Comparative Assessment of First-Row 3d Transition Metals (Ti-Zn) Supported on CeO₂ Nanorods for CO₂ Hydrogenation

Abstract

Herein, motivated by the excellent redox properties of rod-shaped ceria (CeO₂-NR), a series of TM/CeO₂ catalysts, employing the first-row 3d transition metals (Ti, V, Cr, Mn, Fe, Co, Ni, Cu, and Zn) as active metal phases, were comparatively assessed under identical synthesis and reaction conditions to decipher the role of active metal in the CO₂ hydrogenation process. Notably, a volcano-type dependence of CO₂ hydrogenation activity/selectivity was disclosed as a function of metal entity revealing a maximum for the Ni-based sample. Ni/CeO₂ is extremely active and fully selective to methane (Y_CH4 = 90.8% at 350 °C), followed by Co/CeO₂ (Y_CH4 = 45.2%), whereas the rest of the metals present an inferior performance. No straightforward relationship was disclosed between the CO₂ hydrogenation performance and the textural, structural, and redox properties, whereas, on the other hand, a volcano-shaped trend was established with the relative concentration of oxygen vacancies and partially reduced Ce³⁺ species. The observed trend is also perfectly aligned with the previously reported volcano-type dependence of atomic hydrogen adsorption energy and CO₂ activation as a function of 3d-orbital electron number, revealing the key role of intrinsic electronic features of each metal in conjunction to metal–support interactions.

1. Introduction

Global warming has emerged as a pressing environmental issue due to the observed continuous increase in global temperature of 0.2 °C per decade, which is directly linked to the escalation of greenhouse gas emissions [1,2]. As the concentration of greenhouse gasses in the atmosphere continues to rise, the detrimental impact on the environment is greatly enhanced. Specifically, carbon dioxide (CO₂), which originates mainly from fossil fuel energy sources, is a major contributor to the accumulation of greenhouse gasses. To mitigate greenhouse gas emissions, it is imperative to optimize the utilization of fossil fuels and implement effective Carbon Capture and Utilization (CCU) technologies [1,3,4].

A sustainable CCU strategy involves the reduction of captured CO₂ via its reaction with hydrogen, a process known as CO₂ hydrogenation. This method presents an effective means of valorizing CO₂ emissions and efficiently storing surplus power from non-intermittent renewable energy sources (such as solar and wind) in the form of “green” hydrogen [5,6,7]. Coupling a CCU process with a water electrolysis unit producing “green” hydrogen inhibits the release of CO₂ emissions, while at the same time converting CO₂ into an energy carrier, such as synthetic natural gas (SNG), which consists mainly of methane and can be integrated in the present natural gas grid network, thus closing the carbon cycle [8,9]. Under atmospheric pressure conditions, it can be achieved by either the mildly endothermic reverse water–gas shift (rWGS) reaction (Equation (1)), which yields CO, or the highly exothermic methanation reaction, commonly known as the “Sabatier reaction” (Equation (2)), which produces CH₄ [10,11,12].

$${\mathrm{C}\mathrm{O}}_2+{\mathrm{H}}_2\leftrightarrow \mathrm{C}\mathrm{O}+{\mathrm{H}}_2\mathrm{O},\Delta \mathrm{H}=+41.2 \mathrm{k}\mathrm{J}/\mathrm{m}\mathrm{o}\mathrm{l}$$

(1)

$${\mathrm{C}\mathrm{O}}_2+{4\mathrm{H}}_2\leftrightarrow {\mathrm{C}\mathrm{H}}_4+{2\mathrm{H}}_2\mathrm{O},\Delta \mathrm{H}=-164.7 \mathrm{k}\mathrm{J}/\mathrm{m}\mathrm{o}\mathrm{l}$$

(2)

Extensive research has explored various catalytic systems, often summarized in the literature covering catalysts for reactions like rWGS [13,14,15,16,17,18] and CO₂ methanation [17,19,20,21,22,23]. Notably, composite catalytic systems involving metals supported on reducible metal oxides (e.g., CeO₂ and ZrO₂) or combinations of them have been extensively studied. Among the investigated oxide materials, ceria (CeO₂) has gained significant attention as a supporting carrier due to its exceptional properties such as oxygen storage capacity, oxygen mobility, strong metal–support interactions, and rapid change between its two oxidation states (Ce³⁺ and Ce⁴⁺) [15,24,25,26,27,28]. Although ceria-based noble metals (Ru, Rh, and Pd) have demonstrated excellent catalytic activity, their high cost and limited availability make them less preferable from a techno-economic point of view [29,30,31,32,33,34]. Therefore, recent efforts have focused on the rational design of cost-efficient and highly active non-noble metal catalysts, with particular emphasis on the 3d transition metals, which can adequately adsorb and activate CO₂ molecules [18,25,35,36,37,38,39,40,41]. In this direction, we recently showed that by appropriately adjusting the size and shape of ceria carrier via the use of appropriate synthetic routes, extremely active TMs/CeO₂ catalysts can be obtained for the CO₂ hydrogenation reaction with similar reactivity to that of noble metals [42]. For instance, it was shown that Ni catalysts supported on ceria nanoparticles of rod-like morphology exhibit excellent hydrogenation performance, ascribed mainly to the enhanced redox properties of ceria nanorods in conjunction with the synergistic interactions between nickel and ceria nanoparticles [25,43,44]. On the other hand, Cu/CeO₂ samples demonstrate excellent rWGS performance, revealing the key role of the metal entity in CO₂ adsorption/activation and in turn in activity and selectivity towards CO or CH₄ [12,42].

In the case of the rWGS reaction, two main reaction pathways have been considered, namely the redox and the associative mechanism. In the first case, CO₂ is initially activated via its adsorption on metal sites or oxygen vacancies, followed by the formation of carbonyl species that are finally desorbed as CO_(g). On the other hand, during the associative mechanism, the adsorbed CO₂ reacts with H adatoms towards the formation of active intermediate species (such as formates) that finally decompose to CO and H₂O [6,15,16,45,46,47,48]. The main difference between the two rWGS reaction mechanisms is whether the dissociated hydrogen atoms engage in the reaction and generate carbon-containing intermediates (e.g., carboxyls and formates) with the adsorbed CO₂ on the catalyst surface [15,49,50]. Similarly, the CO and formate-mediated routes have been proposed for CH₄ production. In the first route, the chemisorbed CO derived by CO₂ dissociation is further hydrogenated to CH₄, whereas, in the second one, formate species (HCOO*) react with hydrogen adatoms [51,52,53]. In view of these aspects, CO₂ dissociation is considered a key step in the overall reaction process, affected to a great extent by the electronic nature of active sites and the charge transfer to CO₂ molecule [47,51,52].

In view of the above discussion, various transition metals have been studied in the context of CO₂ hydrogenation reactions. Remarkably, the first-row transition metals have shown a capacity to adsorb and activate CO₂ by facilitating charge transfer from the metal phase to the CO₂ molecule [40,54]. Furthermore, computational studies have played a crucial role in enhancing our comprehension of the structural aspects and energetic characteristics of the interplay between metal surfaces and CO₂ chemistry. To this end, Ray and Deo [55] have proposed the local d-density of states (d-DOS) at the Fermi level of the pristine surface as a potential descriptor for CO₂ methanation over alumina-supported Ni and Ni-based alloy catalysts. In particular, the catalytic activity was linearly correlated with the number of electronic excitations from occupied to unoccupied states at the lowest energy cost [55], with this electronic property demonstrating the surface’s ability to react to an outside perturbation (e.g., a reactant present above the surface) by displaying a population of unoccupied states [56,57]. Density functional theory (DFT) studies reveal that metals with less than 2 empty 3d orbitals exhibit a unique CO₂ hydrogenation selectivity, favoring formate formation via a specific co-adsorption configuration of a hydrogen atom and CO₂ [58]. In particular, a “volcano” curve on barrier energy in the formate pathway was established with Ni being located at the peak with the lowest barrier energy [58].

Despite the intense interest in the field, there is no—to the best of our knowledge—systematic experimental work on the impact of 3d transition metals on the key activity descriptors and in turn on the CO₂ hydrogenation performance under identical synthesis and catalytic evaluation conditions. The latter could bypass the variability of the synthesis procedure and reaction conditions followed in the relevant literature, revealing solely the role of the metal entity. In this regard, a series of TM/CeO₂ catalysts with a constant atomic ratio of M/Ce = 0.25 were prepared, using the first-row 3d transition metals (Ti, V, Cr, Mn, Fe, Co, Ni, Cu, and Zn) as active phases for ceria nanorods (CeO₂-NR). The catalytic materials were thoroughly characterized in terms of their textural (N₂ adsorption at –196 °C), structural (X-ray diffraction, XRD), morphological (Scanning electron microscopy/Energy dispersive spectroscopy, SEM/EDS, Transmission electron microscopy, TEM), surface (X-ray photoelectron spectroscopy, XPS) and redox (Temperature-programmed reduction, H₂-TPR) properties in an attempt to gain insight into the role of metal entity in CO₂ hydrogenation. Notably, no particular correlation was disclosed between the aforementioned textural, structural, and redox properties and the catalytic performance. However, based on the XPS analysis, a volcano-shaped trend among the relative concentration of oxygen vacancies, the amount of Ce³⁺ species, and the CO₂ conversion rate and CH₄ selectivity were established. Moreover, a volcano-type dependence of both activity and selectivity as a function of 3d-orbital electron number was disclosed, with Ni located at the peak of the volcano curve. These findings can be interpreted on the basis of the intrinsic electronic features of each metal in relation to their ability to activate CO₂ and dissociate H₂, in compliance with relevant theoretical studies.

2. Results

2.1. Textural and Structural Characterization (BET (Brunauer–Emmett–Teller), XRD)

The main textural and structural characteristics of bare CeO₂-NR and TM/CeO₂ samples (atomic ratio M/Ce = 0.25) are summarized in

Table 1. Textural and structural characteristics of CeO₂ and TM/CeO₂ samples.

Sample	Nominal Metal Loading	EDS Analysis		BET Analysis	XRD Analysis		TEM Analysis
		Atomic Ratio M/Ce	Metal Content (wt%)	SBET (m2/g)	Average Crystallite Size (nm)		MxOy Particle Size (nm)
					CeO2	MxOy
CeO2-NR	-	-	-	79	15	-	-
Ti/CeO2	6.5	0.24	6.3	-	11	20	16
V/CeO2	6.9	0.28	7.6	-	14	45	28
Cr/CeO2	7.0	0.26	7.2	-	11	- 1	10
Mn/CeO2	7.4	0.22	6.5	-	11	- 1	15
Fe/CeO2	7.5	0.21	6.3	69	10	7	11
Co/CeO2	7.9	0.26	8.1	72	14	16	15
Ni/CeO2	7.9	0.25	7.8	72	14	23	10
Cu/CeO2	8.5	0.25	8.6	75	12	43	16
Zn/CeO2	8.7	0.24	8.3	76	12	44	41

¹ Not calculated due to the absence of the corresponding XRD peaks.

. Bare CeO₂ demonstrates a BET surface area of 79 m²/g while a slight decrease in the BET surface area is observed upon the addition of the transition metals into ceria support.

XRD patterns of bare CeO₂-NR and TM/CeO₂ samples are presented in Figure 1. The main diffraction peaks of bare CeO₂ correspond to the following planes: (111), (200), (220), (311), (222), (400), (331), and (420) and are attributed to the face-centered cubic (fcc) fluorite structure of ceria (Fm3m symmetry, no. 225) [59,60]. Furthermore, in the case of TM/CeO₂ samples (M: Ti, Fe, Co, Ni, Cu, and Zn), smaller peaks of their corresponding oxides are detected, demonstrating the existence of TiO₂, Fe₂O₃, Co₃O₄, NiO, CuO, and ZnO, respectively, with no other crystal phases existing, apart from ceria. In contrast, Mn/CeO₂ and Cr/CeO₂ samples do not present any diffraction peaks, except from the ceria crystal phase, a fact that can be attributed to the low metal loadings in conjunction with the high metal dispersion, rendering the calculation of their crystallite sizes impossible. In the XRD pattern of V/CeO₂, apart from ceria characteristic peaks, a series of XRD peaks located at 2θ values around 18.2°, 24.0°, 32.4°, 34.2°, 39.0°, 43.5°, 46.4°, 47.9°, 49.2°, 55.5°, and 60.2° were observed, which are related to the (101), (200), (112), (220), (301), (103), (321), (312), (400), (420) and (332) planes of CeVO₄ [61,62], respectively. Therefore, it is evident that in the case of V/CeO₂, a mixed crystal phase has been formed via the impregnation method, including CeO₂ and CeVO₄.

The average crystallite sizes of CeO₂ and M_xO_y phases for both bare ceria and TM/Ceria samples are calculated by Scherrer’s equation (Equation (4)) and are presented in Table 1. The crystallite size of ceria does not significantly change with the impregnation of the metal entity (11–15 nm), indicating that ceria structural properties remain unaltered upon metal addition. The crystallite sizes of the active metal phases are following the order: CeVO₄ (45 nm) > ZnO (44 nm) > CuO (43 nm) > NiO (23 nm) > TiO₂ (20 nm) > Co₃O₄ (16 nm) > Fe₂O₃ (7 nm).

2.2. Morphological Characterization (SEM/TEM)-Elemental Analysis (SEM/EDS)

Elemental analysis was performed by SEM along with energy dispersive X-ray spectrometry (SEM/EDS) on the TM/CeO₂ samples and the results (atomic ratio of M/Ce, metal content in weight percentage) are presented in Table 1. The atomic ratio of M/Ce is in good agreement with the nominal one. Additionally, the SEM images of the TM/CeO₂ samples are depicted in Figure 2. From SEM/EDS analysis, it is revealed that there is a uniform distribution of ceria and metal phases on the entire surface of the mixed oxides.

The morphological characteristics of all samples were investigated in more detail by TEM analysis, as shown in Figure 3. It is apparent that CeO₂ support retains its nanorod morphology upon the addition of the metal entity, with isolated M_xO_y particles of irregular shape being detected in all TM/CeO₂ samples. Moreover, the M_xO_y particle sizes (Table 1), along with the particle size distribution (Figure 3) were determined by TEM analysis. M_xO_y particle size varied widely, ranging between 10 and 41 nm (Table 1).

2.3. Reducibility Studies (H₂-TPR)

In order to gain insight into the impact of the metal entity on the reducibility, TPR measurements were performed in the temperature range of 100–1000 °C, using H₂ as a reducing agent. More particularly, indicative catalytic systems (vide infra), in terms of conversion and selectivity, were chosen for the reducibility studies, namely, the Ni/CeO₂ and Co/CeO₂ catalysts, as highly active for the Sabatier reaction and the Fe/CeO₂ and Cu/CeO₂ ones, as representatives for the rWGS reaction. As depicted in Figure 4, bare CeO₂-NR exhibits two broad reduction peaks centered at 545 °C and 788 °C which are attributed to the surface (O_s) and bulk (O_b) oxygen reduction in ceria, respectively [63,64]. The Fe/CeO₂ sample exhibits four reduction peaks at 390 °C, 465 °C, 588 °C, and 759 °C with the peaks at 465 °C and 759 °C being ascribed to ceria surface oxygen and bulk oxygen reduction, respectively, while the peaks at 390 °C and 588 °C corresponding to the reduction in the iron species, namely Fe₂O₃ to Fe₃O₄ and Fe₃O₄ to Fe⁰ [65]. As for the Co/CeO₂ sample, it exhibits two main reduction peaks at 318 °C and 388 °C, ascribed to the stepwise reduction of Co₃O₄ to CoO and CoO to metallic Co, respectively [66]. Concerning the Cu/CeO₂ sample, the low-temperature peak at 181 °C is attributed to the reduction of finely dispersed CuO_x species strongly interacting with the ceria surface, while the peak at 217 °C can be ascribed to the reduction of Cu₂O to Cu⁰ due to the interaction of the Cu–[O_x]–Ce structure [67,68,69]. Nickel–ceria nanorods exhibit three peaks; the peak at 220 °C can be attributed to the reduction of surface oxygen located at the defects of the ceria surface, the peak at 288 °C corresponds to the reduction of adsorbed oxygen species associated with the formation of Ni–O–Ce structure, while the peak at 353 °C is ascribed to the reduction of the well-dispersed NiO phase interacting strongly with the ceria support [70,71]. All TM/CeO₂ samples exhibit a high-temperature peak in the range of 747–793 °C which is attributed to the ceria sub-surface oxygen reduction. As illustrated in Figure 4, the addition of the metal phase facilitates ceria surface oxygen reduction.

Furthermore, as depicted in

Table 2. Redox properties of CeO₂-NR and representative TM/CeO₂ samples.

Sample	H2 Consumption (mmol H2 g−1) 1	Theoretical H2 Consumption (mmol H2 g−1) 2	Peak Temperature (°C)
CeO2-NR	0.6	-	-	545	-	788
Fe/CeO2	1.6	1.9	390	465	588	759
Cu/CeO2	1.8	1.3	181	-	217	793
Ni/CeO2	1.8	1.3	220	288	353	747
Co/CeO2	2.4	1.7	318	-	388	789

¹ Estimated by the area of the corresponding TPR peaks, which is calibrated against a known amount of CuO standard sample. ² Calculated as the amount of H₂ required for the complete reduction in fully oxidized M_xO_y to M⁰ on the basis of the metal nominal loading.

, hydrogen consumption surpasses the theoretical amount for all TM/CeO₂ samples, apart from the Fe/CeO₂ sample, revealing that the addition of the metallic phase enhances the reducibility of ceria nanorods, due to synergistic metal–support interactions.

2.4. Surface Analysis (XPS)

In order to assess the impact of surface chemistry on the intrinsic properties and catalytic performance, XPS analysis was conducted over the optimum catalytic materials (TM/CeO₂, TM: Fe, Cu, Co, Ni) in their reduced states. In Figure 5a, the Ce 3d XPS spectra are depicted with the Ce 3d_3/2 and Ce 3d_5/2 states corresponding to the u and v lines, respectively. The three pairs of peaks u (902.1 eV), u″ (908.5 eV), u‴ (917.3 eV) and v (883.2 eV), v″ (889.8 eV), v‴ (899.5 eV) characterize the Ce⁴⁺ species, while the u′ (904.7 eV) and v′ (886.2 eV) peaks are attributed to the Ce³⁺ species [72,73,74].

The concentration of Ce³⁺ species can be calculated from the following equation (Equation (3)) [72]:

$$\left[{\mathrm{C}\mathrm{e}}^{3+}\right]={\displaystyle \frac{\mathrm{C}\mathrm{e}\left(\mathrm{I}\mathrm{I}\mathrm{I}\right)}{\mathrm{C}\mathrm{e}\left(\mathrm{I}\mathrm{I}\mathrm{I}\right)+\mathrm{C}\mathrm{e}\left(\mathrm{I}\mathrm{V}\right)}}$$

(3)

where Ce(III) and Ce(IV) denote the corresponding sums of the integrated peak areas associated with the Ce³⁺ and Ce⁴⁺ XPS signals, respectively (

Table 3. XPS results of representative reduced samples.

Sample	Oads/Olat	Ce3+ (%)
CeO2-NR	0.59	27.2
Fe/CeO2	0.43	26.9
Cu/CeO2	0.54	25.5
Co/CeO2	0.63	29.9
Ni/CeO2	0.65	30.4

). Ni catalyst exhibits the highest amount of Ce³⁺ species (30.4%), followed by Co (29.9%), Fe (26.9%) and Cu (25.5%). The significant role of Ce³⁺ species in the CO₂ hydrogenation process (activation of CO₂ and formation of active intermediates) has been revealed in the literature [75,76,77], with these sites being associated with the abundance of oxygen vacancies and the extent of metal–support interactions [78]. Cao et al. [79] demonstrated a volcano-type curve between activity and surface Ce³⁺ fractions for CO₂ hydrogenation to CO over ceria catalysts. Moreover, under CO₂ hydrogenation reaction conditions, a Ni catalyst supported on mesoporous ceria nanofiber assembly was found to lead to the formation of active Ce³⁺ sites located at the interface, thus boosting the catalytic performance and promoting the formate pathways [80].

Figure 5b illustrates the O 1s XPS spectra with two peaks being observed; the peak at 529 eV represents the lattice oxygen (O_lat), while the peak at 531 eV is ascribed to surface adsorbed oxygen species (O_ads) [81,82]. Analysis of the O 1s XPS spectra of the reduced samples led to the determination of the O_ads/O_lat ratio which constitutes a useful tool for the estimation of the relative oxygen vacancy concentration [54,81]. As shown in Table 3, the Ni catalyst presents the highest O_ads/O_lat ratio, with the following order being observed: Ni/CeO₂ > Co/CeO₂ > Cu/CeO₂ > Fe/CeO₂, in good agreement with the order of Ce³⁺ species (Table 3).

Figure 6a presents the Cu 2p XPS spectrum of the reduced Cu/CeO₂ catalyst which includes two main peaks at 953.7 eV and 933.8 eV, ascribed to Cu 2p_1/2 and Cu 2p_3/2 states, respectively. The existence of satellite peaks at 943 and 962 eV confirms the presence of Cu²⁺ species [69,83]. The Co 2p XPS spectrum for the reduced Co/CeO₂ sample (Figure 6b) exhibits two main peaks of Co 2p_1/2 (796.6 eV) and Co 2p_3/2 (780.6 eV), as well as shake-up satellites at 786 eV which are characteristics of Co²⁺ [84,85]. Moreover, Figure 6c depicts the Fe 2p XPS spectrum of the reduced Fe/CeO₂ catalyst. There are two peaks at 711 and 724 eV, assigned to the Fe 2p_3/2 and Fe 2p_1/2, respectively, and two satellite peaks at 719 and 734 eV, designating the presence of Fe³⁺ [86,87]. The satellite peaks overlap with the Auger Ce MNN peaks. The formation of Fe²⁺ species results from the iron oxide–ceria interaction via an interfacial redox process: xFe₂O₃ + (2 − y)CeO_2−x → xFe₂O_{3 − y} + (2 − y)CeO₂ [86,88,89]. In addition, Figure 6d illustrates the Ni 2p XPS spectrum of the reduced Ni/CeO₂ sample, with two peaks at 855 eV (Ni 2p_3/2) and at 873 eV (Ni 2p_1/2), along with a broad satellite peak at 861 eV, associated with the presence of Ni³⁺ species (Ni₂O₃) [90,91]. According to the XPS results of the reduced catalysts, metal species exist in different oxidation states which can be related to the strength of the metal–support interactions, as indicated in the literature [85,90]. Nevertheless, the difficulty in establishing reliable relationships between metal oxidation states and catalytic performance should be underscored, on the ground of ex situ nature of XPS in conjunction with different metal entities employed as active phases in the present work.

2.5. Catalytic Performance

Figure 7 depicts the CO₂ conversion and CO selectivity performance in the temperature range of 200–500 °C of all TM/CeO₂ samples. For comparison purposes, the corresponding performance of bare ceria nanorods is also included. The main conclusions that can be drawn by the comparative assessment of the first-row 3d transition metals supported on CeO₂ nanorods are the following:

(i)

The CO₂ conversion of TM/CeO₂ is strongly dependent on the metal entity, following the general trend: Ni > Co > Cu > Fe > Zn > Cr ≈ Ti ≈ V ≈ Mn.

(ii)

The Ni/CeO₂ catalyst exhibits by far the best performance overall, in terms of maximum conversion and light-off temperature, offering ~90% conversion at 300 °C, which is close to equilibrium conversion (94%) under the employed experimental conditions. Equally importantly, Cu/CeO₂ was the most active rWGS material, attaining a CO₂-to-CO conversion of 47% at 450 °C, approaching equilibrium conversion (48.5%).

(iii)

The early 3d transition metals (Ti, V, Cr, Mn) are almost inactive (i.e., X_CO2 < 10% at T < 400 °C), exhibiting an even lower conversion performance as compared to bare ceria nanorods.

(iv)

All TM/CeO₂ systems, except for Ni and Co, are mainly selective to CO. In particular, the Ni-based catalyst is completely selective to CH₄ in the whole temperature range investigated, whereas the Co-based sample exhibits intermediate selectivity values depending on temperature. All other TM/CeO₂ catalysts are completely selective to CO.

(v)

In terms of CO₂ conversion and CO/CH₄ selectivity, no catalyst deactivation occurred during short-term (24 h) stability experiments over the optimum catalysts (Fe, Co, Ni, Cu) supported on ceria nanorods (not shown for brevity).

In order to gain a more thorough insight into the particular role of the metal entity in CO₂ hydrogenation performance, the normalized reaction rate of CO₂ consumption (r_CO2) and the corresponding selectivity to methane (S_CH4) are plotted at a specific reaction temperature (350 °C), as at this temperature noticeable differences can be unveiled among the samples (Figure 8). Markedly, a volcano-type dependence of CO₂ conversion and CH₄ selectivity was disclosed for first-row 3d transition metals, i.e., Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn with a maximum obtained for Ni. In particular, the Ni/CeO₂ catalyst exhibits by far the optimum hydrogenation performance in terms of CO₂ consumption rate (~800 μmol g_Ni⁻¹s⁻¹) and CH₄ selectivity (100%), followed by Co (r_CO2 = 445 μmol g_Co⁻¹s⁻¹, S_CH4 = 90%). The 3d early transition metals (Ti–Fe), as well as the 3d late metals (Cu and Zn), exhibit much lower reactivity (r_CO2 < 200 μmol g_M⁻¹s⁻¹), while being selective to CO (Figure 8).

3. Discussion

In the present study, the CO₂ hydrogenation performance of first-row 3d transition metals (Ti–Zn) supported on CeO₂ nanorods (atomic ratio M/Ce = 0.25) were comparatively assessed employing identical synthesis and catalytic evaluation procedures, in an attempt to decipher the role of metal entity and metal–support interactions in the catalytic activity and selectivity. The as-prepared samples were characterized by various physicochemical methods to gain insight into the impact of metal nature on the structural, textural, morphological, and redox properties. On the basis of the obtained results, it could be argued that Ni/CeO₂—and to a lower extent Co/CeO₂—is active for the Sabatier reaction, whereas the rest of the transition metals exhibit a much inferior reactivity while being selective to CO (rWGS). Remarkably, a volcano-type dependence of CO₂ hydrogenation activity and selectivity was disclosed as a function of metal entity (Ti–Zn) revealing a maximum for the Ni-based sample. In particular, Ni/CeO₂ is highly active and fully selective to CH₄ (Y_CH4 = 90.8% at 350 °C), followed by Co (Y_CH4 = 45.2% at 350 °C). The rest of the transition metals exhibit a much lower CO₂ conversion, being also selective to CO instead of CH₄. In an attempt to disclose possible structure-performance relationships and to reveal key activity descriptors, the obtained catalytic findings are discussed below in relation to the present characterization results and the intrinsic characteristics of each metal, taking into account the main reaction pathways suggested for Sabatier and rWGS reactions.

Firstly, the rWGS and Sabatier reaction mechanisms are considered, both constituting a topic of intensive debate. For the rWGS reaction, two main pathways have been in general considered, i.e., the redox and the associative mechanism, depending on the direct or not participation of H₂ in CO₂ activation [47,48]. In the redox mechanism, the gaseous CO₂ is initially activated via its adsorption on an electron donor site, such as a metal or oxygen vacancy, followed by the formation of carbonyl species that are finally desorbed as CO_(g). In this specific route, the role of H₂ is limited to reducing the oxidized sites and sustaining the reduction/oxidation cycle on the catalyst surface [6,49]. Conversely, within the associative mechanism, CO₂ chemisorbed species react with H adatoms derived from H₂ dissociation for the formation of intermediate species (such as formates) that eventually decompose to CO and H₂O [6,13,41].

Similarly, two main paths have been proposed for CO₂ methanation, i.e., the CO and formate route. The former is related to the dissociation of CO_2(g) to CO_(ad) which is fully hydrogenated to CH_4(ad) [51,52]. In the second route, formate species (HCOO*) instead of CO is the main intermediate, followed by a reaction with chemisorbed hydrogen adatoms [51,52,92]. With regard to the role of metal–ceria catalysts in the mechanism of hydrogen activation, it is considered that gas-phase H₂ is activated via dissociation on metal particles that are increasingly active towards hydrogen activation in the vicinity of ceria surfaces. The adsorbed hydrogen adatoms rapidly migrate to the ceria support via spillover phenomena and can react with low-energy and abundant C-containing intermediates onto CeO_2–x and result in gaseous methane molecules [93]. In this regard, in situ DRIFT experiments have revealed the involvement of the direct formate hydrogenation pathway in CO₂ methanation over a nickel–ceria catalyst prepared via sol–gel, with this synthetic procedure promoting oxygen vacancy formation and exposure of the Ni (111) crystal plane [94].

In principle, the CO₂ activation pathway is strongly affected by the metal entity and metal–oxygen interaction [95]. In this regard, it has also been found that the adsorption strength of CO₂ on metal surfaces is affected by the electronic nature of active sites, i.e., the d-band center of the metal as well as the charge transfer to a CO₂ molecule [96]. According to the d-band center model, the adsorption energy of a given molecule can uniformly decrease (increase) from one transition metal surface to another in which the d-electron number increases (decreases). Nevertheless, exceptions from the model can occur possibly due to the adsorbate’s large electronegativity and/or the substrate’s nearly full d-band [97].

Moreover, it is considered that M–CO bond strength determines the selectivity of CO₂ hydrogenation to either CO or CH₄; weak interaction results in the desorption of CO as a final product, whereas strong interaction facilitates its further hydrogenation to CH₄. In this regard, it has been found via computational studies that the CO adsorption energy can be considered as a selectivity descriptor [6]. In specific, metal atoms with high CO desorption energy (such as Pd, Pt, Ni, and Rh) tend to mainly catalyze the methanation reaction, whereas group 11 metals (Cu, Ag, Au) with lower CO desorption energy produce CO as the main product. These findings are in complete agreement with the present ones (Figure 8), revealing the key role of metal nature in CO₂ activation.

In light of the above discussion, CO₂ activation in conjunction with H₂ dissociation on metal sites could be considered as the main factors determining activity and selectivity. In view of this fact, in a comprehensive work, Sun et al. [58] theoretically explored 3d transition metals in relation to their activity and selectivity in CO₂ hydrogenation by taking into account the adsorption configurations of H₂, CO₂, and the co-adsorption of H atom and CO₂ on metal sites. In Figure 9, adopted from the aforementioned article, the atomic hydrogen adsorption energy (Figure 9a), the barrier energy (Figure 9b), and the CO₂ charge change (Figure 9c) are depicted as functions of 3d-orbital electron transfer during the CO₂ hydrogenation.

It is evident that there is a volcano-type dependence of barrier energy in relation to the number of 3d-orbital electrons, with Ni located at the peak with the lowest energy. This implies the superiority of Ni for CO₂ hydrogenation among the other 3d metals. Moreover, the change in charge from IS (initial state) to TS (transition states) on adsorbed CO₂ also exhibits a volcano-shaped relationship in agreement with the barrier energy, confirming that the barrier energy is closely correlated to electron transfer. Besides, the adsorption energy of H atoms is perfectly correlated with the barrier energy of CO₂ hydrogenation, being again the lowest for Ni [58].

Notably, a perfect agreement between the theoretically predicted trend for CO₂ hydrogenation via the formate pathway (Figure 9) and the present experimental findings over supported catalysts (Figure 8) was revealed for the first time, implying the key role of the energy barrier for CO₂ activation and H₂ dissociation. In other words, the intrinsic characteristics of each metal related to the d-band center and the charge transfer to CO₂ or H₂ molecules can be considered as the main factors, affecting the activation of CO₂ and its consequent hydrogenation to methane or other products.

The above is further supported by the lack of straightforward relationships between the structural and redox properties of TM/CeO₂ samples explored in the present work. In specific, the majority of TM/CeO₂ samples exhibit similar textural and structural properties (Table 1), despite their widely separated catalytic performance. Moreover, both the metal and ceria crystallite sizes are not affected, to a great extent, by metal entities, suggesting similar interfacial interactions. For instance, although Ni, Co, and Cu possess similar surface areas (72–75 m²/g) and crystallite sizes for ceria support (12–14 nm) and metal active phase (10–16 nm), they exhibit completely different performance, in terms of activity and selectivity. Similarly, taking into account the present H₂-TPR results, it can be inferred that the reducibility cannot be considered by itself as the key activity/selectivity descriptor during the CO₂ hydrogenation process. In other words, the differences in the redox behavior, in terms of H₂ uptake and TPR peak temperatures (Table 2), cannot be solely accounted for the dissimilar hydrogenation performance of explored samples.

However, taking into account the XPS results, it is noteworthy that a volcano-shaped trend was revealed among the oxygen vacancy and Ce³⁺ species concentrations and the CO₂ conversion rate and CH₄ selectivity at 350 °C for the optimum rod-shaped reduced catalysts (Fe, Cu, Co, and Ni), following the order: Fe/CeO₂ < Cu/CeO₂ < Co/CeO₂ < Ni/CeO₂ (Figure 10).

These results clearly demonstrate the fundamental role of the metal entity and in turn of metal–support interactions towards determining the relative abundance of oxygen vacancies and partially reduced cerium species, both essential for CO₂ activation. In particular, Lee et al. [98] proposed that Ce³⁺ sites with oxygen vacancies over nickel–ceria catalyst led to the activation of CO₂ towards the formation of carbonate species that are hydrogenated to formate and finally to methane. In a similar manner, nickel–ceria nanorods exhibited higher methanation activity than their nanocubes counterparts in the low-temperature range (200–250 °C) due to their abundance in oxygen vacancies which have a direct effect on the formation of active intermediates [36]. The {110} facets mainly exposed on ceria nanorods have the ability to generate and stabilize the oxygen vacancy sites which are essential in CO₂ to CH₄ conversion [99]. Nevertheless, the pivotal role of metal–support interactions in catalytic activity should be underscored, as the experimental results of the present study clearly reveal the significance of the metal entity and its interaction with the rod-shaped ceria support in generating oxygen vacancies and Ce³⁺ species. In a similar manner, it was demonstrated that oxygen vacancies and moderate basic sites function as contributing factors in CO₂ activation, with the Ni metal having a pivotal role in H₂ activation and dissociation [100].

Collectively, on the basis of the present results, it can be inferred that Ni is the best metal for methanation reaction, due to its specific d-band center and its ability to activate the CO₂ and H₂ molecules. In this regard, any attempt to optimize the CO₂ hydrogenation performance of Ni-based samples should be focused on the selection of appropriate supporting carriers, and promoters, and fine-tuning engineering approaches towards optimizing the ability of Ni sites (via metal–support interactions) to promote the aforementioned process. In this regard, we recently showed that a compromise between the extent of Ni–ceria perimeter and the competitive presence of larger Ni particles is a prerequisite for maximizing the CO₂ hydrogenation to methane [43]. The latter was ascribed, on the ground of a structure sensitivity analysis, to the presence of highly active edge and kink sites instead of inactive terrace sites. Hence, on the basis of the aforementioned discussion, it could be argued that CO₂ activation could be more favorably promoted on under-coordination sites with an “optimum” d-band center and charge transfer ability.

4. Materials and Methods

4.1. Materials Synthesis

The chemicals used in the present work were Ce(NO₃)₃·6H₂O (Acros Organics, Geel, Belgium, purity 99.5%), tetrabutyl titanate (TBOT, Sigma-Aldrich, St. Louis, MO, USA, purity ≥ 97%), NH₄VO₃ (Supelco, Merck, Darmstadt, Germany), Cr(NO₃)₃·9H₂O (Sigma-Aldrich, St. Louis, MO, USA, purity 99%), Mn(NO₃)₂·4H₂O (Sigma-Aldrich, St. Louis, MO, USA, purity ≥ 97.0%), Fe(NO₃)₃·9H₂O (Sigma-Aldrich, St. Louis, MO, USA, purity ≥ 98%), Co(NO₃)₂·6H₂O (Sigma-Aldrich, St. Louis, MO, USA, purity ≥ 98%), Ni(NO₃)₂·6H₂O (Alfa Aesar, ThermoFisher, Kandel, Germany, purity 98%), Cu(NO₃)₂·3H₂O (Sigma-Aldrich, St. Louis, MO, USA, purity 99-104%), Zn(CH₃COO)₂·2H₂O (Sigma-Aldrich, St. Louis, MO, USA, purity ≥ 99%) NaOH (Honeywell Fluka, Seelze, Germany, purity ≥ 98%) and ethanol (Supelco, Merck, Darmstadt, Germany).

Bare ceria nanorods (CeO₂-NR) were hydrothermally synthesized, as thoroughly described in our previous work [101]. Ceria-based transition metal samples (TM/CeO₂, TM: Ti, V, Cr, Mn, Fe, Co, Ni, Cu, and Zn) were synthesized by the wet impregnation method, using aqueous solutions of metal nitrate precursors to obtain a nominal metal/cerium atomic ratio of 0.25. The obtained suspensions were heated under stirring until water evaporation, dried at 90 ºC for 12 h, and finally calcined at 500 °C for 2 h (heating ramp 5 °C/min) [44].

4.2. Materials Characterization

The textural properties of samples were assessed by N₂ adsorption–desorption isotherms at −196 °C (Nova 2200e Quantachrome flow apparatus, Boynton Beach, FL, USA). X-ray diffraction patterns of as-prepared samples were obtained by a Rigaku diffractometer (model RINT 2000, Tokyo, Japan). The crystallite size of the samples was calculated by applying Scherrer’s equation (Equation (4)):

$${\mathrm{D}}_{\mathrm{X}\mathrm{R}\mathrm{D}}\left(\mathrm{n}\mathrm{m}\right)={\displaystyle \frac{\mathrm{{\rm K}}\mathrm{\lambda }}{\mathrm{c}\mathrm{o}\mathrm{s}\mathrm{\theta }}}$$

(4)

where λ is the X-ray wavelength in nm, K is Scherrer’s constant, B is the line broadening and θ is the Bragg angle. By using Scherrer’s equation (Equation (4)), on the most intense diffraction peak, i.e., for CeO₂ (2θ: 28.5°), the primary particle size of this crystal phase was determined. Morphological/elemental analysis was carried out by scanning electron microscopy (SEM, JEOL JSM-6390LV, JEOL Ltd., Akishima, Tokyo, Japan) operating at 20 keV, equipped with an energy dispersive X-ray spectrometry (EDS) system and transmission electron microscopy (TEM) on a JEM-2100 instrument (JEOL, Tokyo, Japan). Temperature Programmed Reduction (H₂-TPR) experiments were carried out in a fully automated AMI-200 Catalyst Characterization Instrument (Altamira Instruments, Pittsburgh, PA, USA) under an H₂ atmosphere, in order to assess the reducibility of the samples. XPS analysis was conducted on Kratos AXIS Ultra HSA, Manchester, United Kingdom with VISION software (http://www.casaxps.com/kratos/, accessed on 10 September 2024) for data acquisition and CASAXPS software (http://www.casaxps.com/, accessed on 10 September 2024) for data analysis.

4.3. Catalytic Evaluation Studies

Catalytic studies were carried out in a quartz fixed-bed U-shaped reactor (i.d. = 1 cm), loaded with 200 mg of catalyst diluted with 200 mg of inert SiO₂. The sample’s pre-treatment included in situ reduction at 400 °C for 1 h under pure H₂ flow (40 cm³ min⁻¹), followed by flushing with He (10 cm³ min⁻¹) until room temperature. The experiments were conducted at 1 bar and in the temperature range of 200–500 °C at intervals of 15–20 °C and a heating rate of 1 °C/min. The total volumetric feed flow was 100 cm³ min⁻¹, corresponding to a weight hourly space velocity (WHSV) of 30 L·g⁻¹·h⁻¹. The gas feed constituted a pure H₂/CO₂ mixture at a molar ratio of 80/20. Calculations for the thermodynamic equilibrium were derived using the mathematical model RGibbs in Aspen Plus software (https://www.aspentech.com/en/products/engineering/aspen-plus, accessed on 10 September 2024. Aspen Technology, Inc., Bedford, MA, USA). The analysis of the reactor outlet mixture was performed in a gas chromatograph (Shimadzu GC-14B, Shimadzu, Kyoto, Japan) with He as the carrier gas, equipped with a thermal conductivity detector (TCD) for the detection of CO and CO₂, a flame ionization detector (FID) for monitoring CH₄ and two separation columns (Molecular Sieve 13X, Restek GmbH, Hesse, Germany and Porapak QS, Restek GmbH, Hesse, Germany).

The reactor effluent was passed through a cold trap submerged in an ice bath in order to condensate the H₂O produced by the reactions. The only carbonaceous products detected in the reactor outlet stream were CH₄ and CO. The conversion of carbon dioxide (Χ_CO2), product selectivities (S_CO and S_CH4), and yields (Y_CO and Y_CH4) were calculated as follows (Equations (5)–(9)), where [i] stands for the concentration of the respective gas at the outlet of the reactor:

$${\mathrm{X}}_{{\mathrm{C}\mathrm{O}}_2}=100\times {\displaystyle \frac{\left[\mathrm{C}\mathrm{O}\right]+{[\mathrm{C}\mathrm{H}}_4]}{{[\mathrm{C}\mathrm{O}}_2]+\left[\mathrm{C}\mathrm{O}\right]+{[\mathrm{C}\mathrm{H}}_4]}}$$

(5)

$${\mathrm{S}}_{\mathrm{C}\mathrm{O}}=100\times {\displaystyle \frac{\left[\mathrm{C}\mathrm{O}\right]}{\left[\mathrm{C}\mathrm{O}\right]+{[\mathrm{C}\mathrm{H}}_4]}}$$

(6)

$${\mathrm{S}}_{{\mathrm{C}\mathrm{H}}_4}=100\times {\displaystyle \frac{{[\mathrm{C}\mathrm{H}}_4]}{\left[\mathrm{C}\mathrm{O}\right]+{[\mathrm{C}\mathrm{H}}_4]}}$$

(7)

$${\mathrm{Y}}_{\mathrm{C}\mathrm{O}}={\mathrm{X}}_{{\mathrm{C}\mathrm{O}}_2}\times {\mathrm{S}}_{\mathrm{C}\mathrm{O}}$$

(8)

$${\mathrm{Y}}_{{\mathrm{C}\mathrm{H}}_4}={\mathrm{X}}_{{\mathrm{C}\mathrm{H}}_4}\times {\mathrm{S}}_{{\mathrm{C}\mathrm{H}}_4}$$

(9)

The macroscopic reaction rates were defined in terms of the rate of moles of CO₂ consumed per mass of the catalyst, r_m;

$${\mathrm{r}}_{\mathrm{m}}\left(\mathrm{m}\mathrm{o}\mathrm{l} {\mathrm{C}\mathrm{O}}_2·{\mathrm{g}}^{-1}·{\mathrm{s}}^{-1}\right)={\displaystyle \frac{{{[\mathrm{C}\mathrm{O}}_2]}_{\mathrm{i}\mathrm{n}}\times {{\mathrm{F}}_{\mathrm{i}\mathrm{n}}\times \mathrm{X}}_{{\mathrm{C}\mathrm{O}}_2}}{100\times 60\times {\mathrm{m}}_{\mathrm{c}\mathrm{a}\mathrm{t}}\times {\mathrm{V}}_{\mathrm{m}}}}$$

(10)

where m_cat is the mass of the catalyst in grams, and V_m is the gas molar volume at 25 °C and 1 bar (24,436 cm³/mol).

5. Conclusions

In the present study, a series of TM/CeO₂ catalysts (atomic ratio M/Ce = 0.25) employing the first-row 3d transition metals (Ti, V, Cr, Mn, Fe, Co, Ni, Cu, and Zn) as active metals were catalytically evaluated in CO₂ hydrogenation. The superiority of the Ni/CeO₂ sample was clearly revealed in terms of CO₂ conversion rate and CH₄ selectivity, followed by Co/CeO₂. The 3d early transition metals (Ti–Fe) and the 3d late metals (Cu and Zn) exhibit much lower activity while being selective to CO. The significance of the metal entity in the CO₂ hydrogenation performance was clearly established via a volcano-type dependence of both activity and selectivity to methane as a function of 3d-orbital electron number, with Ni located at the top of this volcano curve. Moreover, a distinct volcano-shaped trend among the relative abundance of oxygen vacancies and Ce³⁺ species with the CO₂ conversion and CH₄ selectivity was disclosed. The present experimental findings agree perfectly with the theoretically predicted trend for CO₂ hydrogenation over 3d transition metals as a function of the energy barrier for CO₂ activation and H₂ dissociation, implying the key role of intrinsic electronic properties of metal entities in conjunction with metal–support interactions rather than of textural/structural characteristics.