CO2 Hydrogenation over Nanoceria-Supported Transition Metal Catalysts: Role of Ceria Morphology (Nanorods versus Nanocubes) and Active Phase Nature (Co versus Cu)

In this work we report on the combined impact of active phase nature (M: Co or Cu) and ceria nanoparticles support morphology (nanorods (NR) or nanocubes (NC)) on the physicochemical characteristics and CO2 hydrogenation performance of M/CeO2 composites at atmospheric pressure. It was found that CO2 conversion followed the order: Co/CeO2 > Cu/CeO2 > CeO2, independently of the support morphology. Co/CeO2 catalysts demonstrated the highest CO2 conversion (92% at 450 °C), accompanied by 93% CH4 selectivity. On the other hand, Cu/CeO2 samples were very selective for CO production, exhibiting 52% CO2 conversion and 95% CO selectivity at 380 °C. The results obtained in a wide range of H2:CO2 ratios (1–9) and temperatures (200–500 °C) are reaching in both cases the corresponding thermodynamic equilibrium conversions, revealing the superiority of Co- and Cu-based samples in methanation and reverse water-gas shift (rWGS) reactions, respectively. Moreover, samples supported on ceria nanocubes exhibited higher specific activity (µmol CO2·m−2·s−1) compared to samples of rod-like shape, disclosing the significant role of support morphology, besides that of metal nature (Co or Cu). Results are interpreted on the basis of different textural and redox properties of as-prepared samples in conjunction to the different impact of metal entity (Co or Cu) on CO2 hydrogenation process.


Introduction
It is widely accepted amongst the scientific community that the increasing trend of CO 2 emissions in the Earth's atmosphere since the onset of industrialization is the key attributor for the planet temperature rise over the last two centuries [1]. Global temperature is projected to rise by the year 2040 by 1.5 • C in comparison with the pre-industrial levels, according to the latest Intergovernmental Panel on Climate Change (IPCC) report on the impacts on global warming [2,3]. Efforts of mitigation of the aforementioned environmental issue can be simplified into three general approaches: (i) complete performance over a wide range of H 2 :CO 2 ratios (1-9) and temperatures (200-500 • C), in conjunction with the thermodynamic analysis performed under different reaction conditions. The as-prepared samples were synthesized hydrothermally and characterized by N 2 adsorption-desorption, X-ray diffraction (XRD), transmission electron microscopy (TEM) and temperature programmed reduction (TPR) methods and they were catalytically evaluated in the CO 2 hydrogenation reaction at atmospheric pressure. The obtained results are interpreted on the basis of a thermodynamic analysis at different reaction conditions, in conjunction to structural and surface characterization results that reveal the key role of metal entity and support morphology both on CO 2 conversion and products selectivity.

Materials Synthesis
Initially, bare ceria nanoparticles were synthesized by the hydrothermal method, as described in our previous work [68]. Briefly, for the synthesis of bare ceria nanostructures, appropriate amounts of Ce(NO 3 ) 3 ·6H 2 O and NaOH were initially dissolved in double deionized water, then mixed under vigorous stirring for 1 h and aged for 24 h at 90 • C for ceria nanorods and at 180 • C for ceria nanocubes. Eventually, the resulting solids were recovered by centrifugation, washed thoroughly with double deionized water until pH reached a value of 7, in order to remove any co-precipitated salts, and finally washed with ethanol to avoid agglomeration of the nanoparticles. Then, the precipitate was dried at 90 • C for 12 h, followed by calcination at 500 • C for 2 h, under air flow (heating ramp 5 • C/min). The bare ceria samples are designated as CeO 2 -NX (NX: NR (nanorods), NC (nanocubes)).
Cu/CeO 2 -NX and Co/CeO 2 -NX catalysts were synthesized by the wet impregnation method, using aqueous solutions of Cu(NO 3 ) 2 ·2.5H 2 O and Co(NO 3 ) 2 ·6H 2 O, respectively, in order to obtain a metal/cerium atomic ratio of 0.25, corresponding to a Cu loading of ca. 8.5 wt.% and a Co loading of ca. 7.9 wt.%. Then, the obtained suspensions were heated under stirring until water evaporation, dried at 90 • C for 12 h and finally calcined at 500 • C for 2 h under air flow (heating ramp 5 • C/min).

Materials Characterization
The textural properties of the materials were evaluated by the N 2 -adsorption isotherms at −196 • C, using an ASAP 2010 (Micromeritics, Norcross, GA, USA) apparatus (ReQuimTe Analyses Laboratory, Universidade Nova de Lisboa, Portugal). Samples were previously degassed at 300 • C for 6 h. The specific surface area (S BET ) was calculated by the Brunauer-Emmett-Teller (BET) equation. Structural characterization was carried out by means of X-ray diffraction (XRD) in a PAN'alytical X'Pert MPD (PANanalytical, Almelo, Netherlands) equipped with a X'Celerator detector and secondary monochromator (Cu Kα, λ = 0.154 nm, 50 kV, 40 mA; data recorded at a 0.017 • step size, 100 s/step), located at Universidade de Trás-os-Montes e Alto Douro, Vila Real, Portugal. The collected spectra were analyzed by Rietveld refinement using PowderCell software (by Werner Kraus and Gert Nolze, http://www.ccp14.ac.uk), allowing the determination of crystallite sizes by means of the Williamson-Hall plot. The samples were imaged by transmission electron microscopy (TEM). The analyses were performed on a Leo 906E apparatus (Austin, TX, USA), at 120 kV. Samples were prepared by ultrasonic dispersion and a 400 mesh formvar/carbon copper grid (Agar Scientific, Essex, UK) was dipped into the solution for the TEM analysis. The redox properties were assessed by temperature programmed reduction (TPR) experiments in an AMI−200 Catalyst Characterization Instrument (Altamira Instruments, Pittsburgh, PA, USA), employing H 2 as a reducing agent. In a typical H 2 -TPR experiment, 50 mg of the sample were heated up to 1100 • C (10 • C/min), under H 2 flow (1.5 cm 3 ) balanced with He (29 cm 3 ).

Catalytic Evaluation Studies
Catalytic tests for CO 2 hydrogenation were carried out in a quartz fixed-bed U-shaped reactor (internal diameter, ID = 1 cm). The reactor was placed inside a high temperature furnace, equipped with a thermocouple and a Proportional Integral Differential (PID) temperature controller. In each experiment, the catalyst bed consisted of a mixture of 200 mg catalyst diluted with 200 mg of inert SiO 2 . Prior to experiments, catalysts were reduced in situ at 400 • C for 1 h under pure H 2 flow (50 cm 3 /min), followed by flushing with He (10 cm 3 /min). The experiments were conducted at atmospheric pressure and in the temperature range of 200-500 • C at intervals of 20-25 • C. The heating rate was 1 • C/min, adequate for the establishment of steady-state conditions before each measurement. To ensure the reproducibility of the obtained conversion/selectivity values, all tests were conducted twice, without noticeable differences (<5% difference between measurements at the same temperature).
All gases used were of 99.999% purity, provided and certified by Air Liquide Hellas S.A. The total flow rate of the feed gas was 100 cm 3 /min, corresponding to a gas hourly space velocity (GHSV) of 20,000 h −1 . Gas feed comprised of H 2 /CO 2 mixtures at different molar ratios (1)(2)(3)(4)(5)(6)(7)(8)(9). The analysis of gases was performed by a gas chromatograph with He as the carrier gas, equipped with a thermal conductivity detector (TCD) for detection of CO and CO 2 , a flame ionization detector (FID) for monitoring CH 4 and two separation columns (Molecular Sieve 13X and Porapack QS). A cold trap submerged in a water bath was connected to the reactor effluent in order to condensate H 2 O produced by the reactions.
Carbon dioxide conversion, X CO 2 , and product selectivities, S CO and S CH4 , were calculated as follows (Equations (3)-(5)): (4) where [i] in and [i] out represent the concentrations of reactants (i = CO 2 ) or products (i = CO or CH 4 ) at the inlet and outlet of the reactor, respectively. F in and F out are the total flow rates (cm 3 /min) at the inlet and outlet of the reactor, respectively. Detection of other carbonaceous products as well as elemental carbon was either negligible or nonexistent, thus only CO 2 , CH 4 and CO were included in the calculations of carbon balance, which closes within ±6%. Reaction rates were defined in terms of the rate of moles of CO 2 consumed per both mass (r m ) and surface area (r s ) of the catalyst: r s mol CO 2 ·m −2 ·s −1 = r m /S BET (7) where m cat (g) is the mass of the catalyst, S BET (m 2 /g) is the surface area of the catalyst and V m is the gas molar volume at 25 • C and 1 bar (24,436 cm 3 /mol). The thermodynamic equilibrium calculations were elaborated using the minimization of Gibbs free energy mathematical model (RGibbs) in Aspen Plus ® software (Aspen Technology, Inc., Bedford, MA, USA) and the simulation results are included in the catalytic performance plots. The following components were included in the model: CO 2 , H 2 , CO, H 2 O, CH 4 , CH 3 OH, CH 3 OCH 3 , HCOOH, C 2 H 4 and C 2 H 6 . Only the first five substances appeared to be formed in a significant amount in the equilibrium mixture, when all possible reactions between CO 2 and H 2 were considered. Properties of the substances were determined using the Soave-Redlich-Kwong (SRK) equation of state, a widely used method in gas-processing simulation.

Morphological Characterization (TEM)
In order to gain insight into the morphology of bare CeO 2 and M/CeO 2 samples, TEM analysis was performed. The NR samples (Figure 1a-c) display ceria in a uniform rod-like morphology, while the existence of nanocubic morphology is evident in the samples denoted as NC (Figure 1d-f). Apparently, the incorporation of the active metal phase into the CeO 2 lattice has no effect on the support morphology, since distinctive rod and cubic particles are still clearly observed after the incorporation of the metal phase on the well-defined support structure. This also confirms the findings of XRD studies, where the structural features of catalysts remained essentially unaffected after the incorporation of cobalt and copper into the ceria support. Nanomaterials 2019, 9, x FOR PEER REVIEW 6 of 21 the substances were determined using the Soave-Redlich-Kwong (SRK) equation of state, a widely used method in gas-processing simulation.

Morphological Characterization (TEM)
In order to gain insight into the morphology of bare CeO2 and M/CeO2 samples, TEM analysis was performed. The NR samples (Figure 1a-c) display ceria in a uniform rod-like morphology, while the existence of nanocubic morphology is evident in the samples denoted as NC (Figure 1d-f). Apparently, the incorporation of the active metal phase into the CeO2 lattice has no effect on the support morphology, since distinctive rod and cubic particles are still clearly observed after the incorporation of the metal phase on the well-defined support structure. This also confirms the findings of XRD studies, where the structural features of catalysts remained essentially unaffected after the incorporation of cobalt and copper into the ceria support.

Textural and Structural Characterization (Brunauer-Emmett-Teller (BET), X-ray Diffraction (XRD))
The textural, structural and redox properties of bare ceria as well as of M/CeO2 catalysts are presented in Table 1. Bare ceria supports, i.e., CeO2-NR and CeO2-NC, exhibit a BET surface area of 79 and 37 m 2 /g, respectively. The incorporation of transition metals into ceria carriers slightly decreases the BET area. However, the order obtained for bare supports remained unaffected, that is, nanorod samples exhibit higher surface area than the corresponding nanocubic ones, regardless of the nature of the metal phase incorporated into the ceria carrier.

Textural and Structural Characterization (Brunauer-Emmett-Teller (BET), X-ray Diffraction (XRD))
The textural, structural and redox properties of bare ceria as well as of M/CeO 2 catalysts are presented in Table 1. Bare ceria supports, i.e., CeO 2 -NR and CeO 2 -NC, exhibit a BET surface area of 79 and 37 m 2 /g, respectively. The incorporation of transition metals into ceria carriers slightly decreases the BET area. However, the order obtained for bare supports remained unaffected, that is, nanorod samples exhibit higher surface area than the corresponding nanocubic ones, regardless of the nature of the metal phase incorporated into the ceria carrier.   220) and (311) planes mostly prevailing in the samples structure. Moreover, all M/CeO2 samples exhibit smaller peaks indexed to the corresponding oxide due to the incorporation of metal phase (~8 wt.% Co or Cu) into ceria support. Peaks corresponding to CuO crystal phases at 2θ = 35.3°, 38.2° and a less distinguishable peak at 62° are observed for Cu/CeO2 samples, indicating heterodispersion or aggregation of copper species on the surface of ceria [70]. For Cu/CeO2-NC, the double peak at 43-44° as well as the peak at 50° are present due to the sample holder. The Co/CeO2 samples show small peaks characteristic of Co3O4 at 2θ ~ 36°, 44° and 66° [71].  Table 1. Ceria particles are following the order: CeO2-NC (27 nm) > Co/CeO2-NC (24 nm) > Cu/CeO2-NC (19 nm) > CeO2-NR (15 nm) > Co/CeO2-NR (14 nm) > Cu/CeO2-NR (12 nm). These values are close to those reported in literature for ceriasupported materials [52,72]. It is worth mentioning that ceria crystallite size of all samples with nanocubic morphology is higher than that of the corresponding samples with rod-like morphology. Moreover, no significant changes were observed on ceria particle sizes upon the incorporation of Cu or Co, implying that the structural features of ceria remain unaffected by metal addition, as further verified by TEM analysis (see above). Regarding the crystallite size of active metal phases, the CuO size is 52 and 43 nm for Cu/CeO2-NC and Cu/CeO2-NR, respectively. The corresponding values for Co3O4 phase are much lower, i.e., 19 and 16 nm, denoting a better dispersion of cobalt compared to copper on ceria support. Similar conclusions have been previously obtained in a series of CeO2 supported transition metal catalysts (Fe, Co, Ni, Cu) synthesized by the incipient wet impregnation method [73]. It is also worth mentioning that the crystallite size of both active phases (Co3O4 or CuO) and ceria nanoparticles follow the same trend, implying that the structural characteristics of CeO2  [52,72]. It is worth mentioning that ceria crystallite size of all samples with nanocubic morphology is higher than that of the corresponding samples with rod-like morphology. Moreover, no significant changes were observed on ceria particle sizes upon the incorporation of Cu or Co, implying that the structural features of ceria remain unaffected by metal addition, as further verified by TEM analysis (see above). Regarding the crystallite size of active metal phases, the CuO size is 52 and 43 nm for Cu/CeO 2 -NC and Cu/CeO 2 -NR, respectively. The corresponding values for Co 3 O 4 phase are much lower, i.e., 19 and 16 nm, denoting a better dispersion of cobalt compared to copper on ceria support. Similar conclusions have been previously obtained in a series of CeO 2 supported transition metal catalysts (Fe, Co, Ni, Cu) synthesized by the incipient wet impregnation method [73]. It is also worth mentioning that the crystallite size of both active phases (Co 3 O 4 or CuO) and ceria nanoparticles follow the same trend, implying that the structural characteristics of CeO 2 can determine the particle size of metal oxides (Co 3 O 4 or CuO), in agreement with relevant literature studies [74,75]. Moreover, it should be noted that the crystallite size of both ceria and Co 3 O 4 or CuO phases follows the reverse order of BET area, indicating an agglomeration upon the decrease of surface area.

Redox Properties (Hydrogen Temperature-Programmed Reduction (H 2 -TPR))
The redox properties of as-prepared catalysts were evaluated by means of hydrogen temperature-programmed reduction (H 2 -TPR) studies. Figure 3 presents the reduction profiles of all samples. In Table 1 the redox properties, in terms of H 2 consumption (mmol/g) and maximum temperature of main TPR peaks, are summarized. The reduction profiles of bare CeO 2 samples ( Figure 3a) consist of two broad peaks centered at 545-590 • C and 790-810 • C, which are attributed to ceria surface (O s ) and bulk oxygen (O b ) reduction, respectively [76,77]. Both peaks are exhibited in lower temperatures for the nanorod CeO 2 with the effect being more intense for the ceria surface oxygen. Similar behavior is displayed also for the impregnated Co-and Cu-based samples, where both peaks are shifted to substantially lower temperatures, indicating the beneficial role of active phases in the overall redox properties of the as-prepared samples.

Redox Properties (Hydrogen Temperature-Programmed Reduction (H2-TPR))
The redox properties of as-prepared catalysts were evaluated by means of hydrogen temperature-programmed reduction (H2-TPR) studies. Figure 3 presents the reduction profiles of all samples. In Table 1 the redox properties, in terms of H2 consumption (mmol/g) and maximum temperature of main TPR peaks, are summarized. The reduction profiles of bare CeO2 samples ( Figure 3a) consist of two broad peaks centered at 545-590 °C and 790-810 °C, which are attributed to ceria surface (Os) and bulk oxygen (Ob) reduction, respectively [76,77]. Both peaks are exhibited in lower temperatures for the nanorod CeO2 with the effect being more intense for the ceria surface oxygen. Similar behavior is displayed also for the impregnated Co-and Cu-based samples, where both peaks are shifted to substantially lower temperatures, indicating the beneficial role of active phases in the overall redox properties of the as-prepared samples. The corresponding TPR profiles of Co/CeO2 and Cu/CeO2 samples along with those of bare Co3O4 and CuO, are shown in Figure 3b,c, respectively. Interestingly, all M/CeO2 samples exhibit reduction peaks at significantly lower temperatures than bare oxides, which implies the synergy between the metal phase and CeO2 towards an improved reducibility. In particular, the Co/CeO2  The corresponding TPR profiles of Co/CeO 2 and Cu/CeO 2 samples along with those of bare Co 3 O 4 and CuO, are shown in Figure 3b,c, respectively. Interestingly, all M/CeO 2 samples exhibit reduction peaks at significantly lower temperatures than bare oxides, which implies the synergy between the metal phase and CeO 2 towards an improved reducibility. In particular, the Co/CeO 2 samples exhibit two main reduction peaks at 318-335 • C (peak a) and 388-405 • C (peak b), attributed to the subsequent reduction of Co 3 O 4 to CoO and CoO to metallic Co, respectively [78]. As for the Cu/CeO 2 samples, the low-temperature peak (peak a) in the range of 181-194 • C is attributed to the reduction of finely dispersed CuO x species strongly interacting with the ceria surface [68]. The peak at higher temperature (peak b) can be attributed to the formation of larger CuO clusters on the ceria surface [79].
It is obvious that the reduction of M/CeO 2 samples occurs at a significantly lower temperature than those of bare ceria samples, due to the synergy between the two oxide phases that weakens the metal-oxygen bonds [80]. It is also of worth pointing out the facilitation of the ceria surface oxygen reduction in the presence of metal phase (Co or Cu), which results in overlapping bands in the low-temperature range of the TPR profiles of M/CeO 2 samples (Figure 3b,c).
In order to gain further insight into the impact of support morphology, as well as of the nature of the active phase on the redox properties of the as-prepared samples, the H 2 consumption in the low temperature range (100-600 • C), corresponding to the surface oxygen reduction of both active phase and ceria, has been estimated ( Table 1). CeO 2 -NR exhibits higher values (0.59 mmol H 2 ·g −1 ) than CeO 2 -NC (0.41 mmol H 2 ·g −1 ), implying the enhanced reducibility and oxygen mobility of ceria-nanorods. These findings are in complete agreement with the in situ Raman analysis performed in our previous work, in which nanorods exhibited the highest amount in defects and oxygen vacancies [68]. In a similar manner, the M/CeO 2 samples of nanorod morphology exhibit higher H 2 consumption than nanocubes in all cases, indicating their abundance in loosely-bound oxygen species. Regarding the impact of metal entity, the total hydrogen consumption follows, independently on support morphology, the order: Co/CeO 2 > Cu/CeO 2 > CeO 2 , matching the catalytic performance in terms of CO 2 conversion (see below). It should also be noted that hydrogen consumption exceeds the theoretical amount required for the complete reduction of the oxidized transition metal phases for all M/CeO 2 samples ( Table 1). The latter implies the facilitation of ceria capping oxygen reduction in the presence of a metal entity, further corroborating the synergistic effect between the metal and ceria carrier.

CO 2 Hydrogenation Activity
The catalytic performance of the as-prepared samples in the CO 2 hydrogenation reaction was investigated in the temperature range of 200-500 • C. A commercial CeO 2 sample (Fluka, S BET = 15 m 2 /g), denoted as CeO 2 -comm, was also tested as a reference sample. Figure 4 depicts CO 2 conversion of all samples in the temperature range investigated, as compared to the thermodynamic equilibrium CO 2 conversion profiles for methanation and rWGS reactions. In Table 2 the results from catalytic evaluation studies, in terms of both CO 2 conversion and normalized rates at 400 • C, are presented comparatively for all samples. investigated in the temperature range of 200-500 °C. A commercial CeO2 sample (Fluka, SBET = 15 m 2 /g), denoted as CeO2-comm, was also tested as a reference sample. Figure 4 depicts CO2 conversion of all samples in the temperature range investigated, as compared to the thermodynamic equilibrium CO2 conversion profiles for methanation and rWGS reactions. In Table 2 the results from catalytic evaluation studies, in terms of both CO2 conversion and normalized rates at 400 °C, are presented comparatively for all samples.   The beneficial effect of the synthesis method on the catalytic activity is obvious, as all nano-ceria samples exhibit far better CO 2 conversion values than CeO 2 -comm. Moreover, the incorporation of a metal phase into ceria leads to a dramatic improvement of the catalytic activity. Specifically, Co/CeO 2 -NX, Cu/CeO 2 -NX, CeO 2 -NX and CeO 2 -comm catalysts exhibited CO 2 conversion values at 400 • C of approximately 86%, 52%, 20% and 8%, respectively. The observed order of CO 2 conversion generally correlates with the amount of consumed hydrogen calculated by H 2 -TPR in Table 1.
The increase in temperature obviously increases the CO 2 conversion, but to a different extent for each sample. As expected, CO 2 conversion initially increases and reaches a plateau for the highly selective towards methane Co/CeO 2 -NX catalysts, favoring the exothermic CO 2 methanation (Equation (2)) below 500 • C, in agreement with the thermodynamic calculations. A different trend was shown for bare ceria and Cu/CeO 2 -NX samples, as CO 2 conversion increases steadily, but to a lesser extent, as compared to Co/CeO 2 -NX samples, approaching in the case of Cu-based catalysts the equilibrium for the rWGS reaction. Bare ceria carriers, although clearly favoring the rWGS reaction, demonstrated CO 2 conversion values that are well below the corresponding values predicted by thermodynamics.
The impact of the nature of the metal phase (Cu, Co) on the CO 2 hydrogenation performance is further evaluated on the basis of the selectivity towards CH 4 and CO, depicted in Figure 5. Apparently, commercial ceria, nano-ceria and Cu/CeO 2 -NX samples are all highly selective to CO (>90%). Thus, maximum CO 2 conversion values for these samples are expected to be closer to those for the rWGS reaction equilibrium, a much less favorable reaction than CO 2 methanation, as indeed shown by the corresponding equilibrium curve (dotted line) in Figure 4. Similar results for CO production over CeO 2 [59,81] and Cu-based catalysts are reported in literature (see Table 3). Interestingly, Cu/CeO 2 -NR reaches equilibrium conversion values at ca. 380 • C, well below that reported for many rWGS catalysts. On the contrary, the addition of cobalt into CeO 2 leads to a completely different trend, since the selectivity towards CH 4 for all Co/CeO 2 samples is approximately 95% at temperatures higher than ca. 400 • C. However, at temperatures below 400 • C, selectivity towards CO is significant, decreasing rapidly for higher temperatures.
Interestingly, Cu/CeO2-NR reaches equilibrium conversion values at ca. 380 °C, well below that reported for many rWGS catalysts. On the contrary, the addition of cobalt into CeO2 leads to a completely different trend, since the selectivity towards CH4 for all Co/CeO2 samples is approximately 95% at temperatures higher than ca. 400 °C. However, at temperatures below 400 °C, selectivity towards CO is significant, decreasing rapidly for higher temperatures.  In light of the above results, it should be highlighted that the as-prepared cobalt catalysts are highly efficient in methane production at temperatures as low as 400 • C, under the reaction conditions employed, being superior, or at least comparable, to the most active Co-based catalysts reported in literature (see Table 3). On the other hand, the as-prepared copper catalysts are highly selective to CO, reaching rWGS equilibrium values at ca. 380 • C. In a similar manner, Dai et al. [82] reported 100% selectivity to CO for Cu/CeO 2 , attributed to the presence of Cu 0 active sites, whereas selectivity to CH 4 for Co/CeO 2 increased markedly between 350 • C and 400 • C. Table 3. Comparison of the CO 2 reduction performance of Cu-and Co-based catalysts for the rWGS and CO 2 methanation reactions, respectively, at atmospheric pressure.

Sample
T ( • C) % X CO2 % S CO % S CH4 H 2 :CO 2 %wt. Cu or Co Ref. The obtained differences in activity/selectivity between Cu-and Co-based catalysts can be corroborated by taking into account the underlying mechanism of Cu-and Co-catalyzed CO 2 hydrogenation reaction, in conjunction with the present characterization results. In particular, the dissociation of adsorbed CO, considered as the rate-determining step of the CO 2 methanation process [92,93], has been proposed to proceed by two main pathways: direct CO ads dissociation and H-assisted CO ads dissociation. The first mechanism is proposed to occur over group VIII metal-based catalysts, such as Co [94,95]. In view of this fact, Liu et al. [48] demonstrated that Co showed more favorable thermodynamics and lower CO 2 decomposition barriers for CO 2 reduction compared to Cu. In a similar manner, a close relationship between the CO 2 and H 2 adsorption capacity of Co/KIT-6 catalysts and their CO 2 conversion/selectivity performance has been revealed [87]. In particular, high H 2 adsorption capacity can provide a large number of active H species for the further hydrogenation of intermediate species (such as HCOO − ) to methane, whereas the low H 2 adsorption and activation capacity is favorable to CO formation. Considering the fact that as-prepared Co/CeO 2 -NX possess a higher reducibility than Cu/CeO 2 -NX in terms of H 2 uptake (Table 1), this could be an explanation for the high methane selectivity exhibited for the Co-based samples.

Cu-Catalyzed rWGS Reaction
By contrast, a redox mechanism has been proposed in the literature for rWGS reaction over Cu-based catalysts [29,96]. Specifically, Cu 0 atoms can act as active sites for the dissociation of CO 2 and the Cu 2 O formed is subsequently reduced by hydrogen to regenerate metallic Cu species. Hydrogen was proposed to be only a reducing agent without direct participation in the formation of intermediate species in the rWGS reaction [97]. The scheme can be simplified to the following reactions (Equations (8) and (9)): The facile reduction of Cu/CeO 2 -NX catalysts towards reduced copper species at temperatures lower than ca. 300 • C, corroborated by the H 2 -TPR results (Figure 3), can possibly favor this redox mechanism, thus being the reason for the high selectivity towards CO even at low temperatures. However, hydrogenation of carbon dioxide proceeds through a rather complex pathway, thus more detailed work has to be undertaken in order to elucidate the differences in the mechanisms of both CO 2 methanation and rWGS and in the product distribution depending on the metal active phase of the catalyst. To this end, mechanistic studies are currently under development in our laboratory.
Regarding the impact of support morphology, it can be concluded that the CO 2 conversion and selectivity of the samples are not significantly affected by the support morphology (NR vs. NC), as shown in Figures 4 and 5. In particular, a slightly better conversion performance is obtained for Cu/CeO 2 -NR samples as compared to Cu/CeO 2 -NC, whereas for bare CeO 2 and Co/CeO 2 minor differences in conversion performance between NC and NR were observed. This can also be demonstrated by the mass-normalized reactions rates, as values for NRs are very close to those for NCs (Table 2).
It should be mentioned, however, that NC samples exhibit two to three times lower BET surface area as compared to NR samples (Table 1), which should be further accounted for the different catalytic activity of NC vs. NR samples. In order to more precisely gain insight into the impact of support morphology on the intrinsic reactivity, the specific activity normalized per unit of surface area (µmol CO 2 ·m −2 ·s −1 ) was thus calculated (Equation (7)) and is summarized in Table 2. It is evident that the area-normalized reaction rates of nanocubic samples are considerably higher compared to their nanorod counterparts. Specifically, Co/CeO 2 -NC exhibits the highest rate (1.07 µmol CO 2 ·m −2 ·s −1 ), followed by Cu/CeO 2 -NC (0.51 µmol CO 2 ·m −2 ·s −1 ) and CeO 2 -NC (0.18 µmol CO 2 ·m −2 ·s −1 ). Samples with nanorod morphology exhibit almost half of the reaction rates shown by the nanocubic samples, with values of 0.40, 0.25 and 0.09 µmol CO 2 ·m −2 ·s −1 for Co/CeO 2 -NR, Cu/CeO 2 -NR and CeO 2 -NR, respectively.
In a similar manner, it was found that Ru catalysts supported on ceria nanocubes showed higher CO 2 methanation activity, in comparison with ceria nanorods and nanopolyhedra, on the basis of specific rate [98]. Kovacevic et al. [81] synthesized CeO 2 nanoparticles with distinct morphology for the rWGS reaction and reported that nanocubes were more active than nanorods in terms of CO produced per surface area, owing to the greater inherent reactivity of (100) crystal planes enclosing cubes, in contrast with the inherently less reactive (111) facets exposed by rods and particles.

Effect of H 2 :CO 2 Ratio
The effect of the H 2 :CO 2 feed ratio (9:1, 6:1, 4:1 and 1:1) on the conversion and selectivity performance was next explored over Co/CeO 2 -NR and Cu/CeO 2 -NR samples. Reactant feed ratio is an important factor in the CO 2 hydrogenation process, as it modifies the thermodynamic equilibrium of the system [99] and could increase selectivity of a specific product [10,100]. Moreover, knowledge on a catalyst performance under different reaction mixtures is important, especially with the scope of employing the catalytic system in an industrial-scale process, where hydrogen flow may vary, due to the inherent fluctuations associated with the available excess RES power [101].
The CO 2 conversion along with the selectivities to CH 4 and CO, as a function of temperature, for the different H 2 /CO 2 feed molar ratios investigated, are depicted in Figure 6. The corresponding equilibrium curves for each reaction system are also shown for comparison. It is evident that the increase of H 2 :CO 2 ratio exerted a beneficial effect on the CO 2 conversion values, according to thermodynamics.
observed for CO selectivity. Traces of C2 hydrocarbons were also observed for Co/CeO2-NR, especially under H2-limiting conditions. In general, significant CO2 conversion to CH4 can be achieved on Co/CeO2-NR, even when hydrogen availability decreases, as long as the system is operating far from sub-stoichiometry.
For Cu/CeO2-NR, CO2 conversion at 450 °C increased from 24.2% to 58.1% as the feed ratio increased from 1:1 to 9:1, respectively, in agreement with the equilibrium CO2 conversion to CO. Considering that equimolar amounts of CO2 and H2 are required for the stoichiometric rWGS reaction to take place, any excess amount of hydrogen in the feed is expected to facilitate CO production. Regardless of the hydrogen inlet concentration, CO2 conversion approached the respective equilibrium values within ~5% at around 380 °C. Several catalytic systems employed for rWGS have been reported to function under thermodynamic limitation at the medium-high or high temperature regime [103][104][105]. As for the selectivity to CH4, it increased slightly, from about 1.0% for a H2:CO2 ratio of 1:1 to 5.1% for a ratio of 9:1. Correspondingly, CO selectivity decreased slightly with an increase in H2:CO2 ratio, with selectivity values at 500 °C equal to 95.3%, 97.1%, 98.2% and 99.5%, at H2:CO2 ratios of 9, 6, 4 and 1, respectively. Contrary to Co/CeO2-NR, no other carbonaceous products were identified in the case of Cu/CeO2, at the reaction conditions examined. Low methane selectivity values exhibited by Cu/CeO2-NR, even under excess hydrogen in the reactant stream, further substantiate the possibility of a redox mechanism over this catalyst. Excess hydrogen may directly re-reduce Cu2O species formed by CO2 dissociation over metallic Cu, rather than fully hydrogenating intermediate species, thus leading to high CO production, as discussed above.

Stability Tests
Further examination of the catalytic performance of Co/CeO2-NR and Cu/CeO2-NR was conducted in short-term stability experiments under isothermal conditions, in order to assess the catalysts lifetime characteristics and to detect possible deactivation phenomena. Herein, the aforementioned samples were further tested for 12 h at a constant temperature of 450 °C, with a feed ratio of H2:CO2 equal to 9:1 and a GHSV of 20,000 h −1 . The results of CO2 conversion and selectivity values as a function of time are presented in Figure 7.
Copper-based catalysts are generally known for their inferior catalytic performance under hightemperature operation, due to the sintering of copper particles resulting in a decrease in active sites [106,107]. However, CO2 conversion on Cu/CeO2-NR remained relatively stable for the duration of the experiment, fluctuating between an average value of 52%, decreasing very slightly from the value of 56% found for the fresh catalyst during the corresponding light-off tests. It is also noted that the thermodynamic value for the rWGS reaction, under the employed conditions, is 63%. Moreover, the selectivity towards CO stabilized to ca. 96% after approximately 1 h. Thus, the Cu/CeO2-NR remained highly active towards CO production and exhibited very low methane yield even after 12 h at 450 °C, under hydrogen excess conditions, further corroborating the fact that CO2 methanation is largely In particular, for Co/CeO 2 -NR catalysts, CO 2 conversion values at 450 • C of 23, 61, 81, and 90% at a H 2 :CO 2 ratio of 1, 4, 6 and 9, were obtained, respectively. When equal amounts of CO 2 and H 2 were fed into the reactor, CO 2 conversion first increased up to 14% at 360 • C, decreased until 400 • C and then re-increased to a value of 26% at 500 • C. This behavior is expected thermodynamically, as sub-stoichiometric conditions for CO 2 methanation were employed and rWGS was thus favored, allowing the conversion to exceed the theoretical quasi-limit calculated by including only CH 4 as a product (green dotted line in Figure 6a). Similar results were found on thermodynamic analyses for CO 2 methanation in literature [36,100,102]. It is worth mentioning that Co/CeO 2 -NR almost reaches maximum CO 2 conversion values at ca. 400 • C for over-stoichiometric conditions and ca. 420 • C in the case of stoichiometric ratio. In addition, CO 2 conversion approached values as high as 90% of the thermodynamic equilibrium ones, implying the superiority of Co/CeO 2 catalysts for the Sabatier reaction.
Higher H 2 :CO 2 feed ratios are likely to promote methane formation instead of the reverse WGS reaction, since Equation (2) is more dependent on hydrogen. Indeed, at 450 • C, selectivity values towards CH 4 equal to 97.5%, 93.6%, 88.2%, and 38.1% were observed for a feed ratio of 9:1, 6:1, 4:1 and 1:1, respectively (Figure 6b). CH 4 selectivity decreased significantly at temperatures higher than 380 • C when the H 2 :CO 2 ratio was unity, reaching a value of 26% at 500 • C. The opposite trend was observed for CO selectivity. Traces of C 2 hydrocarbons were also observed for Co/CeO 2 -NR, especially under H 2 -limiting conditions. In general, significant CO 2 conversion to CH 4 can be achieved on Co/CeO 2 -NR, even when hydrogen availability decreases, as long as the system is operating far from sub-stoichiometry.
For Cu/CeO 2 -NR, CO 2 conversion at 450 • C increased from 24.2% to 58.1% as the feed ratio increased from 1:1 to 9:1, respectively, in agreement with the equilibrium CO 2 conversion to CO. Considering that equimolar amounts of CO 2 and H 2 are required for the stoichiometric rWGS reaction to take place, any excess amount of hydrogen in the feed is expected to facilitate CO production. Regardless of the hydrogen inlet concentration, CO 2 conversion approached the respective equilibrium values within~5% at around 380 • C. Several catalytic systems employed for rWGS have been reported to function under thermodynamic limitation at the medium-high or high temperature regime [103][104][105]. As for the selectivity to CH 4 , it increased slightly, from about 1.0% for a H 2 :CO 2 ratio of 1:1 to 5.1% for a ratio of 9:1. Correspondingly, CO selectivity decreased slightly with an increase in H 2 :CO 2 ratio, with selectivity values at 500 • C equal to 95.3%, 97.1%, 98.2% and 99.5%, at H 2 :CO 2 ratios of 9, 6, 4 and 1, respectively. Contrary to Co/CeO 2 -NR, no other carbonaceous products were identified in the case of Cu/CeO 2 , at the reaction conditions examined. Low methane selectivity values exhibited by Cu/CeO 2 -NR, even under excess hydrogen in the reactant stream, further substantiate the possibility of a redox mechanism over this catalyst. Excess hydrogen may directly re-reduce Cu 2 O species formed by CO 2 dissociation over metallic Cu, rather than fully hydrogenating intermediate species, thus leading to high CO production, as discussed above.

Stability Tests
Further examination of the catalytic performance of Co/CeO 2 -NR and Cu/CeO 2 -NR was conducted in short-term stability experiments under isothermal conditions, in order to assess the catalysts lifetime characteristics and to detect possible deactivation phenomena. Herein, the aforementioned samples were further tested for 12 h at a constant temperature of 450 • C, with a feed ratio of H 2 :CO 2 equal to 9:1 and a GHSV of 20,000 h −1 . The results of CO 2 conversion and selectivity values as a function of time are presented in Figure 7. Figure 6. Effect of H2:CO2 ratio on theoretical and experimental CO2 conversion and selectivity to CH4 (triangle) and CO (square) values for (a,b) Co/CeO2-NR and for (c,d) Cu/CeO2-NR. Dotted and dashed lines represent equilibrium CO2 conversion values for the methanation and rWGS reactions, respectively.

Stability Tests
Further examination of the catalytic performance of Co/CeO2-NR and Cu/CeO2-NR was conducted in short-term stability experiments under isothermal conditions, in order to assess the catalysts lifetime characteristics and to detect possible deactivation phenomena. Herein, the aforementioned samples were further tested for 12 h at a constant temperature of 450 °C, with a feed ratio of H2:CO2 equal to 9:1 and a GHSV of 20,000 h −1 . The results of CO2 conversion and selectivity values as a function of time are presented in Figure 7.
Copper-based catalysts are generally known for their inferior catalytic performance under hightemperature operation, due to the sintering of copper particles resulting in a decrease in active sites [106,107]. However, CO2 conversion on Cu/CeO2-NR remained relatively stable for the duration of the experiment, fluctuating between an average value of 52%, decreasing very slightly from the value of 56% found for the fresh catalyst during the corresponding light-off tests. It is also noted that the thermodynamic value for the rWGS reaction, under the employed conditions, is 63%. Moreover, the selectivity towards CO stabilized to ca. 96% after approximately 1 h. Thus, the Cu/CeO2-NR remained highly active towards CO production and exhibited very low methane yield even after 12 h at 450 °C, under hydrogen excess conditions, further corroborating the fact that CO2 methanation is largely unfavorable over this catalyst.
In the course of the methanation reaction, the stability of a catalyst is closely linked to coking and metal sintering [108,109]. For Co/CeO2-NR, it was demonstrated that CO2 conversion remained very stable at ~90%, whereas the thermodynamic value is calculated at 99.5% (Figure 7). At the same time, the selectivity towards CH4 remained stable at 96% for the whole duration of experiment, revealing the superiority of Co/CeO2-NR sample in CO2 methanation process, in terms of conversion, selectivity and stability.  Copper-based catalysts are generally known for their inferior catalytic performance under high-temperature operation, due to the sintering of copper particles resulting in a decrease in active sites [106,107]. However, CO 2 conversion on Cu/CeO 2 -NR remained relatively stable for the duration of the experiment, fluctuating between an average value of 52%, decreasing very slightly from the value of 56% found for the fresh catalyst during the corresponding light-off tests. It is also noted that the thermodynamic value for the rWGS reaction, under the employed conditions, is 63%. Moreover, the selectivity towards CO stabilized to ca. 96% after approximately 1 h. Thus, the Cu/CeO 2 -NR remained highly active towards CO production and exhibited very low methane yield even after 12 h at 450 • C, under hydrogen excess conditions, further corroborating the fact that CO 2 methanation is largely unfavorable over this catalyst.
In the course of the methanation reaction, the stability of a catalyst is closely linked to coking and metal sintering [108,109]. For Co/CeO 2 -NR, it was demonstrated that CO 2 conversion remained very stable at~90%, whereas the thermodynamic value is calculated at 99.5% (Figure 7). At the same time, the selectivity towards CH 4 remained stable at 96% for the whole duration of experiment, revealing the superiority of Co/CeO 2 -NR sample in CO 2 methanation process, in terms of conversion, selectivity and stability.

Conclusions
The present results revealed the strong effect of metal phase (Cu or Cu) and reaction conditions on the CO 2 hydrogenation performance of nanoceria-based M/CeO 2 catalysts of different morphology (nanorods or nanocubes). It was shown that hydrothermally synthesized nano-ceria carriers exhibited better catalytic activity in the hydrogenation of CO 2 than commercial ceria. More importantly, incorporating Co and Cu into the nanostructured ceria support led to a significant increase in the catalytic activity, with CO 2 conversion following the order Co/CeO 2 > Cu/CeO 2 > CeO 2 . This order correlates with the hydrogen consumption estimated by H 2 -TPR, implying that the enhanced reducibility-linked to the different metal phase-probably favors the CO 2 hydrogenation reactivity. The main product of the reaction was determined by the nature of the metal entity incorporated into cerium oxide, with Co/CeO 2 exhibiting high selectivity towards CH 4 and CO 2 conversion values close to equilibrium. Cu/CeO 2 exhibited remarkable CO selectivity even under hydrogen excess conditions and very close to equilibrium CO 2 conversion values for the rWGS reaction. From a practical point of view, Co/CeO 2 catalysts demonstrated an excellent conversion and selectivity performance, offering~85% yield to methane at 450 • C. On the other hand, Cu/CeO 2 samples were very selective for CO production, exhibiting 52% CO 2 conversion and 95% CO selectivity at ca. 400 • C.
In both cases a stable conversion and selectivity performance (either for CO 2 methanation or rWGS reaction) was attained in short-term (12 h) stability tests. Regarding the impact of ceria morphology, the samples supported on ceria nanocubes exhibited higher specific activity (µmol CO 2 ·m −2 ·s −1 ), as compared to samples of rod-like shape, revealing the significant role of support morphology, besides that of metal nature (Co or Cu). These results are considered to be very promising in the sense of employing these catalysts in real-scale processes with variable H 2 :CO 2 ratios, where renewable hydrogen from excess RES power can be used to efficiently and selectively convert CO 2 to CO or CH 4 .