Supported Vanadium Carbide Catalysts for Reverse Water Gas Shift and Methanol Steam Reforming: Activity, Stability, and Coking Pathways

Abstract

A series of supported vanadium carbide (VC _x ) catalysts were prepared, characterized, and tested for the carbon dioxide and methanol activation via the Reverse Water Gas Shift (RWGS) and Methanol Steam Reforming (MSR) reactions, respectively. Crystallite sizes of VC _x ranging from 9 to 36 nm were obtained depending on the support used (γ-Al₂O₃, SiO₂, CeO₂, ZrO₂ and TiO₂). In both reactions, the supported catalysts exhibited superior performance compared to the bulk VC _x sample. In the RWGS reaction, all catalysts showed high CO selectivity, with VC _x /Al₂O₃ demonstrating the best performance and no significant deactivation after 100 h at 873 K. Under MSR conditions, VC _x /ZrO₂ achieved the highest methanol conversion. However, all catalysts suffered from significant deactivation due to coke formation, with CH₄ as the main product instead of the desired H₂ and CO₂ from full steam reforming. Density Functional Theory (DFT) calculations revealed that methanol decomposition is more facile than CO₂ decomposition on both stoichiometric VC and carbon-deficient V₈C₇ surfaces, particularly in the presence of carbon vacancies, leading to coke formation in the form of partially hydrogenated C _x H _y * species. These findings indicate that VC _x catalysts are more susceptible to coking under MSR than RWGS conditions, in line with experimental observations, and highlight the critical role of the carbide surface structure and vacancy concentration in coke formation.

1. Introduction

Transition metal carbides (TMCs), particularly those derived from early transition metals in Groups 4–6, have attracted significant attention for their distinctive catalytic properties. Since the discovery of the Pt-like behavior of tungsten carbide (WC), TMCs have been extensively investigated in the fields of surface science and catalysis. ,− These materials have emerged as promising, earth-abundant, and cost-effective alternatives to the precious noble metals traditionally used in several catalytic reactions. This is largely due to their similar electronic structures, excellent chemical and physical resistance, and high tolerance to coking and sulfur poisoning, all of which enhance their durability in catalytic processes. ,,

While molybdenum and tungsten carbide-based catalysts have been widely studied, − vanadium carbide (VC _x ) catalysts remain relatively unexplored. Single-crystal studies have shown that stoichiometric vanadium carbide (VC) enhances C–H bond activation in alkanes, while interacting less strongly with CC bonds in alkenes compared to metallic vanadium. This unique reactivity is similar to that of Pt-group metals and is also observed for Mo₂C and WC. Notably, VC _x catalysts have been demonstrated to be catalytically active for several reactions, including ammonia decomposition, oxygen reduction reaction (ORR), hydrogen evolution reaction (HER), and n-butane dehydrogenation. −

In previous work, we highlighted the key role of carbon vacancies in VC _x during the Reverse Water Gas Shift (RWGS) reaction (CO₂ + H₂ → CO + H₂O; Δ_r H° = 41.2 kJ mol^–1 at 298 K), where they enhance both activity and CO selectivity, while suppressing side reactions such as CO₂ and CO methanation. Density Functional Theory (DFT) calculations further revealed that the carbon-deficient V₈C₇ phase adsorbs both CO₂ and H₂ more strongly and facilitates their dissociation, lowering the CO₂ dissociation barrier from 1.53 eV on VC to 0.62 eV on V₈C₇, and the H₂ dissociation barrier from 0.65 eV on VC to 0.16 eV on V₈C₇. , Additionally, studies of Al₂O₃-supported VC _x nanoparticles demonstrated that the presence of carbon vacancies and large VC _x -Al₂O₃ interfacial areas are beneficial for the RWGS reaction. These supported catalysts exhibited only minor deactivation under reaction conditions. ,

In contrast, under methanol steam reforming (MSR) reaction conditions, we revealed that bulk VC _x is highly selective toward CH₄ formation instead of H₂ + CO₂ (CH₃OH + H₂O → CO₂ + 3H₂; Δ_r H° = 49.7 kJ mol^–1 at 298 K) − and undergoes severe deactivation due to carbon deposition. This high selectivity toward CH₄ has also been observed on stoichiometric VC single crystals, suggesting that the complete decomposition of methanol (CH₃OH) proceeds through C–O bond cleavage of intermediate methoxy species, yielding CH₄ and surface O.

Building on this background, we have prepared, characterized, and evaluated a series of VC _x -based catalysts supported on various oxides (γ-Al₂O₃, SiO₂, CeO₂, ZrO₂ and TiO₂) for both the RWGS and MSR reactions as case studies, with the aim of investigating their catalytic behavior, stability and deactivation pathways. We show how the structural features of the supported VC _x nanoparticles correlate with their catalytic behavior under these conditions. In addition, we conduct DFT calculations to elucidate and compare the coke-formation pathways for both RWGS and MSR reactions on VC and V₈C₇ surface models. This combined experimental and theoretical work aims to provide new insights into the structure–reactivity relationships and coking mechanisms of vanadium carbide catalysts in industrially relevant processes.

2. Methodology

2.1. Preparation

γ-Al₂O₃ (Alfa Aesar, 226 m ² g ^–1 ), SiO₂ (Degussa, 200 m ² g ^–1 ), CeO₂ (Tecnan, 90 m ² g ^–1 ), ZrO₂ (Tecnan, 50 m ² g ^–1 ) and TiO₂ (Tecnan, 117 m ² g ^–1 anatase/rutile = 78/22 wt %/wt) were used as support materials. The supports were first immersed in 50 mL of ethanol to form a suspension. Then, an equimolar amount of vanadium oxytriisopropoxide (VO­(isopropoxide) ₃, Alfa Aesar 96%) and 4,5-dicyanoimidazole (C ₅ H ₂ N ₄, Manchester Organics 97%) was added to yield ∼25 wt % of V in the final catalyst, based on procedures previously reported for the preparation of bulk and alumina-supported VC _x catalysts. ,, The suspension was stirred at room temperature under Ar until the ethanol had evaporated. Afterward, the solid was treated in a tubular furnace under a flow of Ar up to 1373 K (2.5 K min^–1) for 5 h and then cooled down. The catalysts were named VC _x /Al₂O₃, VC _x /SiO₂, VC _x /CeO₂, VC _x /ZrO₂ and VC _x /TiO₂ indicating the support used in the preparation. For comparison, a bulk VC _x sample was prepared following the same procedure but in the absence of support.

2.2. Characterization

The crystal structures of the samples were characterized by X-ray diffraction (XRD) analysis with a PANalytical X’Pert PRO MPD Alpha1 powder diffractometer, using a Cu Kα radiation source (λ = 1.5406 Å). The crystallite size was calculated using the Debye–Scherrer equation. Nitrogen adsorption–desorption was performed at 77 K, using a Micromeritics Tristar II 3020 instrument. Before the measurement, the catalyst was degassed at 525 K for 5 h under N₂. The pore size distribution was determined by applying the Barrett–Joyner–Halenda (BJH) method.

Inductively coupled plasma atomic emission spectrometry (ICP-AES) was used for the analysis of the chemical composition of the catalysts. The ICP-AES measurements were carried out using a PerkinElmer Optima 3200RL apparatus. X-ray photoelectron spectroscopy (XPS, PerkinElmer PHI-5500 Multitechnique System, Physical Electronics) was used to analyze the surface of the samples. All spectra were collected using an Al X-ray source (hν = 1486.6 eV and 350 W). Calibration of the instrument was confirmed using Au as a reference and the binding energy (BE) of the C 1s peak at 284.8 eV.

Transmission electron microscopy (TEM) images were collected employing a JEOL J2010 F microscope operated at up to 200 kV. Energy dispersive X-ray (EDX) analysis was carried out in an Oxford instrument INCA x-sight. H₂-temperature-programmed reduction (H₂-TPR) experiments were performed in a Micromeritics Autochem II 2920 equipped with a thermal conductivity detector (TCD). Samples were pretreated at 363 K under He for 1 h and then exposed to an H₂/Ar (12% v/v) flow, and the temperature was then increased up to 1073 at 10 K min^–1.

Raman spectroscopy analysis was performed in a Jobin-Yvon LabRam HR 800 instrument, with an optical Olympus BXFM microscope with a 532 nm laser and a CCD detector. The laser power was restricted to 1.25 mW to avoid undesired laser-heating effects during spectra acquisition. Thermogravimetric analysis (TGA) was performed on a NETZSCH STA449 F3 Jupiter instrument coupled to a Pfeiffer mass spectrometer (MS). Samples were heated from 298 to 1073 K (5 K min^–1) under air flow. MS signals were recorded during the analysis.

2.3. Catalytic Tests

2.3.1. RWGS Reaction

The RWGS catalytic tests were performed in a Microactivity unit (PID Eng&Tech), using 300 mg of catalyst, a catalytic bed of 1 mL, and SiC as the diluting agent. The catalytic bed was loaded into a tubular reactor. The samples were heated under N₂ flow from room temperature up to 525 K. Then, they were exposed to a reactant gas mixture of CO₂/H₂/N₂ = 1/3/1 (mol/mol) at a constant gas hourly space velocity (GHSV) of 3000 h^–1.

All catalysts were studied within a temperature range of 523 to 873 K at 0.1 MPa; the catalysts were kept for 2.5 h at each temperature, which was reached after 10 min of heating; the first measurement was taken after 30 min at a given temperature. A Varian 450-GC, equipped with a TCD and two FIDs, was used for the online analysis of the products. CO₂ conversion and product distribution were determined at each temperature from the average of 4 measurements. The main products obtained were CO and CH₄. The CO₂ conversion (X _CO2 ) and the selectivity to carbon-product i (S _i ) were determined as follows (eqs and )­

$$X_{{\mathrm{C}\mathrm{O}}_2}(\%)=\left(1-\frac{{(C_{{\mathrm{C}\mathrm{O}}_2})}_{\mathrm{o}\mathrm{u}\mathrm{t}\mathrm{l}\mathrm{e}\mathrm{t}}}{{(C_{{\mathrm{C}\mathrm{O}}_2})}_{\mathrm{o}\mathrm{u}\mathrm{t}\mathrm{l}\mathrm{e}\mathrm{t}}+\sum {(C_i)}_{\mathrm{o}\mathrm{u}\mathrm{t}\mathrm{l}\mathrm{e}\mathrm{t}}}\right)\times 100$$ 1

$$S_i(\%)=\frac{{(C_i)}_{\mathrm{outlet}}}{\sum {(C_i)}_{\mathrm{outlet}}}\times 100$$ 2

where C _i and C _CO2 denote the molar concentrations of the i product (CH₄ or CO) and CO₂, respectively.

2.3.2. MSR Reaction

The MSR catalytic tests were performed using a Microactivity-Reference unit (PID Eng&Tech), similar to the setup used for the RWGS tests described above, but equipped with a GILSON liquid pump for injecting the reactant mixture H₂O/CH₃OH = 1/1 (molar ratio) at a constant flow and at atmospheric pressure.

The preheated liquid mixture (473 K) was mixed with N₂, H₂O/CH₃OH/N₂ = 1/1/1.2 (molar ratio) and flowed through the catalytic bed. A liquid–gas separator working at 277 K allowed the condensation of vapors at the outlet of the system. A total of 300 mg of catalyst were diluted in a catalytic bed of 1 mL with SiC. All tests were carried out at 573–723 K, 0.1 MPa and a GHSV of 2500 h^–1. Each reaction temperature was reached after 10 min of heating, and the first measurement was taken after 30 min at each temperature. The CH₃OH conversion and product distribution were determined at each temperature from the average of 4 different measurements. The catalysts were kept for 1.5 h at each temperature. At 723 K, the catalysts were kept for 20 h under reaction to study the stability of these materials. The gaseous products were analyzed online employing a Varian 4900 micro-GC equipped with three channels with micro TCDs.

The CH₃OH conversion (X _CH3OH) was defined as follows (eq )­

$$X_{{\mathrm{CH}}_3\mathrm{OH}}(\%)=\frac{\sum a_i·{{(\eta }_i)}_{\mathrm{outlet}}}{{{(\eta }_{\mathrm{C}{\mathrm{H}}_3\mathrm{OH}})}_{\mathrm{inlet}}}\times 100$$ 3

where a _i is the number of carbon atoms per molecule of the product i (CH₄, CO, CO₂, HCHO, C₂H₄ and C₂H₆), η _i is the number of moles of the product i, and η_CH3OH is the number of moles of methanol. The product distribution is given as molar fraction (Y _i ) of products obtained (eq )­

$$Y_i(\%)=\frac{{(C_i)}_{\mathrm{outlet}}}{\sum {(C_i)}_{\mathrm{outlet}}}\times 100$$ 4

where C _i is the concentration of the product i.

2.4. Computational Methods

The Vienna Ab Initio Simulation Package (VASP, version 5.4.4) was employed to carry out periodic DFT calculations. The exchange-correlation effects were captured using the Perdew–Burke–Ernzerhof (PBE) functional together with the D3 dispersion correction. , The effect of core electrons was approximated using the Projector Augmented Wave (PAW) method. , All calculations used an energy cutoff of 415 eV and electronic and ionic convergence criteria of 10^–5 eV and 10^–2 eV Å^–1, respectively.

The slab models for the stoichiometric (VC) and C-deficient (V₈C₇) phases were obtained from the corresponding optimized bulk structures. The VC slab consist of a 3 × 3 supercell, while the V₈C₇ slab used a 1 × 1 supercell, both including 4 layers with the two bottommost layers fixed to bulk positions (Figure S1). Note that the V₈C₇ phase consists of ordered vacancies and has been reported to be more stable than the stoichiometric VC phase. All slab calculations were carried out using a Γ-centered 5×5×1 k-point mesh. A vacuum layer of at least 18 Å was included to eliminate interactions between the periodic slab images, and dipole corrections were applied in the z-direction.

All transition states (TS) were located using the ML-NEB module included in CatLearn. A vibrational analysis was performed by means of the finite difference method using a displacement of 0.02 Å, in order to confirm the nature of each TS by verifying the presence of a single imaginary frequency. All free energy contributions were computed using the ideal-gas model for gas-phase molecules (including translations, rotations and vibrations as necessary for each species), and the harmonic oscillator model for the adsorbed species, as implemented in the ASE thermochemistry module.

3. Results

3.1. Characterization of the Catalysts

Table shows the vanadium content and S _BET values for the supported VC _x catalysts prepared using the procedures detailed in Section . The S _BET values obtained for all catalysts were in the range of 86–271 m² g^–1. The S _BET of VC _x /Al₂O₃ and VC _x /CeO₂ decreased relative to their respective fresh supports (see Section for S _BET of fresh supports). However, for VC _x /SiO₂, VC _x /ZrO₂ and VC _x /TiO₂, an increase in S _BET was observed. All catalysts were mainly mesoporous materials with average pore widths ranging over 6–17 nm (Table and Figure S2).

1. Several Characteristics of Supported VC _x Catalysts before and after RWGS and MSR Reaction.

			S BET (m2 g–1)			pore size (nm)
catalyst	V content (wt %)	crystallite size of VC x from XRD (nm)	fresh	used RWGS	used MSR	fresh	used RWGS	used MSR
VC x /Al2O3	22.1	9	210	101	24	9	11	9
VC x /SiO2	21.3	9	216	92	37	7	9	10
VC x /CeO2	22.5	36	86	45	9	6	6	6
VC x /ZrO2	21.6	25	119	59	8	6	6	21
VC x /TiO2	23.0	20	158	82	48	17	18	16
VC x		11	271	219	5	7	10	35

The XRD peaks located at 2θ = 37.7, 43.4, 63.1, 75.6 and 79.7° correspond to reflections characteristic of cubic vanadium carbides, which may include V₈C₇ (JCPDS 35-0786) and VC (JCPDS 01-073-0476) (Figure ). Because the diffraction patterns of these two phases are very similar, their distinction based solely on XRD is challenging. Additionally, no crystalline VO _x species were detected (Figure ). In a prior study, we investigated VC _x samples synthesized using a similar procedure to the one employed in this current work, and found that the resulting bulk material consisted predominantly of V₈C₇ rather than stoichiometric VC, as confirmed by high-resolution STEM analysis. Given that the supported VC _x catalysts were synthesized following the same route, it can be inferred that they also contain a significant proportion of V₈C₇, while the fraction of stoichiometric VC likely increases with particle sintering.

1.

XRD patterns of the (a–d) supported and (d) bulk VC _x catalysts used in this work and their corresponding support materials.

During the preparation of supported catalysts, materials were treated at 1373 K, and some supports underwent structural changes. A partial transformation of γ-Al₂O₃ cubic phase (JCPDS 10-0425) into monoclinic (JCPDS 35-0121) and rhombohedral (JCPDS 10-0173) phases was detected, as a result of the high temperature employed during the preparation of VC _x /Al₂O₃ (Figure a). , For VC _x /TiO₂, the presence of (Ti_0.9V_0.1)₂O₃ rhombohedral solid solution (JCPDS 01-071-0275) is deduced (Figure b). In the case of ZrO₂ with both tetragonal (JCPDS 50-1089) and monoclinic (JCPDS 37-1484) phases, a small increase in the amount of monoclinic phase is proposed to have occurred during the preparation of VC _x /ZrO₂ (Figure c). In VC _x /CeO₂, cubic CeO₂ (JCPDS 34-0394) was found (Figure c). The crystallite sizes of VC _x , measured by XRD for all catalysts, are listed in Table . Both VC _x /Al₂O₃ and VC _x /SiO₂ catalysts exhibited slightly smaller sizes (9 nm) compared to the bulk VC _x catalyst (11 nm). Conversely, higher crystallite sizes were obtained for all the other catalysts (Table ). This variation in VC _x crystallite size can be attributed to the surface area of the material support, as supports with higher S _BET values restrict the sintering of the VC _x particles during preparation, leading to the formation of smaller VC _x nanoparticles.

The samples exhibited Raman bands at 1350 and 1598 cm^–1, which were attributed to residual carbon formed during the preparation process (Figure S3). Additionally, the Raman bands at 97, 143, 192, 285, 694, and 995 cm^–1 could be associated with the presence of oxy-vanadium species. , Furthermore, for VC _x /CeO₂, the band at 461 cm^–1 is attributed to CeO₂, and those at 122, 264, 375, 785, and 854 cm^–1 to CeVO₄. − The poorly defined bands at 269, 413, and 603 cm^–1 in the Raman spectrum of VC _x /TiO₂, are related to the presence of rutile TiO₂.

The TEM analyses confirmed the presence of smaller VC _x particles in VC _x /Al₂O₃ (9.0 nm) and VC _x /SiO₂ (8.9 nm) compared to the other three catalysts (Figure S4), in accordance with XRD results. Furthermore, STEM-EDX analysis revealed a homogeneous distribution of V along the support in all cases. Characterization by HRTEM allowed the determination of d-spacings of 0.241 and 0.208 nm, assigned to the (222) and (400) facets of V₈C₇ or the (111) and (200) facets of stoichiometric VC, respectively.

H₂-TPR experiments indicated, in all cases, a relatively small amount of H₂ consumption (Table ). In the case of VC _x /Al₂O₃, VC _x /ZrO₂ and VC _x /CeO₂ samples, the H₂ consumption peak at 732–750 K is attributed to the reduction of surface mono- or polymeric oxy-vanadium­(V) species (Figure a). If amorphous V₂O₅ had been formed, it would be reduced at about 852 K and in the case of crystalline V₂O₅ at even higher temperatures. ,− The peak at 1010 K in the H₂-TPR of VC _x /CeO₂ is related to the reduction of VCeO₄. The presence of these species was previously determined by Raman analysis. The slight shifts in the reduction peaks can be attributed to the interaction of VO _x species formed on VC _x with the support. H₂ consumption peaks centered at about 500–600 K in the profiles of VC _x /SiO₂ and bulk VC can be associated with the reduction of vanadium oxy-carbide species, as proposed in prior studies (Figure a). , The presence of various species including oxy-carbide and oxy-vanadium species is related to the H₂-consumption profile of VC _x /TiO₂ (Figure a).

2. H₂ Consumption Determined by TPR, V/M (mol/mol) Ratio from Chemical Analysis (ICP) and XPS, and Dispersion Factor Determined by XPS.

			(V/M)XPS (M = Al,Si,Ce,Zr,Ti)(mol/mol)
catalyst	H2 consumption (molH2/molV)	(V/M)ICP (M = Si,Ce,Zr,Al,Ti) (mol/mol)	fresh	used RWGS	dispersion factor
VC x /Al2O3	0.198	0.388	0.735	0.648	1.894
VC x /SiO2	0.181	0.447	1.664	1.739	3.723
VC x /CeO2	0.050	2.714	2.654	3.139	0.978
VC x /ZrO2	0.160	0.862	1.242	1.801	1.441
VCx/TiO2	0.046	0.529	0.674	0.959	1.274
VC x	0.015

^aAssuming 100% of VC on the sample.

^bCalculated from the fresh catalysts.

2.

(a) H₂-TPR profiles, (b) C 1s XPS levels, and (c) V 2p XPS levels of the supported VC _x catalysts.

XPS was used for the surface characterization of samples (Figure b,c and S5–S9). Specifically, the C 1s core level spectra revealed bands at 282.9–283.2 eV, assigned to the carbide species, (VC _x ) (Figure b). ,− ,, The peak at 284.8 eV was attributed to adventitious carbon and C–C bonds belonging to residual carbon formed during the preparation process. Shoulders observed at BE higher than 284.8 eV were associated with C–O, CO and OC–O bonds. − The V 2p spectra were deconvoluted into 3 or 4 doublets (V 2p _3/2 – V 2p _1/2 ), depending on the sample (Figure c). Peaks at low BE, ranging from 513.5–514.0 eV, were attributed to VC _x . ,− ,, V 2p _3/2 bands at higher BE are assigned to oxy-vanadium species. ,− ,, XPS spectra corresponding to O 1s and characteristic bands of the supports, Al 2p, Si 2p, Ce 3d, Ti 2p and Zr 3d (Figure S5–S9), are discussed in the Supporting Information.

Using the V/M mol/mol (M = Al, Si, Ce, Zr, Ti) determined by XPS and chemical analysis it is possible to calculate the dispersion factor (D) that provides a semiquantitative assessment of the vanadium carbide dispersion onto the support (eq )

$$D=\frac{{(V/M)}_{\mathit{XPS}}}{{(V/M)}_{\mathit{ICP}}}$$ 5

The dispersion factors obtained from both XPS and chemical analysis (ICP) are summarized in Table . VC _x /SiO₂ and VC _x /Al₂O₃ exhibited the highest dispersion factors among all samples. This observation may be attributed to their elevated S _BET values and the small crystallite sizes of VC _x in these two samples.

3.2. RWGS Reaction

The catalytic behavior of the supported catalysts is depicted in Figure . The catalysts showed a high CO₂ conversion ranging from 31 to 52% at 873 K (equilibrium conversion = 60% at 873 K) depending on the support used (Figure a,b). All catalysts exhibited increasing CO selectivity with temperature in the studied range (573–873 K), approaching 100% at 873 K (Figure c), with CH₄ as the main byproduct. For VC _x /TiO₂, the anomalous profile of CO₂ conversion as a function of temperature may be related to a structural transformation of the support during the catalytic test, as discussed later (Figure b). The bulk VC _x catalyst showed a higher CO₂ conversion than the supported VC _x catalysts in the whole range of temperatures considered. However, supported catalysts produced a much higher amount of CO per mol of V (mol_CO mol_V ^–1 h^–1) when compared to bulk VC _x (Figure d).

3.

Catalytic behavior of supported VC _x catalysts in the RWGS reaction. (a) Schematic of the reaction, (b) CO₂ conversion, (c) CO selectivity, (d) CO production, (e) apparent activation energy in the range of 673–773 K, and (f) stability test over VC _x /Al₂O₃ at 873 K for 100 h. Reaction conditions: CO₂/H₂/N₂ = 1/3/1; GHSV = 3000 h^–1; P = 0.1 MPa; m = 300 mg.

Figure e depicts the Arrhenius plots and the apparent activation energies (E _a) in the 673–773 K range for supported and bulk VC _x catalysts. The VC _x /CeO₂ and VC _x /ZrO₂ samples, with crystallite sizes of 36 and 25 nm, respectively, showed much higher E _a than VC _x /Al₂O₃, VC _x /SiO₂, and bulk VC _x , which had smaller crystallite sizes (9–11 nm) (Figure e and Table ). The catalytic behavior of the supported VC _x samples in the RWGS reaction is largely governed by the structural characteristics of the carbide phase on each support. For catalysts containing small VC _x particles, a higher number of carbon vacancies is expected, consistent with previous observations where smaller VC _x crystallites exhibited predominantly C-deficient V₈C₇ domains rather than stoichiometric VC. The superior performance of VC _x /Al₂O₃ and VC _x /SiO₂ is therefore attributed mainly to their higher density of carbon vacancies, which, as discussed earlier, enhance the adsorption of CO₂ and H₂ and lower the corresponding dissociation barriers. These results suggest that the active sites involved in CO₂ and H₂ activation are under-coordinated vanadium atoms located near carbon vacancies.

Additionally, the contribution of support-VC _x interfacial effects cannot be ruled out. For instance, in Mo _x C/Al₂O₃ systems, the surface OH groups of Al₂O₃ have been shown to enhance the RWGS activity by facilitating CO activation through the formation of bicarbonate and formate intermediates at the carbide-support interface, ultimately promoting CO production.

Overall, the VC _x /Al₂O₃ catalyst displayed a CO₂ conversion of 52% at 873 K with nearly 100% CO selectivity, matching or surpassing the activity of other non-noble-metal catalysts reported in the literature (Table S1). This result highlights VC _x /Al₂O₃ as a cost-effective and earth-abundant alternative to traditional noble-metal systems for RWGS reaction.

The VC _x crystallization on the different supports is likely influenced by the interaction between the support and the sol–gel vanadium precursor. This interaction could determine the final VC _x crystallite size. Within this context, for Mo₂C-based catalysts prepared by impregnation and subsequent carburization, a high influence of the support on their characteristics and performance in the RWGS has also been found.

After RWGS catalytic tests, XRD patterns of VC _x /Al₂O₃, VC _x /SiO₂, VC _x /CeO₂ and bulk VC _x catalysts were similar to those of the corresponding fresh ones, (Figures a and S10). However, for the other catalysts, some variations of the crystalline phases associated with the support were noticed. The presence of (Ti_0.9V_0.1)₂O₃ cannot be deduced from the XRD pattern of used VC _x /TiO₂, and an increase in the amount of anatase and rutile phases after the RWGS test can be deduced, compared to the fresh catalyst (Figure S10). For VC _x /ZrO₂, a partial transformation of tetragonal ZrO₂ onto the monoclinic phase during the catalytic test can be proposed (Figure S10). In all cases, after the RWGS catalytic test, no variations in the crystallite sizes of VC _x were determined by XRD with respect to those of the fresh catalysts. Moreover, the presence of surface carbide species was determined in the used catalysts (Figures b,c and S6–S9); see C 1s and V 2p _3/2 at 282.8–283.1 and 513.7–514.0 eV, respectively, attributed to VC _x . However, an increase in the amount of surface oxy-vanadium species took place during the RWGS test, as can be deduced from V 2p spectra (Figures c and S6–S9). The Al 2p, Si 2p, Ce 3d, Ti 2p and Zr 3d spectra characteristic of the different supports were also found for the postreaction catalysts (Figures S5–S9).

4.

Characterization of used VC _x /Al₂O₃ catalyst after RWGS reaction: (a) XRD, (b) C 1s XPS level, (c) V 2p XPS level, and (d) TGA-MS analysis under air of fresh and used VC _x /Al₂O₃. XPS of fresh VC _x /Al₂O₃ was plotted for comparison. Reaction conditions: T = 523–873 K, CO₂/H₂/N₂ = 1/3/1; GHSV = 3000 h^–1; P = 0.1 MPa; m = 300 mg.

Finally, a RWGS stability test was performed using the most efficient catalyst, VC _x /Al₂O₃, at 873 K for 100 h. Constant values of approximately 51% CO₂ conversion and 99.96% CO selectivity were found (Figure f). Moreover, in order to determine the possible coke formation during the RWGS, TGA-MS analysis of fresh and used VC _x /Al₂O₃ was conducted (Figure d). The TGA-MS profiles of the fresh and used catalysts were similar. Initially, the loss of H₂O was noted. Then, CO₂ formation took place with maxima at about 623 K, likely related to the combustion of residual carbonaceous deposits formed during the catalyst preparation. This indicates that after the long RWGS stability test, no additional carbon deposition occurred on the VC _x /Al₂O₃ catalyst.

3.3. MSR Reaction

All supported catalysts exhibited higher methanol conversion than bulk VC _x under the reaction conditions employed (Figure a,b). In all cases, the conversion was low at 573 K and increased with temperature, with the highest value observed for VC _x /ZrO₂ (Figure b). Across the temperature range studied, CH₄ was always the main product evolved (Figures b,c and S11), as previously observed for bulk VC _x , which could be produced via the methanol decomposition reaction outlined as follows

$$C{H}_{3}O{H}_{(\mathit{ad})}\to C{H}_{4(g)}+O_{(\mathit{ad})}$$ 6

5.

Catalytic behavior of supported VC _x catalysts in the MSR reaction: (a) Schematic of MSR reaction, (b) CH₃OH conversion, (c) product molar fraction over VC _x /ZrO₂, (d) product molar fraction over bulk VC _x , and (e) stability test over VC _x /ZrO₂ at 723 K for 20 h. Reaction conditions: CH₃OH/H₂O/N₂ = 1/1/1.2; GHSV = 2500 h^–1; P = 0.1 MPa; m = 300 mg.

Subsequently, the adsorbed O* species generated during methanol decomposition may react with H₂, regenerating the original VC _x surface. These results indicate that VC _x promotes the decomposition of CH₃OH through C–O bond cleavage of intermediate methoxy species, producing CH₄ and surface oxygen.

Among supported catalysts, VC _x /ZrO₂ exhibited the highest H₂ yield (Figures c and S11). For this sample, a slight decrease in CH₄ formation and a corresponding increase in H₂ formation was observed with increasing temperature (Figure c). This behavior can be attributed to the ability of ZrO₂ support to stabilize surface methoxy groups, thereby promoting H₂ formation and suppressing methanation, as previously observed for Mo₂C/ZrO₂ catalysts during MSR. Differences in product distribution likely depend on the nature of the support, as properties such as acidity and the presence of defect sites can play a significant role in the reaction mechanism.

In addition to CH₄ and H₂, CO, CO₂ and HCHO were found in variable amounts depending on the catalyst used. Moreover, very small amounts of C₂H₄ and C₂H₆ were also observed (Figures c and S11). The dehydrogenation of methanol (eq ) leads to HCHO formation, which can decompose producing CO and H₂ (eq ).

$$\mathrm{C}{\mathrm{H}}_3\mathrm{O}{\mathrm{H}}_{(\mathrm{g})}\to \mathrm{HCH}{\mathrm{O}}_{(\mathrm{g})}+{\mathrm{H}}_{2(\mathrm{g})}$$ 7

$$\mathrm{HCH}{\mathrm{O}}_{(\mathrm{g})}\to \mathrm{C}{\mathrm{O}}_{(\mathrm{g})}+{\mathrm{H}}_{2(\mathrm{g})}$$ 8

On the other hand, once CO is formed, it could alternatively react with H₂O producing CO₂ and H₂. This reaction, as well as the HCHO decomposition, would be favored by increasing the temperature. Furthermore, at low temperatures, the product distribution could be influenced by the Water Gas Shift (WGS) reaction equilibrium (eq ).

$$\mathrm{C}{\mathrm{O}}_{(\mathrm{g})}+{\mathrm{H}}_2{\mathrm{O}}_{(\mathrm{g})}\rightleftarrows \mathrm{C}{\mathrm{O}}_{2(\mathrm{g})}+{\mathrm{H}}_{2(\mathrm{g})}$$ 9

Although VC _x /ZrO₂ was the most active catalyst for MSR among those tested in this work, and even outperformed many catalysts reported in the literature (achieving 100% methanol conversion at 673 K), is less suitable for producing H₂ and CO/CO₂-rich products, which are the desired products in conventional reforming processes typically catalyzed by Cu-based catalysts (e.g., Cu/ZnO/Al₂O₃, see Table S2). However, its high CH₄/H₂ yield ratio suggest potential in alternative applications, such as converting stored chemical energy (methanol) back into methane, which can subsequently be utilized within existing natural gas infrastructure.

To assess the stability of the samples, the reaction time was extended to 20 h at 723 K (Figures e and S12). All catalysts displayed severe deactivation. This behavior was previously observed for bulk VC _x - and Mo₂C-based catalysts under MSR conditions. ,, The evolution of the product distribution over time is also presented for all catalysts in Figures e and S12. As expected, in most cases, the decrease in the methanol conversion was accompanied by a significant variation in the product distribution. However, VC _x /ZrO₂, which showed the highest H₂ yield and a methanol conversion of 64% after 20 h at 723 K, showed almost no variation in CH₄ and H₂ concentrations over time (Figure e). For a further analysis of carbon deposits formed on VC _x /ZrO₂ during the MSR reaction, TGA-MS experiments were performed before and after the MSR catalytic test (Figure a); after the initial loss of H₂O, CO₂ evolution was observed in both cases. A larger amount of CO₂ was formed in the used catalyst at slightly higher temperature (623–650 K) than for the fresh sample (573–623 K). This indicates the formation of carbon deposits on VC _x /ZrO₂ during the MSR reaction. Moreover, the simultaneous evolution of water at this high temperature points to the presence of C _x H _y species.

6.

Characterization of used VC _x /ZrO₂ catalyst after MSR reaction: (a) TGA-MS analysis under air of fresh and used VC _x /ZrO₂, (b) XRD pattern, (c) V 2pO 1s XPS levels (combined) and (d) C 1s XPS level. XPS of fresh catalyst was plotted for comparison. Reaction conditions: T = 573–723 K, CH₃OH/H₂O/N₂ = 1/1/1.2; GHSV = 2500 h^–1; P = 0.1 MPa; m = 300 mg.

The characterization of catalysts by XRD after the MSR (Figures b and S13), pointed out that in all cases the VC _x crystallite size on the used and the fresh catalyst was similar. Additionally, no XRD peaks corresponding to crystalline VO _x species were detected in any case. However, XRD patterns of VC _x /ZrO₂ and VC _x /TiO₂ indicated some structural changes in the supports after the MSR reaction (Figures b and S13d). For VC _x /ZrO₂, more intense peaks corresponding to monoclinic ZrO₂ were observed (Figure b). For VC _x /TiO₂, small peaks corresponding to rutile and anatase TiO₂ were detected after the MSR reaction, likely resulting from a partial transformation of the initial solid solution, (Ti_0.9V_0.1)₂O₃, present in the fresh VC _x /TiO₂ (Figure S13d). Furthermore, the presence of carbonaceous deposits can be inferred from the appearance of a broad peak at 2θ = 25.0° in all cases, which is consistent with the results obtained for bulk group 5 TMC catalysts. CH₄ decomposition likely contributed to coke formation. Additionally, a significant reduction in the surface area (S _BET) values of used catalysts was observed when compared to fresh catalysts (Table ), which could be related (at least in part) to the coke deposition during the MSR catalytic test.

Figures c and S14 display the XPS profiles of the V 2p and O 1s core levels for the catalysts after the MSR reaction. The intensity of the V 2p _3/2 band at 513.5–513.8 eV, associated with the presence of VC _x species, decreased compared to the fresh catalysts, while the signal at 517.1–517.5 eV, attributed to oxy-vanadium species, increased. Additionally, new bands appearing at 537–542 eV (O 1s) were attributed to adsorbed CH₃OH and H₂O species. The C 1s core level spectra of the used catalysts (Figures d and S15) showed that the peak corresponding to carbidic species (C 1s at ∼ 283.0 eV) is difficult to be observed, likely due to overlap with the main C 1s peak at 284.8 eV, whose intensity increased significantly relative to the fresh samples. This increase can be attributed to the formation of carbon deposits during the MSR catalytic test, as previously discussed.

3.4. DFT Calculations to Understand Coke Formation

To elucidate the experimental observations of coke formation, DFT calculations were performed to investigate the decomposition mechanisms of CO₂ and CH₃OH on VC and V₈C₇ slab models. Initially, we identified all plausible pathways for atomic carbon formation on V₈C₇ from CO₂ and CH₃OH. For CO₂ decomposition, only the direct C–O dissociation pathway (CO₂ → CO → C) was considered. In contrast, CH₃OH decomposition may proceed via multiple pathways, depending on which bond is cleaved initially. We explored several bond-breaking elementary steps in methanol decomposition, calculated the corresponding energy barriers, and mapped the lowest energy pathway (Figure S16).

Our results indicated that the most favorable route for C* formation from CH₃OH on both VC and V₈C₇ begins with the breaking of the O–H bond, forming CH₃O* (Figure ). On VC, this pathway continues with C–H bond cleavage to form CH₂O* and subsequently CHO*, followed by C–O bond scission to produce CH*, and ultimately C*. In contrast, on V₈C₇, the CH₃O* species preferentially undergoes C–O bond scission prior to C–H activation, yielding CH₃* first, which is subsequently dehydrogenated to form C*. Notably, on the V₈C₇ model, the vacancy site is not always available for subsequent bond dissociation steps. Specifically, after CO₂* dissociation, the vacancy is occupied by an O* atom, so the next C–O bond scission occurs adjacent to the vacancy. Similarly, for CH₃OH, the vacancy site facilitates three of the five bond dissociation steps, while the remaining two occur at sites adjacent to the vacancy. The transition state geometries of the most favorable reaction steps are presented in Figure S17.

7.

(a, b) Potential energy diagrams and (c, d) free energy diagrams for (a, c) CO₂ decomposition and (b, d) CH₃OH decomposition. Gray and green lines correspond to VC and V₈C₇, respectively. The free energy diagrams have been computed at 600 K and 0.1 MPa and the values reported are in units of eV. The reference states used for the potential and free energy diagrams are the gas-phase reactant species.

Comparing the potential and Gibbs free energy profiles for CO₂ and CH₃OH decomposition reveals that methanol decomposition is significantly more favorable than CO₂ decomposition on both VC and V₈C₇ (Figure ). This finding aligns with our experimental observations, including stability tests, XRD, XPS, and TGA-MS profiles. Specifically, C–O bond cleavage in adsorbed CO* during CO₂ decomposition has high potential energy barriers of 2.67 eV on VC and 2.64 eV on V₈C₇ (Figure a), making it unlikely to occur under the reaction conditions tested (Figure ). In contrast, when carbon forms bonds with hydrogen atoms, as in CH₃O*, the C–O bond becomes significantly weaker. For example, the C–O bond dissociation of CH₃O* on V₈C₇ has a potential energy barrier of only 1.11 eV (Figure b), which is accessible under reaction conditions (Figure ).

The presence of carbon vacancies systematically reduces the energy barriers for both CO₂ and CH₃OH decompositions and stabilizes C* and CH* species. Specifically, the overall potential energy barrier for C* formation from CO₂ decreases from 3.18 eV on VC to 1.72 eV on V₈C₇, and for C* formation from CH₃OH, it decreases from 2.30 eV on VC to 1.15 eV on V₈C₇. Therefore, coke formation is expected to be more pronounced on VC _x samples with a higher proportion of the V₈C₇ phase.

Furthermore, the energy profiles suggest that coke deposits under MSR conditions likely consist of C _x H _y * species rather than solely C* atoms, as the CH* → C* dissociation requires overcoming high potential energy barriers of 2.03 eV on VC and 1.81 eV on V₈C₇. This is supported by the TGA-MS profile of the used VC _x /ZrO₂ catalyst in MSR, which showed the presence of H₂O at the CO₂ release temperature (Figure a), indicating that the formed coke may comprise C _x H _y * species. The CH* intermediate could initiate the formation of long, partially hydrogenated C _x H _y rings or chains, which can deactivate the VC _x catalyst. Overall, the DFT calculations demonstrated that VC _x catalysts are more prone to coking during CH₃OH decomposition than during CO₂ decomposition.

4. Conclusions

In this study, a series of vanadium carbide catalysts supported on Al₂O₃, SiO₂, TiO₂, CeO₂ and ZrO₂ were successfully synthesized and evaluated for CO₂ and CH₃OH conversion under the RWGS and MSR conditions, respectively.

Under RWGS conditions, all catalysts exhibited high selectivity for CO, reaching up to 100% at 873 K. Compared to bulk VC _x , the supported catalysts demonstrated significantly higher CO production per mol of V. This enhanced activity is attributed to their smaller VC _x particle sizes and consequently higher concentrations of carbon vacancies, which facilitate the adsorption and dissociation of CO₂ and H₂, as well as to favorable support-VC _x interfacial effects. Among the studied samples, VC _x /Al₂O₃ showed the best performance, exhibiting high stability and no evidence of carbon deposition after 100 h on stream at 873 K. The superior behavior of this catalyst could be attributed to the −OH groups on the Al₂O₃ surface, which likely assist CO₂ activation via bicarbonate and formate intermediates at the carbide-support interface.

In the MSR reaction, the supported VC _x catalysts achieved higher methanol conversions than bulk VC _x , with VC _x /ZrO₂ being the most active in the 673–723 K range. However, in all cases, CH₄ was the main product, and the catalysts suffered substantial deactivation due to coke formation.

Postreaction characterization confirmed the presence of VC _x phases in all spent catalysts via XRD, with no detectable crystalline VO _x species. XPS analysis revealed that surface VC _x was preserved under RWGS conditions but underwent partial oxidation during MSR.

DFT calculations offered mechanistic insights into coke formation pathways on VC and V₈C₇ surfaces. The computed energy landscapes revealed that methanol decomposition proceeds via pathways with significantly lower energy barriers than CO₂ decomposition, and thus CH₃OH is a more potent precursor for coke. The presence of carbon vacancies was found to further lower the energy barriers, thereby increasing the likelihood of coke formation on V₈C₇-rich VC _x samples. Additionally, the results suggest that coke species formed under MSR conditions are partially hydrogenated C _x H _y * intermediates rather than pure C* atoms.

Overall, this work demonstrates that supported VC _x catalysts are highly active and selective for the RWGS, while bulk VC _x exhibits lower activity. Conversely, under MSR conditions, VC _x catalysts are more susceptible to coking, particularly when V₈C₇ phases dominate. The higher catalytic activity of the supported materials arises from the combined effects of smaller crystallite size (and hence more carbon vacancies) and specific metal–support interfacial interactions that enhance CO₂ and CH₃OH activation. These findings highlight the importance of tailoring the catalyst surface structure and metal–support interface to minimize coke formation and improve the long-term catalytic stability of transition-metal carbide catalysts.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsami.5c16601.

• This includes additional details on the DFT slab models, N₂ adsorption–desorption isotherms, pore size distribution, Raman spectra, TEM characterization, XPS profiles, XRD patterns, product distribution, catalytic tests, methanol and CO₂ decomposition pathways and transition state configurations (PDF)

A.P.: Conceptualization, Methodology, Validation, Formal analysis, Investigation, WritingOriginal draft, Writingreview and editing. S.S.Y.: Formal analysis, Investigation, Writing-Original draft. H.P.: Validation, Formal analysis, Investigation, WritingOriginal draft, Writingreview and editing. P.R.d.l.P.: Methodology, Resources, Writingreview and editing, Supervision, Project administration. M.S.: Methodology, Resources, Writingreview and editing, Supervision, Project administration. N.H.: Methodology, Resources, Writing-review and editing, Supervision, Project administration.

The authors declare no competing financial interest.