Supported Nanostructured MoxC Materials for the Catalytic Reduction of CO2 through the Reverse Water Gas Shift Reaction

MoxC-based catalysts supported on γ-Al2O3, SiO2 and TiO2 were prepared, characterized and studied in the reverse water gas shift (RWGS) at 548–673 K and atmospheric pressure, using CO2:H2 = 1:1 and CO2:H2 = 1:3 mol/mol reactant mixtures. The support used determined the crystalline MoxC phases obtained and the behavior of the supported nanostructured MoxC catalysts in the RWGS. All catalysts were active in the RWGS reaction under the experimental conditions used; CO productivity per mol of Mo was always higher than that of unsupported Mo2C prepared using a similar method in the absence of support. The CO selectivity at 673 K was above 94% for all the supported catalysts, and near 99% for the SiO2-supported. The MoxC/SiO2 catalyst, which contains a mixture of hexagonal Mo2C and cubic MoC phases, exhibited the best performance for CO production.


Introduction
In addition to capture and storage of CO 2 , nowadays there is a clear interest in its use as an out-stream chemical feedstock in order to actively contribute to the reduction of CO 2 emissions; CO 2 can be considered a cheap carbon C1 source for upgrading rather than a waste with consequences in global warming [1][2][3][4]. However, the direct transformation of CO 2 to useful products is difficult. The high chemical stability of CO 2 difficult its catalytic transformation, the developing of new materials capable of efficiently bind and activate this molecule is nowadays an active research area. An interesting CO 2 utilization approach is its reduction to CO, employing H 2 as a reducing agent via the reverse water gas shift (RWGS) reaction [5][6][7][8]: The reduction of CO 2 to CO with renewable H 2 can be regarded as a simple and easy path for CO 2 recycling, which would allow its reuse at a large scale. After the RWGS step and H 2 O separation, a CO 2 /CO/H 2 out-stream mixture can be produced. This outstream can be used as syngas input for other well-established chemical processes, such as Fischer-Tropsch (FT) or methanol synthesis [9][10][11][12][13][14][15].
The RWGS reaction can be carried out using noble metal-based catalysts [5,10,16]. Due to the similar properties of transition metal carbides (TMCs) and Pt-based catalysts, the formers have been proposed as catalysts for different processes in which Pt-based catalysts are active [17,18]. One of these processes is the CO 2 reduction to CO, which has been analyzed over different TMCs using theoretical and experimental approaches [19][20][21][22][23][24][25].
The preparation of TMCs is usually carried out using carburization methods. These methods apply high temperature and/or pressure conditions in the presence of a reducing atmosphere, usually mixtures of H 2 and carbon-containing gases (CO, CH 4 , C 2 H 4 ) [25][26][27][28]. Due to the increased interest in TMC-based catalysts, in recent years, greener preparation methods have been explored [21,22,[29][30][31]. In an earlier investigation, we studied the preparation of bulk Mo x C catalysts using different molybdenum and carbon precursors and following sol-gel based routes; the bulk Mo x C catalysts generated, contained different crystalline phases, which influenced their catalytic behavior in the RWGS reaction [31].
The deposition onto a support of the appropriate TMC active phase can be an interesting approach to improve the catalytic behavior of bulk TMCs materials, which usually show low surface area values. Supported Mo x C phases have been used as catalysts in different processes such as CH 4 dry reforming [32], hydrazine decomposition [33], thiophene hydrodesulfurization [34], propene and tetralin hydrogenation [35] and Fischer-Tropsch synthesis [36]. However, supported Mo x C catalysts have not been much studied in the RWGS reaction [37][38][39]. Porosoff et al. have reported the promoter effect of K in Al 2 O 3 -supported Mo 2 C-based catalysts containing MoO 2 and/or metallic Mo, which were prepared by carburization with CH 4 /H 2 at 873 K [38]. Sub-nanosized molybdenum carbide clusters highly dispersed onto N-doped carbon/Al 2 O 3 , prepared by carbonization of MoO 3 with glucose, were more performant in the RWGS than bulk β-Mo 2 C [39]. Recently, the preparation of SiO 2 -and SBA-15-supported Mo 2 C-based catalysts (20% wt Mo), using different routes of Mo incorporation to the support and a final carburization process with CH 4 /H 2 , has been studied [40]. The preparation method and the support influenced the composition of Mo x C y crystalline phases developed and therefore the catalytic performance of the material in the RWGS [40]. The preparation of Mo x C-based catalysts supported onto γ-Al 2 O 3 , SiO 2 and MFI-type zeolites by incipient wetness impregnation of ammonium molybdate and carburization with CH 4 /H 2, have led to catalysts with different Mo-containing species such as Mo 2 C, MoO 3 and Mo 0 ; the phases developed and the catalytic performance in the RWGS of the materials depended also on the support characteristics [41].
Here, Mo x C phases were generated onto γ-Al 2 O 3 , SiO 2 and TiO 2 by a thermal treatment of the solid obtained from the interaction between a MoCl 5 /urea solution and the corresponding oxide. The crystalline Mo x C phases obtained depended on the support used in the preparation and determined the catalytic behavior of materials in the RWGS.

Preparation of Catalysts
Commercial γ-Al 2 O 3 (Alfa Aesar, Haverhill, MA, US, 226 m 2 g −1 ), SiO 2 (Degussa, Frankfurt, Germany, 200 m 2 g −1 ) and TiO 2 (Tecnan, Navarra, Spain, 117 m 2 g −1 , anatase/rutile, 78/22% wt) were employed as supports. Urea (Alfa Aesar, Haverhill, MA, US, 99%), which was used as carbon source, was added to a solution of MoCl 5 (Alfa Aesar, Haverhill, MA, US, 99.6%) in ethanol with a urea/MoCl 5 = 7 molar ratio [21,29,31]. The viscous solution was contacted with the respective powdered support. The resulting solid was dried at 333 K, and then treated under Ar flow up to 1073 K for 3 h. The samples were cooled down to room temperature under Ar and then exposed to air without passivation. Mo x C/Al 2 O 3 , Mo x C/TiO 2 and Mo x C/SiO 2 catalysts with about 26% wt of Mo were prepared by using the proper amount of molybdenum and carbon precursors. A reference catalyst (unsupported), containing only bulk hexagonal Mo 2 C was prepared following a similar method but in the absence of support [21]. For characterization purposes, the commercial supports were also separately treated up to 1073 K (3 h) under Ar.

Characterization of Catalysts
The Mo content of samples was determined by inductively coupled plasma mass spectrometry using a Perkin Elmer Optima 3200RL apparatus (Santa Clara, CA, US). The N 2 adsorption-desorption isotherms were recorded at 77 K using a Micromeritics Tristar II 3020 equipment. Prior to the measurements, the samples were outgassed at 523 K for Nanomaterials 2022, 12, 3165 3 of 15 5 h. The specific surface area (S BET ) was calculated by multi-point BET analysis of N 2 adsorption isotherms. The X-ray powder diffraction (XRD) analysis was performed using a PANalytical X'Pert PRO MPD Alpha1 powder diffractometer (Malvern, UK) equipped with a CuKα 1 radiation. The XRD profiles were collected in the 2θ range of 4 • -100 • with a step size of 0.017 • and counting 50 s at each step. Transmission electron microscopy (TEM-HRTEM) images and energy dispersive X-ray analysis (EDX) were collected employing a JEOL J2010F microscope (Tokyo, Japan) operated at an accelerating voltage to 200 kV. The Raman spectra of the samples were collected using a Jobin-Yvon LabRam HR 800, fitted to an optical Olympus BXFM microscope (Kyoto, Japan) with a 532 nm laser and a CCD detector. X-ray photoelectron spectroscopy (XPS) analysis was performed using a Perkin Elmer PHI-5500 Multitechnique System (Physical Electronics, Chanhassen, MN, US) with an Al X-ray source (hυ = 1486.6 eV and 350 W). Samples were kept in an ultra-high vacuum chamber during data acquisition (5·10 −9 -2·10 −8 Torr). Before XPS measurements, the C 1s BE of adventitious carbon was determined in the same equipment and conditions using Au as reference. The BE values were referred to the mentioned C 1s BE at 284.8 eV.

RWGS Catalytic Tests
The RWGS reaction tests were carried out in a Microactivity-Reference unit (PID Eng&Tech) using a tubular fixed-bed reactor under atmospheric pressure. Approximately, 150 mg of catalyst were diluted with inactive SiC up to 1 mL of catalytic bed. The RWGS was studied at 0.1 MPa, between 548 K and 673 K, by following the temperature sequence: . The first part of the catalytic test: 598 K (3 h)→573 K (3 h)→548 K (10 h) was carried out in order to condition the catalyst under RWGS. The gas hourly space velocity (GHSV) was 3000 h −1 . The effluent was analysed on-line with a gas chromatograph Varian 450-GC equipped with a methanizer and TCD and FID detectors. CO 2 conversion and product distribution at each temperature were determined by the average of at least three measures.

Results and Discussion
As stated above, Al 2 O 3 -, SiO 2 -and TiO 2 -supported Mo x C catalysts with about 26% wt Mo were prepared, characterized and tested in the RWGS reaction. Table 1 shows the Mo content and the S BET of fresh catalysts. For comparison, S BET values of the supports treated at 1073 K under Ar, which are the conditions used in the preparation of catalysts, are also included. In all cases, the S BET of the supports after the thermal treatment at 1073 K was lower than that of the corresponding commercial pristine material; the diminution was about 10% for Al 2 O 3 and SiO 2, meanwhile for TiO 2 the S BET decreased from 117 m 2 g −1 to 13 m 2 g −1 . For TiO 2 , a phase change occurred during the thermal treatment; the rutile weight percentage increased from 22% (pristine material) until 95% after the treatment at 1073 K, as determined from XRD analysis [42]. On the other hand, except for the Mo x C/TiO 2 , the S BET of supported catalysts was lower than that of the corresponding support treated at 1073 K; the formation of Mo x C could prevent in some extension the surface area decrease of the TiO 2 support, which could be related with a different extent of the rutile formation from anatase. The supported catalysts were analyzed by XRD, and the corresponding XRD patterns are shown in Figures 1-3; XRD patterns of the respective supports treated at 1073 K under Ar are also displayed for comparison. From the XRD pattern of Mo x C/Al 2 O 3 (Figure 1), characteristic diffraction peaks of γ-Al 2 O 3 are observed, and the main presence of hexagonal Mo 2 C (JCPDS 00-035-0787) can be deduced; a crystallite size of 28 nm was calculated. The XRD analysis of Mo x C/SiO 2 ( Figure 2) indicates the presence of hexagonal Mo 2 C; however, the observation of diffraction peaks with maxima at 2θ = 36.9 • and 2θ = 42.1 • are attributed to the presence of cubic MoC (JCPDS 03-065-0280). From the intensity of diffraction peaks of both phases and that in reference files, a semiquantitative analysis was performed [43]; the presence of 65% cubic MoC and 35% hexagonal Mo 2 C is determined in the Mo x C/SiO 2 catalyst. Figure 3 shows the corresponding XRD profile of TiO 2 -supported catalyst. Characteristic diffraction peaks of both anatase and rutile TiO 2 phases are clearly observed. The rutile weight percentage with respect to TiO 2 phases calculated from XRD pattern is 51% [42]. As commented above, the formation of Mo x C could prevent the anatase transformation, having the Mo x C/TiO 2 catalyst a higher amount of anatase and a higher surface area than the support treated at 1073 K (Table 1). From the XRD pattern of Mo x C/TiO 2 , the main presence of cubic MoC with poor crystallinity can be proposed, even if the presence of hexagonal Mo 2 C could not be ruled out ( Figure 3).
The catalysts were also characterized by Raman spectroscopy, TEM-HRTEM, STEM-EDX and XPS. Raman spectroscopy was used in order to determine the presence of molybdenum oxide species and/or carbonaceous residues ( Figure S1). The very low intensity bands in the zone 815-990 cm −1 points to the presence of residual MoO 3 [44][45][46], which could be formed by surface oxidation when the samples were exposed to air. For Mo x C/TiO 2 , Raman bands at 260, 429 and 610 cm −1 , assigned to rutile, and at 150 cm −1 assigned to anatase, are clearly visible [47][48][49]. In all cases, the intensity of the bands in the 1200-1700 cm −1 region characteristic of carbonaceous species (D and G bands), is negligible ( Figure S1).  The supported catalysts were analyzed by XRD, and the corresponding XRD patterns are shown in . From the intensity of diffraction peaks of both phases and that in reference files, a semiquantitative analysis was performed [43]; the presence of 65% cubic MoC and 35% hexagonal Mo2C is determined in the MoxC/SiO2 catalyst. Figure 3 shows the corresponding XRD profile of TiO2-supported catalyst. Characteristic diffraction peaks of both anatase and rutile TiO2 phases are clearly observed. The rutile weight percentage with respect to TiO2 phases calculated from XRD pattern is 51% [42]. As commented above, the formation of MoxC could prevent the anatase transformation, having the MoxC/TiO2 catalyst a higher amount of anatase and a higher surface area than the support treated at 1073 K (Table 1). From the XRD pattern of MoxC/TiO2, the main presence of cubic MoC with poor crystallinity can be proposed, even if the presence of hexagonal Mo2C could not be ruled out ( Figure 3).  The catalysts were also characterized by Raman spectroscopy, TEM-HRTEM, STEM-EDX and XPS. Raman spectroscopy was used in order to determine the presence of molybdenum oxide species and/or carbonaceous residues ( Figure S1). The very low intensity bands in the zone 815-990 cm −1 points to the presence of residual MoO3 [44][45][46], which could be formed by surface oxidation when the samples were exposed to air. For MoxC/TiO2, Raman bands at 260, 429 and 610 cm −1 , assigned to rutile, and at 150 cm −1 assigned to anatase, are clearly visible [47][48][49]. In all cases, the intensity of the bands in the 1200-1700 cm −1 region characteristic of carbonaceous species (D and G bands), is negligible ( Figure S1).   The catalysts were also characterized by Raman spectroscopy, TEM-HRTEM, STEM-EDX and XPS. Raman spectroscopy was used in order to determine the presence of molybdenum oxide species and/or carbonaceous residues ( Figure S1). The very low intensity bands in the zone 815-990 cm −1 points to the presence of residual MoO3 [44][45][46], which could be formed by surface oxidation when the samples were exposed to air. For MoxC/TiO2, Raman bands at 260, 429 and 610 cm −1 , assigned to rutile, and at 150 cm −1 assigned to anatase, are clearly visible [47][48][49]. In all cases, the intensity of the bands in the 1200-1700 cm −1 region characteristic of carbonaceous species (D and G bands), is negligible ( Figure S1).   (Figure 4), the presence of hexagonal Mo 2 C with a mean particle size of 21 nm was determined in agreement with XRD results. TEM-HRTEM analysis of Mo x C/SiO 2 ( Figure 5) allowed to confirm the presence of hexagonal Mo 2 C and cubic MoC particles with bimodal distribution and mean particle sizes of 18 nm and 5 nm, respectively ( Figure 5A-C). For Mo x C/TiO 2 (Figure 6), only the presence of the cubic MoC phase with a mean particle size of 4 nm could be determined. The supported Mo x C materials studied in this work follow the recently predicted general trend of size-dependent phase diagrams for bulk Mo and W carbides: fcc phases are generally found at small particle size and hcp phases are prevalent at large particle size [50].
In all cases, STEM-EDX results (see Figures 4C, 5D, and 6C) indicate a homogeneous distribution of Mo on the corresponding support. Figures 4D, 5E and 6D, show the corresponding EDX spectra; N-and Cl-containing species were not detected.
hexagonal Mo2C and cubic MoC particles with bimodal distribution and mean particle sizes of 18 nm and 5 nm, respectively ( Figure 5A-C). For MoxC/TiO2 (Figure 6), only the presence of the cubic MoC phase with a mean particle size of 4 nm could be determined. The supported MoxC materials studied in this work follow the recently predicted general trend of size-dependent phase diagrams for bulk Mo and W carbides: fcc phases are generally found at small particle size and hcp phases are prevalent at large particle size [50].      As stated above, the catalysts were also analyzed by XPS. Al 2p, Si 2p and Ti 2p 3/2 BE at 74.8, 104,0 and 459,3 eV, characteristic of Al 2 O 3 , SiO 2 , and TiO 2 , were found for Mo x C/Al 2 O 3 , Mo x C/SiO 2 and Mo x C/TiO 2 , respectively ( Figure S2). Figure 7 shows the C 1s and Mo 3d XP spectra. The C 1s core level spectra ( Figure 7A) show a maximum at 284.8 eV associated to the adventitious carbon, the component at 283.7-283.8 eV is associated to surface molybdenum carbide species [21,31,[51][52][53][54]. Components extended above 284.8 eV are related to different oxygen containing species [52][53][54][55][56]. The Mo 3d spectra are complex ( Figure 7B); however, they can be deconvoluted into four doublets (Mo 3d 5/2 and Mo 3d 3/2 ). According to literature, the Mo 3d 5/2 /Mo 3d 3/2 intensity ratio was fixed to be 1.5, and the Mo 3d 5/2 -Mo 3d 3/2 BE splitting was set at 3.1 eV [57][58][59]. The 3d 5/2 peaks at the lowest BE region, 228.5-228.7 eV, are attributed to Mo 2+ and Mo 3+ in Mo 2 C and/or oxycarbide species [19,21,31,51]. The Mo 3d 5/2 components at 229.4-229.5, 231.3-232.6 and 233.2 eV, can be assigned to Mo 4+ , Mo 5+ and Mo 6+ surface species, respectively [19,[58][59][60][61], which could be related to the presence of MoC, oxycarbide and/or oxide species. Table 2 shows the contribution of Mo 2+ /Mo 3+ and Mo 4+ species to the total surface Mo n+ species; the Mo x C/SiO 2 catalyst having both Mo 2 C and MoC shows the highest values.  All catalysts were tested in the RWGS using CO2:H2 = 1/3 and CO2/H2 = 1/1 ratios. Catalytic data of unsupported Mo2C, prepared using a similar method to that used in this work but in the absence of support, are also included for comparison [21]. As stated in the experimental section, the first part of the catalytic test: 598 K (3 h)→573 K (3 h)→548 K (10 h) was carried out in order to condition the catalyst under RWGS. Next, when the temperature was increased to 598 K, the CO2 conversion was in all cases higher than that obtained at 598 K in the conditioning step ( Figures 8A and 10A). This behavior could be related with the removal of initially adsorbed surface species. After this first step and regardless the catalyst and the conditions, CO2 conversion increases with the rising of reaction temperature from 598 K to 673 K (Figures 8A and 10A).  All catalysts were tested in the RWGS using CO 2 :H 2 = 1/3 and CO 2 /H 2 = 1/1 ratios. Catalytic data of unsupported Mo 2 C, prepared using a similar method to that used in this work but in the absence of support, are also included for comparison [21]. As stated in the experimental section, the first part of the catalytic test: 598 K (3 h)→573 K (3 h)→548 K (10 h) was carried out in order to condition the catalyst under RWGS. Next, when the temperature was increased to 598 K, the CO 2 conversion was in all cases higher than that obtained at 598 K in the conditioning step ( Figure 8A and Figure 10A). This behavior could be related with the removal of initially adsorbed surface species. After this first step and regardless the catalyst and the conditions, CO 2 conversion increases with the rising of reaction temperature from 598 K to 673 K ( Figure 8A and Figure 10A).  Figure 8A); the corresponding equilibrium CO2 conversion for RWGS at the experimental conditions used is about 37% (at 673 K). MoxC/Al2O3 showed a catalytic activity close to that of the unsupported Mo2C catalyst. Meanwhile, MoxC/TiO2 showed lower values of CO2 conversion than those of unsupported Mo2C [21]. These results contrast with those usually reported for supported metallic catalysts [62,63]. The activity of SiO2-and Al2O3-supported metals in the RWGS is usually lower than that found when reducible supports such as TiO2 or CeO2 are used, which can generate oxygen vacancies that strengths the CO2 adsorption and then the activity in the RWGS [63]. In this work, besides the difference in the  Figures 8 and 9 show the RWGS behavior of catalysts when CO 2 :H 2 = 1/3 is used. Mo x C/SiO 2 presented the highest value of CO 2 conversion (27.5%) at 673 K ( Figure 8A); the corresponding equilibrium CO 2 conversion for RWGS at the experimental conditions used is about 37% (at 673 K). Mo x C/Al 2 O 3 showed a catalytic activity close to that of the unsupported Mo 2 C catalyst. Meanwhile, Mo x C/TiO 2 showed lower values of CO 2 conversion than those of unsupported Mo 2 C [21]. These results contrast with those usually reported for supported metallic catalysts [62,63]. The activity of SiO 2 -and Al 2 O 3 -supported metals in the RWGS is usually lower than that found when reducible supports such as TiO 2 or CeO 2 are used, which can generate oxygen vacancies that strengths the CO 2 adsorption and then the activity in the RWGS [63]. In this work, besides the difference in the surface-area of catalysts, the composition and characteristics of generated Mo x C nanoparticles change as a function of the support. surface-area of catalysts, the composition and characteristics of generated MoxC nanoparticles change as a function of the support. A key process in the RWGS is the cleavage of C-O bond with CO + O formation. In this context molybdenum oxycarbide has been proposed as an intermediate in the RWGS over Mo2C that likely enhances the RWGS rate [25]. We have demonstrated that over a polycrystalline α-Mo2C catalyst, prepared with the method used in the present work, the enhanced CO2 dissociation toward CO + O results from specific surface facets [21]. Next, the easy release of CO and the continuous O removal by H2 to form H2O, results in high RWGS activity. The existence of both, hcp Mo2C and fcc MoC phases in the SiO2-supported catalyst, could result in interphases regions with appropriate characteristics to enhance RWGS on MoxC/SiO2 catalyst. In this context, for different MoxC bulk catalysts, the lowest activation energy in the RWGS was found for a catalyst containing several Mo2C and MoC phases [31].
All the supported catalysts showed high CO selectivity values. When CO2:H2 = 1/3 was used, CO selectivity were always higher than 92% ( Figure 8B). The highest CO selectivity was observed for the MoxC/SiO2 catalyst, achieving at 673 K, 98.5%. Only MoxC/Al2O3 showed CO selectivity values slightly lower than that of unsupported Mo2C ( Figure 8B). CH4 was the main byproduct and only very small amounts of ethylene were formed.
For a proper comparison of the catalysts, the values of CO production were calculated per mol of Mo in the samples; results are shown in Figure 9. All the supported catalysts showed a higher CO production per mol of Mo compared to the unsupported Mo2C catalyst [21]. At the end of the catalytic test, MoxC/SiO2 and MoxC/Al2O3 showed a higher CO production at 648 K than before reaction at 673 K (Figure 9). This could be related with the removal of remaining oxygen surface species during the reaction at 673 K. The highest CO production in the whole range of reaction temperature tested was obtained for MoxC/SiO2; it reached about 17.0 mol CO/mol Mo·h at 673 K.
Catalysts were also tested in the RWGS using a stoichiometric ratio of the reactant mixture, CO2/H2/ = 1/1. Figure 10 shows the variation of CO2 conversion and CO selectivity A key process in the RWGS is the cleavage of C-O bond with CO + O formation. In this context molybdenum oxycarbide has been proposed as an intermediate in the RWGS over Mo 2 C that likely enhances the RWGS rate [25]. We have demonstrated that over a polycrystalline α-Mo 2 C catalyst, prepared with the method used in the present work, the enhanced CO 2 dissociation toward CO + O results from specific surface facets [21]. Next, the easy release of CO and the continuous O removal by H 2 to form H 2 O, results in high RWGS activity. The existence of both, hcp Mo 2 C and fcc MoC phases in the SiO 2 -supported catalyst, could result in interphases regions with appropriate characteristics to enhance RWGS on Mo x C/SiO 2 catalyst. In this context, for different Mo x C bulk catalysts, the lowest activation energy in the RWGS was found for a catalyst containing several Mo 2 C and MoC phases [31].
All the supported catalysts showed high CO selectivity values. When CO 2 :H 2 = 1/3 was used, CO selectivity were always higher than 92% ( Figure 8B). The highest CO selectivity was observed for the Mo x C/SiO 2 catalyst, achieving at 673 K, 98.5%. Only Mo x C/Al 2 O 3 showed CO selectivity values slightly lower than that of unsupported Mo 2 C ( Figure 8B). CH 4 was the main byproduct and only very small amounts of ethylene were formed.
For a proper comparison of the catalysts, the values of CO production were calculated per mol of Mo in the samples; results are shown in Figure 9. All the supported catalysts showed a higher CO production per mol of Mo compared to the unsupported Mo 2 C catalyst [21]. At the end of the catalytic test, Mo x C/SiO 2 and Mo x C/Al 2 O 3 showed a higher CO production at 648 K than before reaction at 673 K ( Figure 9). This could be related with the removal of remaining oxygen surface species during the reaction at 673 K. The highest CO production in the whole range of reaction temperature tested was obtained for Mo x C/SiO 2 ; it reached about 17.0 mol CO/mol Mo·h at 673 K.
Catalysts were also tested in the RWGS using a stoichiometric ratio of the reactant mixture, CO 2 /H 2 / = 1/1. Figure 10 shows the variation of CO 2 conversion and CO selectivity values. As expected, the CO 2 conversion ( Figure 10A) was lower and the CO selectivity ( Figure 10B) higher when a mixture CO 2 /H 2 = 1/1 was used than when the reactant mixture was CO 2 /H 2 = 1/3. Using the CO 2 /H 2 = 1/1 reactant mixture, the highest CO 2 conversion ( Figure 10A) and the highest CO production per mol of Mo (Figure 11), in the whole range of reaction temperature tested, were also found over the Mo x C/SiO 2 catalyst. In this case, at the end of the catalytic test, only for Mo x C/SiO 2 a slightly higher CO production at 648 K than before reaction at 673 K was observed ( Figure 11). values. As expected, the CO2 conversion ( Figure 10A) was lower and the CO selectivity ( Figure 10B) higher when a mixture CO2/H2 = 1/1 was used than when the reactant mixture was CO2/H2 = 1/3. Using the CO2/H2 = 1/1 reactant mixture, the highest CO2 conversion ( Figure 10A) and the highest CO production per mol of Mo (Figure 11), in the whole range of reaction temperature tested, were also found over the MoxC/SiO2 catalyst. In this case, at the end of the catalytic test, only for MoxC/SiO2 a slightly higher CO production at 648 K than before reaction at 673 K was observed ( Figure 11).  It is noteworthy, that after the overall RWGS study carried out, all supported catalysts, showed quite constant values of CO2 conversion and CO selectivity during the last step at 648 K (5 h), under both CO2/H2/ = 1/3 and CO2/H2/ = 1/1 conditions. The apparent activation energies (Ea) for CO production over supported catalysts were calculated according to the Arrhenius plots in the temperature range of 598-648 K; values between 65-78 kJ/mol were obtained ( Table 2). These values are in the range of that recently reported for an alumina supported Mo2C cluster-based catalyst (76.4 kJ/mol) [39]. MoxC/SiO2 showed the lowest Ea for CO production. As stated above, the best performance of MoxC/SiO2 could be related with the coexistence in this catalyst of different MoxC phases, hexagonal Mo2C and cubic MoC, as has been recently suggested for unsupported MoxC catalysts [31]. Moreover, MoxC/SiO2 showed the highest contribution of Mo 2+ /Mo 3+ and Mo 4+ species to the total surface Mo n+ species. For MoxC-based catalysts, an easy reduction under reaction conditions of molybdenum species has been related with their performance in RWGS [41].
Post-reaction catalysts were characterized by BET and XRD. Only a slight decrease in the BET surface area was found after the RWGS reaction ( Table 1). The XRD patterns of fresh (Figures 1-3) and post-reaction catalysts after the test with CO2/H2 = 1/3 ( Figure S3) were similar. Meanwhile, the presence of MoO2 was detected by XRD in post-reaction MoxC/SiO2 and MoxC/TiO2 when the reactant mixture was CO2/H2 = 1/1 ( Figure S4); the oxidation could be prevented under a richer hydrogen atmosphere (CO2/H2 = 1/3) due to an easier removal of the O surface species formed from the CO2 activation over these materials under CO2/H2 = 1/3 conditions [21,31].

Conclusions
Using urea and MoCl5 as carbon and molybdenum sources, different MoxC phases were successfully supported over Al2O3, SiO2 and TiO2. The support determined the developed MoxC phases on the materials and their catalytic behavior in the RWGS. Hexagonal Mo2C nanoparticles on MoxC/Al2O3 and cubic MoC nanoparticles on MoxC/TiO2 were found. Over MoxC/SiO2 both hexagonal Mo2C and cubic MoC nanoparticles were present. In all cases, supported hexagonal Mo2C nanoparticles were larger than cubic MoC ones. It is noteworthy, that after the overall RWGS study carried out, all supported catalysts, showed quite constant values of CO 2 conversion and CO selectivity during the last step at 648 K (5 h), under both CO 2 /H 2 / = 1/3 and CO 2 /H 2 / = 1/1 conditions. The apparent activation energies (E a ) for CO production over supported catalysts were calculated according to the Arrhenius plots in the temperature range of 598-648 K; values between 65-78 kJ/mol were obtained (Table 2). These values are in the range of that recently reported for an alumina supported Mo 2 C cluster-based catalyst (76.4 kJ/mol) [39]. Mo x C/SiO 2 showed the lowest E a for CO production. As stated above, the best performance of Mo x C/SiO 2 could be related with the coexistence in this catalyst of different Mo x C phases, hexagonal Mo 2 C and cubic MoC, as has been recently suggested for unsupported Mo x C catalysts [31]. Moreover, Mo x C/SiO 2 showed the highest contribution of Mo 2+ /Mo 3+ and Mo 4+ species to the total surface Mo n+ species. For Mo x C-based catalysts, an easy reduction under reaction conditions of molybdenum species has been related with their performance in RWGS [41].
Post-reaction catalysts were characterized by BET and XRD. Only a slight decrease in the BET surface area was found after the RWGS reaction ( Table 1). The XRD patterns of fresh (Figures 1-3) and post-reaction catalysts after the test with CO 2 /H 2 = 1/3 ( Figure S3) were similar. Meanwhile, the presence of MoO 2 was detected by XRD in post-reaction Mo x C/SiO 2 and Mo x C/TiO 2 when the reactant mixture was CO 2 /H 2 = 1/1 ( Figure S4); the oxidation could be prevented under a richer hydrogen atmosphere (CO 2 /H 2 = 1/3) due to an easier removal of the O surface species formed from the CO 2 activation over these materials under CO 2 /H 2 = 1/3 conditions [21,31].

Conclusions
Using urea and MoCl 5 as carbon and molybdenum sources, different Mo x C phases were successfully supported over Al 2 O 3 , SiO 2 and TiO 2 . The support determined the developed Mo x C phases on the materials and their catalytic behavior in the RWGS. Hexagonal Mo 2 C nanoparticles on Mo x C/Al 2 O 3 and cubic MoC nanoparticles on Mo x C/TiO 2 were found. Over Mo x C/SiO 2 both hexagonal Mo 2 C and cubic MoC nanoparticles were present. In all cases, supported hexagonal Mo 2 C nanoparticles were larger than cubic MoC ones.
All catalysts showed a stable catalytic behavior and exhibited higher CO production per mol of Mo than the unsupported hexagonal Mo 2 C similarly prepared, under the reaction conditions used (CO 2 /H 2 = 1/3 and CO 2 /H 2 = 1/1; T = 548-673 K). Mo x C/SiO 2 exhibited the highest surface ratio of Mo species with low oxidation states (Mo 2+,3+,4+ ) and the best performance in the RWGS reaction. Over Mo x C/SiO 2 , CO 2 conversion of 27.5% and CO selectivity of 98.5% were achieved at 673 K under CO 2 /H 2 = 1/3; for CO production, an apparent activation energy of 64.9 ± 3.2 kJ mol −1 was determined at 598-648 K under CO 2 /H 2 = 1/1. The catalytic behavior is proposed to be governed by the supported Mo x C phase. The simultaneous presence of hexagonal Mo 2 C and cubic MoC nanoparticles in Mo x C/SiO 2 plays a main role on the catalytic behavior of this catalyst.

Data Availability Statement:
The data present in this study are available on request from the corresponding author.