD profiles (acquired for Mo2C/Al2O3 and Ni-Mo2C) covered almost all the temperature window, which pointing to the presence of basic sites with different strength. The most important part for the CO2-TPD was the desorption occurred at low and moderate temperatures, which reflects a major presence of weak and medium-strength basic sites in these carburized catalysts. Note here moreover that the presence of Ni in the bimetallic
Table 2. Total basicity, i.e. integration of the CO2-TPD profiles, for the different catalysts before and after carburization.
Figure 2. CO2-TPD profiles for the Mo2C/Al2O3 and Ni-Mo2C/Al2O3 carburized catalysts.
Figure 3. H2-TPR profiles for the Mo2C/Al2O3 and Ni-Mo2C/Al2O3 carburized catalysts.
catalysts resulted in enhanced CO2 absorption in comparison to both the reduced and carburized Mo-catalysts.
Figure 3 presents H2-TPR profiles for the carburized catalysts. The Mo2C/ Al2O3 catalyst exhibited two main H2 consumption peaks respectively centered at about 437˚C and 910˚C.
The low temperature reduction peak (437˚C) can be assigned either to the reduction of MoO3 to MoO2 or to the reduction of some high valent Mo species (MoOx), The reduction peak occurring observed around 910˚C results from the reduction of MoO2 to metallic Mo, but can also be ascribed to the reduction of Mo2C species  . If this high temperature peak could be ascribed to the further reduction of MoO2 to metallic Mo, the first low temperature peak corresponding to the reduction of MoO3 to MoO2 should have very similar intensity, i.e. similar H2 consumption, than the high temperature one, which, indeed, is not the case. Therefore, the high temperature peak can be directly linked to the presence of molybdenum carbide and oxycarbide species. In the case of the Ni-Mo2C/Al2O3 catalyst, the peak corresponding to the reduction of high valence Mo-species at low temperature appears shifted to lower temperatures, i.e. around 410˚C, and becomes much weaker. The high temperature peak, corresponding to the reduction of molybdenum carbide and oxycarbide species, also shifts to lower temperature, i.e. around 820˚C. This latter result points out that the presence of Ni affects the carburization process and the type of carburized species formed. In the presence of Ni, the carburization treatment leads most probably to favored carburization of molybdenum and to a lower extent of formation of oxycarbides, which confirming the results obtained through XPS analysis. The H2-TPR profiles acquired for the calcined Mo/Al2O3 and Ni-Mo/Al2O3 catalysts can be found elsewhere  , but they only evidenced the H2-consumption peaks typical of Mo and Ni oxide and mixed oxides species.
The results of the methanation experiments are presented in Table 3. It is worth to note that the Mo/Al2O3 catalyst (reduced) was found to be completely
Table 3. Catalytic activity and selectivity of the different catalysts: CO2 conversion and CH4 selectivity at temperatures from 250˚C to 500˚C.
inactive towards CO2 methanation, and thus we decided to exclude the catalytic results of these experiments from Table 3. Both the reduced Ni-Mo/Al2O3 and the carburized Mo2C/Al2O3 showed very low catalytic activity and very poor CH4 selectivity in CO2 methanation reaction.
As previously observed for this thermodynamically feasible but strongly kinetically hindered reaction, the CO2 conversion generally increases with increasing reaction temperatures, i.e. in the case of the Mo2C/Al2O3 catalyst it increases from 4.1% at 350˚C to 26.5% at 500˚C. Similar CO2 conversions were obtained over the reduced Ni-Mo/Al2O3 catalyst. The activity towards CO2 methanation substantially was improved in the presence of the carburized Ni-Mo2C/Al2O3 catalyst. CO2 conversions were increased from 13.8% at 350˚C to 27.3% at 500˚C, whereas CH4 selectivity was almost 100% at low temperatures, which slightly decreasing with increasing temperature, as predicted by equilibrium thermodynamics. Park et al. reported that CO2 conversion would reach to 40% on Pd/SiO2 catalyst at 450˚C, but on which the selectivity to CH4 was only 10%  . According to the above results of the physico-chemical characterization, first of all, carburization results in a 40-fold increase the basicity, i.e. the CO2 absorption ability of these catalysts. As a consequence, and even if the amount of oxycarbide species formed was found to be important, the carburized Mo2C/Al2O3 catalyst showed already a better activity vis-à-vis the reduced Mo/Al2O3 one. Additionally, the presence of Ni enhanced the carburization of the molybdenum species in the bimetallic Ni-Mo catalysts. The amount of oxycarbides was reduced at the same when Ni and Mo was coexist. All this facts resulted therefore in the further promotion of the activity and selectivity observed on the Ni-Mo2C/Al2O3 catalyst. The formation of a mixed NiAl2O4 phase was also observed through the analysis of the XPS results. The presence of Ni in strong interaction with the alumina support may have also affected the carburization process in the case of the Ni-Mo2C/Al2O3 catalyst.
γ-alumina-supported Mo and Ni-Mo catalysts were prepared and submitted either to reduction or to a carburization treatment, prior to evaluating their catalytic activity in CO2 methanation. The presence of Ni facilitated the formation of the Mo2C species, considerably reducing the formation of oxycarbides. The CO2 absorption substantially increased upon carburization, leading to improved catalytic activity in CO2 methanation. Moreover, the presence of Ni and thus as a consequence of favored carburization, resulted in further enhanced catalytic activity and selectivity. Nevertheless, the carburized catalysts still contained an important amount of oxycarbide species, pointing to incomplete carburization. Though the results are quite promising and prove that Ni-Mo carbide catalysts can be successfully used for CO2 methanation, the carburization treatment needs to be consequently optimized.
Lu Yao would like to acknowledge the Chinese Scholarship Council (CSC) for the financial support for her last year PhD at UPMC Sorbonne Universités.