CO2 Methanation on Supported Rh Nanoparticles: The combined Effect of Support Oxygen Storage Capacity and Rh Particle Size

CO2 hydrogenation toward methane, a reaction of high environmental and sustainable energy importance, was investigated at 200–600 ◦C and H2/CO2 = 4/1, over Rh nanoparticles dispersed on supports with different oxygen storage capacity characteristics (γ-Al2O3, alumina-ceria-zirconia, and ceria-zirconia). The effects of the support OSC and Rh particle size on reaction behavior under both integral and differential conditions were investigated, to elucidate the combined role of these crucial catalyst design parameters on methanation efficiency. A volcano-type variation of methanation turnover frequency was found in respect to support OSC; Rh/ACZ, with intermediate OSC, was the optimal catalyst. The structure sensitivity of the reaction was found to be a combined function of support OSC and Rh particle size: For Rh/γ-Al2O3 (lack of OSC) methanation was strongly favored on small particles—the opposite for Rh/CZ (high OSC). The findings are promising for rational design and optimization of CO2 methanation catalysts by tailoring the aforementioned characteristics.


Introduction
Fossil fuels (coal, oil, and natural gas) remain the dominant source of energy in the industrial, commercial, residential, and transportation sectors. Besides the fact that fossil fuels are finite, fossil fuel utilization leads to the emission of tremendous quantities of CO 2 in the atmosphere: These have consequently risen from~280 ppm before the industrial revolution to~410 ppm nowadays. Moreover, it is predicted that the concentration of CO 2 may reach~570 ppm by the end of the century, if urgent mitigating actions are not implemented. As is well understood, CO 2 is the dominant greenhouse gas and has the most significant contribution to the greenhouse effect, as well as in the concomitant global warming and climate change. Therefore, control of CO 2 emissions is a critical and urgent environmental issue [1][2][3][4][5][6][7][8][9]. A possible solution can be provided by energy models with a reduced environmental footprint that can also be combined with the so-called cyclic economy strategies [4,[9][10][11][12][13][14]. Apart from Rh dispersed on the support with moderate OSC value (Rh/ACZ) was superior in CO 2 methanation. Moreover, a strong dependence of CO 2 methanation on Rh particle size was observed, with a very interesting feature of this structure sensitivity: For Rh particles dispersed on a non-reducible support (i.e., Rh/γ-Al 2 O 3 ), the lower the particle size, the more superior the methanation performance, in full contrast with the feature for rhodium particles dispersed on high OSC support (i.e., Rh/CZ), where the higher the particle size, the more superior the methanation performance. These observations open up new possibilities on improving the methanation efficiency of supported metal catalysts by simultaneous adjusting their OSC and particle size characteristics. The results are adequately rationalized in terms of the (O δ− , δ + ) effective-double-layer account of promotion and metal-support interactions. Table 1 displays the main morphological and physicochemical properties of the support and catalysts studied herein. The BET surface area values (S BET ) of the oxide supports are 178, 149, and 22 m 2 /g for the γ-Al 2 O 3 , ACZ, and CZ, respectively. After Rh deposition, the obtained counterpart catalysts show a slight decreased in S BET values (160, 136, and 17 m 2 /g, respectively), probably due to some pore blocking of the mesoporous oxide supports [65]. Minor losses on the BET surface area can also be observed in the catalysts subjected to further treatment at high temperatures, 750 and 850 • C, under oxidative conditions (Table 1). Rhodium metal loadings measured by ICP-OES was found to be close to the nominal 1 wt % Rh loading (Table 1) [53]. The total oxygen storage capacities (in µmol O 2 /g) of the supports and Rh catalyst counterparts, assessed by H 2 -TPR as the half of the total amount of consumed H 2 in the 25-850 • C temperature interval of the H 2 consumption profiles, are shown in Figure 1. The oxide supports γ-Al 2 O 3 , ACZ, and CZ used encompass a wide range of OSC values, i.e., 0, 101, and 557 µmol O 2 /g, respectively (Table 1), reflecting the non-reducibility of γ-Al 2 O 3 , and the Ce 4+ → Ce 3+ reduction of ACZ and CZ supports appearing in the TPR spectra ( Figure 1a) as two overlapping broad peaks at ca. 300-550 • C and 500-850 • C, a feature characteristic of ceria-containing samples [68,69]. Finally, the moderate reduction on the OSC values found for the thermally aged Rh/CZ catalysts (from 589 on fresh → 476 on treated at 750 °C → 327 μmol O2/g on treated at 850 °C; Table 1 and Figure  1d) was expected, since it is well-known in the literature for CeO2-ZrO2 systems [71]. However, as will be discussed below, the remaining OSC ability of the catalysts is still high enough to induce substantial modifications on their CO2 methanation performance.

Morphological and Reducibility Characteristics of the Materials
Rh nanoparticle sizes measured by both HRTEM (Supplementary Materials Figure S1) and H2 isothermal chemisorption, the arithmetic mean value ( ̅ Rh) obtained by the two independent methods for each catalyst (Supplementary Materials Section S1), are shown in Table 1. These take the values of 1.2, 1.7, and 5.0 nm on the fresh Rh/γ-Αl2O3, Rh/ACZ, and Rh/CZ catalysts, respectively. It is highlighted that Rh nanoparticles dispersed on zero OSC support (γ-Αl2O3) underwent agglomeration after high-temperature oxidative treatment under the sintering protocols #1 and #2: from 1.2 nm on Rh/γ-Al2O3 (fresh) → 1.6 nm on Rh/γ-Al2O3 (treated at 750 °C) → 2.1 nm on Rh/γ-Al2O3 (treated at 850 °C). In contrast, Rh nanoparticles on the high OSC support (CZ) underwent significant redispersion: from 5.0 nm on Rh/CZ (fresh) → 2.3 nm on Rh/CZ (treated at 750 °C) → 2.1 nm on Rh/CZ (treated at 850 °C). These findings are strikingly consistent with the H2-TPR behavior of the corresponding samples ( Figure 1). Incorporation of~1 wt % Rh on the supports increases the OSC of the resulted catalysts to 69, 146, and 589 µmol O 2 /g of the fresh Rh/γ-Al 2 O 3 , Rh/ACZ, and Rh/CZ catalysts (Table 1 and Figure 1a). The corresponding 69, 45, and 32 µmol O 2 /g OSC increments found reflect the contribution of Rh 3+ and Rh 1+ oxide reduction to Rh • and are close to the theoretically calculated amounts of O 2 for these transformations (i.e., 72 µmol O 2 /g for Rh 3+ → Rh • and 29 µmol O 2 /g for Rh 1+ → Rh • ), bearing in mind that Rh particles anchored on γ-Al 2 O 3 can be fully oxidized (Rh 3+ ) in contrast to those dispersed on high OSC supports, which undergo only partial oxidation due to strong interactions between Rh particles and the support that destabilize rhodium oxide phases, as demonstrated by recent XPS studies [53].
It should be stressed, however, that incorporation of Rh on the supports resulted in a significant promotion of the reducibility of ACZ and CZ, as evidenced by the substantial shifts of the TPR peaks, attributed to the reduction of ceria at much lower temperatures (from ca. 300 to 850 • C for the supports → ca. 50-550 • C for the counterpart catalysts (Figure 1a,c,d and Table 1), indicating the strong promotion of H 2 spillover [68,69], which, in the absence of metal, is limited by H 2 dissociation. These shifts are advantageous for the catalytic system under investigation, as the reducibility (i.e., the lattice oxygen lability) of the resulted Rh catalysts is well matched to the favorable temperature range of the CO 2 methanation reaction ( Figure 1). Some additional features of the H 2 -TPR profiles are of particular interest. In the case of fresh Rh/γ-Al 2 O 3 catalyst, the reduction of Rh species appeared as a LT peak (at~150 • C) and a HT broad one (at~500 • C) (Figure 1b and Table 1); the former is attributed to the reduction of Rh 2 O 3 species anchored on the Al 2 O 3 surface, while the latter to the reduction of Rh oxide diffused into alumina and thus in strong interaction with it (RhAl x O y ) [70]. The interesting point is that the TPR peak of Rh 2 O 3 surface species of the samples that have aged at oxidative thermal conditions, Rh/γ-Al 2 O 3 -treated at 750 • C, and Rh/γ-Al 2 O 3 -treated at 850 • C, is shifted to higher temperatures, from 150 to 175 and 215 • C, respectively ( Figure 1b and Table 1), implying more difficult-to-reduce Rh particles-most probably larger particles. As discussed below, this is in agreement with the Rh particle agglomeration found after oxidative thermal aging of Rh particles anchored on γ-Al 2 O 3 support, which is lacking in oxygen storage capacity. In striking contrast, TPR peaks on fresh Rh/CZ catalyst attributed to Rh oxide species (at~105 • C), to superficial reduction of ceria taking place at CZ surfaces (at~175 • C), and to the reduction of bulk ceria (at~380 • C) are all shifted to lower temperatures over the thermally aged Rh/CZ catalysts ( Figure 1d and Table 1). This implies an enhanced promotion of the reducibility of the aged (Rh/CZ-treated at 750 • C and Rh/CZ-treated at 850 • C), which is rationally understood by considering an enhanced metal-support interaction between Rh particles and CZ support resulting from a larger amount of undercoordinated rhodium sites, implying lower-size Rh particles (i.e., Rh redispersion during oxidative thermal aging). The latter is fully consistent with our results for the mean Rh particle sizes obtained by means of TEM and H 2 -chemisorption experiments, included in Table 1 and discussed below.
Finally, the moderate reduction on the OSC values found for the thermally aged Rh/CZ catalysts (from 589 on fresh → 476 on treated at 750 • C → 327 µmol O 2 /g on treated at 850 • C; Table 1 and Figure 1d) was expected, since it is well-known in the literature for CeO 2 -ZrO 2 systems [71]. However, as will be discussed below, the remaining OSC ability of the catalysts is still high enough to induce substantial modifications on their CO 2 methanation performance.
Rh nanoparticle sizes measured by both HRTEM (Supplementary Materials Figure S1) and H 2 isothermal chemisorption, the arithmetic mean value (d Rh ) obtained by the two independent methods for each catalyst (Supplementary Materials Section S1), are shown in Table 1. These take the values of 1.2, 1.7, and 5.0 nm on the fresh Rh/γ-Al 2 O 3 , Rh/ACZ, and Rh/CZ catalysts, respectively. It is highlighted that Rh nanoparticles dispersed on zero OSC support (γ-Al 2 O 3 ) underwent agglomeration after high-temperature oxidative treatment under the sintering protocols #1 and #2: from 1.2 nm on Rh/γ-Al 2 O 3 (fresh) → 1.6 nm on Rh/γ-Al 2 O 3 (treated at 750 • C) → 2.1 nm on Rh/γ-Al 2 O 3 (treated at 850 • C). In contrast, Rh nanoparticles on the high OSC support (CZ) underwent significant redispersion: from 5.0 nm on Rh/CZ (fresh) → 2.3 nm on Rh/CZ (treated at 750 • C) → 2.1 nm on Rh/CZ (treated at 850 • C). These findings are strikingly consistent with the H 2 -TPR behavior of the corresponding samples ( Figure 1).
The resistance to sintering during oxidative thermal treatment-on both mechanisms of particles agglomeration, i.e., large particle migration and coalescence (PMC) and Ostwald ripening (OR)-or even redispersion of metal particles dispersed on supports characterized by moderate and high values of labile lattice oxygen capacity has been thoroughly interpreted in Reference [65]; only a brief description is included herein. The creation of an O δ− electric layer via the spontaneous (thermally driven) O 2− backspillover from the support to the surface of metal particles quenches the PMC mechanism due to the resulting interparticle electrostatic repulsion (anti-PMC). At the same time, metal atoms possibly detached from large metal crystallites are efficiently trapped by surface oxygen vacancies in the support material suppressing surface diffusion of these atomic species on the support and their subsequent attachment to larger particles (anti-OR). The latter can certainly lead to redispersion, the extent of which depends on several factors, including the metal identity, the sintering conditions, and length of time imposed, and mainly the population of the surface oxygen vacancies of the support [65]. Metal nanoparticles dispersed on supports with a lack of labile lattice oxygen (e.g., γ-Al 2 O 3 ) where such phenomena are absent provide little or no resistance to sintering under high-temperature oxidative treatment, as indeed found (Table 1).  (Table 1) so that to have different supporting material, but as close as possible to the mean Rh particle size. The latter was adopted to allow us to investigate and understand the effect of the support OSC isolated from that of particle size. For the sake of comparison, the equilibrium CO 2 conversion, CH 4 and CO yields, and selectivity profiles, predicted by the thermodynamics of the methanation reaction (R1) by using the Outokumpu HSC Chemistry ® program [46], are also included in the figure. The feed composition was kept constant at the stoichiometry of the CO 2 methanation reaction (R1), i.e., 5% CO 2 /20% H 2 balance Ar at 1 bar, as well as the effective mean contact time of the reactants, with rhodium active sites at τ(CO 2 ) = 1.26 s (and τ(H 2 ) = 5.04 s). Due to the different concentration of Rh active sites per mass of catalyst of the Rh(1.6nm)/γ-Al 2 O 3 , Rh(1.7nm)/ACZ, and Rh(2.1nm)/CZ samples, the latter was achieved by adjusting the total flow rates at 78, 60, and 47 N mL/min, respectively.  According to thermodynamics, CH4 yield is restricted at elevated temperatures (Figure 2b), while the Gibbs free-energy of the methanation reaction (R1) receives positive values at temperatures higher than 600 °C. Therefore, getting close to this temperature (typically for T > ~400 °C; Figure 2b), the initially increasing YCH4 passes through a maximum and then lessens, approaching the equilibrium predicted profile. On the other hand, as expected and shown in Figure 2c, CO production is activated at elevated temperatures (ca. 350-400 °C) where the reverse water-gas shift (rWGS) reaction is thermodynamically favored. Then, it keeps increasing upon increasing temperature at the  At the relatively high wGHSVs employed (ca. 56400-93600 N mL/g·h; low effective mean contact time τ(CO 2 )~1.3 s), the CO 2 hydrogenation reaction is practically activated at temperatures higher than~180-200 • C for all three catalysts tested ( Figure 2a); then CO 2 conversion increases with a temperature approaching the thermodynamic equilibrium curve at about 550 • C. The clear superiority of Rh particles dispersed on ACZ support (with intermediate OSC = 101 µmol O 2 /g) on CO 2 conversion efficiency is evident in the whole temperature range investigated (200-600 • C), while Rh particles on CZ support (with the highest OSC = 557 µmol O 2 /g) appear worse, even compared with Rh particles dispersed on γ-Al 2 O 3 with negligible oxygen storage capacity and lability. Interesting differences in the CH 4 and CO product distribution features between the three catalysts are also perceivable (Figure 2b,c). Compared with γ-Al 2 O 3 , the most active Rh(1.7 nm)/ACZ catalyst effectively promotes CH 4 yield (i.e., the CO 2 methanation reaction (R1)) at temperatures up to~425 • C, while at higher temperatures, its Y CH4 profile approaches that of Rh(1.6nm)/γ-Al 2 O 3 . On the other hand, the methanation efficiency of Rh(2.1 nm)/CZ appears significantly depressed and inferior to that of Rh(1.  Table 2). Table 2. Support-mediated promotion characteristics, apparent activation energies, and pre-exponential factors for CO 2 methanation reaction over Rh(1.6nm)/γ-Al 2 O 3 , Rh(1.7nm)/ACZ, and Rh(2.1nm)/CZ catalysts with different OSC of their supports.

Catalyst
Support According to thermodynamics, CH 4 yield is restricted at elevated temperatures (Figure 2b), while the Gibbs free-energy of the methanation reaction (R1) receives positive values at temperatures higher than 600 • C. Therefore, getting close to this temperature (typically for T >~400 • C; Figure 2b), the initially increasing Y CH4 passes through a maximum and then lessens, approaching the equilibrium predicted profile. On the other hand, as expected and shown in Figure 2c, CO production is activated at elevated temperatures (ca. 350-400 • C) where the reverse water-gas shift (rWGS) reaction is thermodynamically favored. Then, it keeps increasing upon increasing temperature at the expense of CH 4 production, thus rendering the "CO 2 methanation system" to a "syngas production system" (Figure 2b-d). It is also evident that the Y CO profile is systematically moved to lower temperatures as the support's OSC increases ( Figure 2c). Thus, CO productivity for each catalyst is ignited according to the following temperature order: i.e., in reverse order, compared to the order of the OSC values of the supports, OSC(CZ) A notable effect of the OSC of the support on the Rh-catalyzed rWGS reaction (R2) is therefore obvious; the higher the OSC of the support, the greater the promotion of rWGS reaction at elevated (>~350 • C) temperatures.
The CH 4 and CO selectivity performances of the Rh(1.6nm)/γ-Al 2 O 3 and Rh(1.7nm)/ACZ catalysts depicted in Figure 2d are in accordance with the CH 4 and CO productivity discussed above. They offer 100% selectivity toward CH 4 (S CH4 ) at the low-to intermediate-temperature range (200 to~350 • C), while at higher temperatures, the selectivity toward CO (S CO ) starts increasing, with a concomitant S CH4 decrease, in accordance with thermodynamics. The symmetry between their S CH4 and S CO curves in respect to the horizontal invisible line at S = 50%, reflects the fact that no other products except CH 4 Catalysts 2020, 10, 944 9 of 25 and CO were obtained by these two catalysts; the carbon balance X CO2 = Y CH4 + Y CO for these two catalysts always closed well with deviations of less than~3 %.
Nevertheless, the selectivity behavior of Rh(2.1nm)/CZ appears to be rather more complicated. For this catalyst, the carbon balance, X CO2 = Y CH4 + Y CO , did not close well, implying the formation of additional C-containing by-products, besides CH 4 and CO. Indeed, such C 2+ products (mostly C 2 H 6 ) were observed in gas chromatographs, but at very low concentrations, inappropriate for quantitative analysis within the detection limits of our analysis system. It is worth noting that formation of C 2+ products in small amounts has been reported in the literature of CO 2 hydrogenation over Rh/TiO 2 and Rh/Nb 2 O 5 catalysts with the main product being C 2 H 6 [72]. An overview and thorough explanation of the effect of the support OSC on the catalysts' light-off and turnover frequency (TOF) performances is given below, following the presentation of the latter, since both should have a common interpretation.

CO 2 Hydrogenation Intrinsic Activity of Rh Nanoparticles
Turnover frequency results for the CO 2 methanation (TOF CH4 ) on Rh(1.6nm)/γ-Al 2 O 3 , Rh(1.7nm)/ACZ, and Rh(2.1nm)/CZ are shown as Arrhenius plots in Figure 3; TOF CH4 is given as moles of methane produced per mole of rhodium active site per second (Rh active sites were estimated on the basis of mean Rh particle size values given in Table 1). Kinetic data were acquired under conditions of low CO 2 (and H 2 ) conversion, typically~5-15%, in order to reflect intrinsic activity, unaffected by reactants concentration gradients, mass, and/or thermal transport constraints (differential reaction conditions). Apparently, Rh dispersed on ACZ with the intermediate OSC shows the highest methanation activity, followed by Rh on γ-Al 2 O 3 and finally by Rh on CZ, with the higher OSC exhibiting the lower activity ( Figure 3). Its TOF CH4 values are also positively compared with values recently reported for Rh catalysts supported on other supports and studied at similar conditions (examples include References [35,40]).     (Table 2). These apparent activation energy values are similar to previously reported values over supported Rh catalysts. Solymosi et al. [73] reported an apparent activation energy value of 68 kJ/mol on Rh/γ-Al 2 O 3 , while more recently Karelovic and Ruiz [40] reported an interesting dependence of E a on Rh particle size for Rh/γ-Al 2 O 3 catalysts, which varied in the range of 61-95 kJ/mol when Rh particle size varied between 3.6 and 15.4 nm with the smaller particles to obey the higher E a values due to the structure sensitivity of the reaction, as the authors confirmed. On the other hand, in cases where Rh particles were dispersed on supports imposing significant metal-support interactions, such as TiO 2 and W 6+ -doped TiO 2 , elevated E a values were typically obtained: 71 kJ/mol on Rh/TiO 2 [35] and 70-103 kJ/mol on Rh/(W 6+ -doped TiO 2 ) [41] catalysts.
Taking into consideration both integral and intrinsic CO 2 hydrogenation performance of the Rh nanoparticles dispersed on the different supports, γ-Al 2 O 3 , ACZ, and CZ (Figures 2 and 3; Table 2), and given the large variation in oxygen ion lability between them that can stimulate possible metal-support interactions [53,66], the main experimental observations are summarized as follows:  Table 2), providing a volcano-type promotion behavior in respect to the OSC of the support or the counterpart catalyst ( Figure 4). (ii) ACZ and CZ promote the reverse water-gas shift reaction (CO formation) in a monotonic manner in respect to their OSC value; the higher the OSC, the higher the promotion of rWGS (Figure 2c).
In particular, CZ also promotes the formation of additional C-containing byproducts, besides CH 4 and CO ( Figure 2d). (iii) The apparent activation energies (E a ) of CH 4 formation and the corresponding pre-exponential factors (TOF o CH4 ) increase monotonically with increasing the OSC of the support, i.e., E a and TOF o CH4 values follow the order Rh/CZ > Rh/ACZ > Rh/γ-Al 2 O 3 ( Figure 3; Table 2).
Catalysts 2020, 10, x FOR PEER REVIEW 11 of 26 (iii) The apparent activation energies (Ea) of CH4 formation and the corresponding pre-exponential factors ( ) increase monotonically with increasing the OSC of the support, i.e., Ea and values follow the order Rh/CZ > Rh/ACZ > Rh/γ-Al2O3 ( Figure 3; Table 2). It is readily understood that the combination of the effects of (i) and (ii) determines the obtained Rh/ACZ > Rh/γ-Al2O3 > Rh/CZ order for the maxima in CH4 yield and the temperatures where these maxima are located (Figure 2b and Table 2), as well as the behavior of the selectivity toward CH4 and CO (Figure 2d). The question arising regards the origin of these effects. It has been previously demonstrated that CeO2-based supports are characterized by substantial oxygen storage capacity (Table 1) due to the labile lattice oxygen and therefore by high concentration of bulk and surface oxygen ion defects (oxygen vacancies, V .. ) [57,[60][61][62]. Therefore, the suggestion of a bi-functional reaction mechanism, commonly employed in mechanistic reaction schemes involving CO2 processing over CeO2-containing catalysts [53], is consistent with the present catalytic system as well. It is readily understood that the combination of the effects of (i) and (ii) determines the obtained Rh/ACZ > Rh/γ-Al 2 O 3 > Rh/CZ order for the maxima in CH 4 yield and the temperatures where these maxima are located (Figure 2b and Table 2), as well as the behavior of the selectivity toward CH 4 and CO (Figure 2d). The question arising regards the origin of these effects. It has been previously demonstrated that CeO 2 -based supports are characterized by substantial oxygen storage capacity (Table 1) due to the labile lattice oxygen and therefore by high concentration of bulk and surface oxygen ion defects (oxygen vacancies, V .. O ) [57,[60][61][62]. Therefore, the suggestion of a bi-functional reaction mechanism, commonly employed in mechanistic reaction schemes involving CO 2 processing over CeO 2 -containing catalysts [53], is consistent with the present catalytic system as well.
According to this mechanism, CO 2 activation (dissociative adsorption) is especially favored on CeO 2 -containing supports incorporating a significant concentration of oxygen vacancies (V .. O ) that act as additional active centers, via the reaction (R3) [53]: Such a contribution of the support to CO 2 activation (R3) should ultimately be beneficial for the overall methanation reaction progress. Specifically, the elevated population of surface oxygen vacancies on ACZ and CZ supports will substantially promote CO 2 scission with a concomitant effective promotion of the rWGS reaction (R2), a key step in the CO 2 methanation pathway. Therefore, the enhancement found in both Y CH4 and Y CO yields on Rh/ACZ catalyst is fully consistent with the effect of its OSC characteristic on the CO 2 hydrogenation performance ( Figure 2). The pronounced CO evolution (Y CO ) by Rh/CZ compared to Rh/ACZ (and even more compared to Rh/γ-Al 2 O 3 , Figure 2c) is a consequence of the higher availability of V .. O centers on the surface of CZ, resulting from its higher OSC, which facilitates reaction (R3). However, the inferior Y CH4 of the Rh/CZ still remains to be interpreted, and it can be readily understood as follows. An additional effect of the occurrence of reaction (R3) is the continuous replenishment of the support with interstitial O 2− ions (O i ) that can spontaneously (thermally driven) backspillover onto the surface of Rh particles creating an effective double layer [O δ− , δ + ] via the step (R5) [53,66]: where Rh * denotes a rhodium active site on the surface of Rh particles.
According to the effective-double-layer account of catalysts promotion and metal-support interactions [66] the as-created O δ− layer on the surface of metal particles ( Figure S2 in the Supplementary Material) acts as an electronic modifier of the catalytic particles by its compensating charge, δ + , altering their work function and chemisorptive properties and hence the intrinsic activity and/or selectivity toward catalytic reactions [52,57,66,74], before the ultimate reaction of O δ− species as sacrificial promoter (O δ− species are quite reactive and therefore consumed by the oxidizable reactants (e.g., H 2 ) that scavenge them [66]). Nevertheless, they are effective because of their continuous replenishment via reactions (R3) and (R5). Apparently, the intensity of the O δ− layer and consequently its concomitant effect on intrinsic activity are in accordance with the oxygen ions capacity and mobility of the support.
In this context, the increased work function, due to the creation of the effective double layer ([O δ− , δ + ]), or equivalently the decreased electron availability on the surface of Rh particles, results in strengthening of the chemisorptive bond of electron donor (electrophobic) adsorbates and weakening that of electron acceptor (electrophilic). Nevertheless, CO adsorption characteristics on catalyst particles play a key role in CO 2 methanation mechanism [22,37,39]. CO chemisorption on Pt-group metals involves both donation and backdonation of electrons. Therefore, it is difficult to estimate which one of its electrophilic or electrophobic character dominates under the specific conditions employed, and therefore to quantify the effect of increasing metal work function on its chemisorption bond strength [54,75] (additional information is provided in the Supplementary Materials, Section S2). The present results, however, point toward the electrophobic character of CO chemisorption bonding in agreement with References [75,76]. Indeed, the monotonic increase in the apparent activation energy of the methanation reaction upon increasing the OSC of the support (observation iii, Table 2), which is associated with the concomitant enhancement in intensity of the O δ− layer, accompanied by the increasing work function of Rh, rationally implies a strengthening of the Rh-CO bond; that is, the electron-donor (electrophobic) character of CO dominates in our catalytic system under the conditions employed. More specifically, the increase in the apparent activation energy of the CO 2 methanation reaction indicates that the reaction proceeds through schemes with higher energetic barriers; a more strongly bonded, thus less reactive, CO is consistent with this observation. This view is also consistent with the observed increase of the entropic (TOF o CH4 ) Arrhenius factor (observation iii, Table 2): Strengthening of the Rh-CO bond implies an increase in the amount of adsorbed CO on the Rh surface, and therefore to an enhancement of the reaction probability, i.e., higher methanation rate, as indeed obtained (observation i). However, over-strengthening of the Rh-CO bond can finally induce self-poisoning of the reaction, i.e., CO-poisoning of the surface by strongly adsorbed, less-reactive CO, which is fully consistent with the inferior methanation activity of Rh/CZ and the concomitant volcano-type dependence of the promotion on the support's OSC (observation i; Figure 4). Apparently, the values of these two compensative factors (E a and pre-exponential factor) of the methanation reaction (R1) are optimized when supports with intermediate OSC values (e.g., ACZ) are used.
Our recent post-DRM-reaction XPS studies on Rh/γ-Al 2 O 3 , Rh/ACZ, and Rh/CZ catalysts [53] have shown that, between the Rh 3+ , Rh 1+ , and Rh • oxidation states of rhodium, the principal Rh species was Rh • ; its relative content decreasing in the order Rh/CZ(100%) > Rh/ACZ(72%) > Rh/γ-Al 2 O 3 (55%). That is, supports with high oxygen ion lability (ACZ and CZ) due to O 2− backspillover that weakens the Rh-O bond destabilize rhodium oxide, promoting its metallic state [53]. Bearing in mind that, under CO 2 hydrogenation and DRM reactions, the catalyst is exposed to a relatively close reaction products environment, and the bifunctional character of CO 2 methanation reaction, which requires reduced metal sites that are active in H 2 dissociation [26,27,37,38], it is reasonable to suggest that the increase in Rh • state on the catalyst containing ACZ and CZ supports is an additional factor promoting the methanation reaction however, minor in effect compared to the aforementioned one related to CO formation and its chemisorption strengthening on metal sites, able to create reverse effects (CO-poisoning) in the case of supports with very high OSC value (CZ).
The work of Liu et al. [24] on the effect of CeO 2 addition on Ni/Al 2 O 3 catalyst for CO 2 methanation is worth mentioning here. By incorporating different amounts of CeO 2 between 0 and 6 wt %, the authors found that methanation is significantly promoted by low CeO 2 loadings, optimized at an intermediate loading of 2 wt %, and then depressed for higher CeO 2 loadings. The CeO 2 -induced promotion was attributed to the enhanced reducibility of the resulted Ni/CeO 2 -Al 2 O 3 catalysts. Although the OSC values of their catalysts are not available for a closer comparison with the present results, the general feature of the reported promotion is in qualitative agreement with the volcano-type behavior found herein as a function of the OSC (reducibility) of our catalysts.

Effect of Rh Particle Size on CO 2 Methanation Performance
In this context, Rh/Al 2 O 3 and Rh/CZ catalysts, as representative extreme cases regarding the OSC value of the supporting material, were selected for a detailed investigation of the effect of mean particle size of Rh on CO 2 methanation reaction, i.e., its structure sensitivity, and possible synergy with the OSC characteristic of the support. Figure 5 illustrates the CO 2 methanation performance (X CO2 , Y CH4 , Y CO , S CH4 , and S CO ) of the three Rh/γ-Al 2 O 3 catalysts different in the mean Rh particle size: 1.2, 1.6, and 2.1 nm. Although the range of particle size values does not show a large variation, the reaction apparently exhibits structure sensitivity, clearly influencing CO 2 conversion (Figure 5a) and CH 4 yield (Figure 5b), but not to any significant extent the CO yield (Figure 5c), thus causing marginal effects on system selectivity ( Figure 5d). Overall, CH 4 formation performance superiority follows the order Rh(1.2nm)/γ-Al 2 O 3 > Rh(1.6nm)/γ-Al 2 O 3 >> Rh(2.1nm)/γ-Al 2 O 3 ; the reverse water-gas shift (rWGS) reaction appears insensitive to Rh particle size.

Rh/γ-Al 2 O 3 Catalysts with Different Mean Rh Particle Size
Catalysts 2020, 10, x FOR PEER REVIEW  14 of 26 catalysts. More specifically, it is attributed to the fact that the thermal sintering method applied to our catalysts for changing the size of the particles leads to significantly narrow particle size distribution [65,77], as compared to methods employing different metal loadings to achieve this which commonly lead to a fairly broader particle size distribution.   Under intrinsic reaction conditions (Figure 6), the trend of methane formation turnover frequency (TOF CH4 ) is similar to the above order, while an increasing trend of the apparent activation energy of the methanation reaction upon decreasing Rh particle size is obtained: 67.5 kJ/mol < 70.5 kJ/mol << 77.9 kJ/mol over Rh(1.2nm)/γ-Al 2 O 3 , Rh(1.6nm)/γ-Al 2 O 3 , and Rh(2.1nm)/γ-Al 2 O 3 , respectively ( Figure 6 and Table 3). catalysts. More specifically, it is attributed to the fact that the thermal sintering method applied to our catalysts for changing the size of the particles leads to significantly narrow particle size distribution [65,77], as compared to methods employing different metal loadings to achieve this which commonly lead to a fairly broader particle size distribution.    The structure sensitivity of Rh/γ-Al 2 O 3 -catalysed CO 2 methanation is known in the literature. Studying Rh/γ-Al 2 O 3 catalysts with a variety of Rh particle sizes (between 3.6 and 15.4 nm) obtained by applying different Rh loadings (between 1 and 5 wt %), Karelovic and Ruiz [40] reported significant structure sensitivity of the titled reaction, which was favored on larger Rh particles but only at low temperatures (135-150 • C). At a higher temperature (200 • C), the authors observed similar CH 4 formation TOFs on all Rh particle sizes, or slightly better on the smaller one. Here, we report a clear beneficial influence on CH 4 formation activity of the smaller particles at temperatures higher than 200 • C. Based on the facts that the catalysts studied herein concern small Rh particle sizes (1.2-2.1 nm), smaller than that in Reference [40], and the temperature window is higher than 200 • C (ca. 200-350 • C), the observed structure sensitivity and its trend are actually in agreement with that reported by Karelovic and Ruiz at similar conditions. An increasing variation of the apparent activation energy, in the range of ca. 61-95 kJ/mol, with decreasing the particle size was also reported by the authors [40]. Changes in E a in the present study have an opposite trend: E a is increased (up to 10 kJ/mol; Figure 6, Table 3) as the mean Rh particle size increases from 1.2 to 2.1 nm. The different trend compared to that found in Reference [40] most probably originates from the different reaction conditions applied, but also on possible differences on the individual particle distribution of the catalysts. More specifically, it is attributed to the fact that the thermal sintering method applied to our catalysts for changing the size of the particles leads to significantly narrow particle size distribution [65,77], as compared to methods employing different metal loadings to achieve this which commonly lead to a fairly broader particle size distribution. Figure 7 illustrates the full CO 2 methanation performance of Rh/CZ catalysts, in respect to variation in mean Rh particle size, between 2.1 to 5 nm. Apparently, the Rh/CZ catalyst exhibits rather stronger structure sensitivity than that previously depicted for Rh/γ-Al 2 O 3 with the following principal features: The CO 2 conversion is significantly depressed with decreased size of Rh particles (Figure 7a), similarly to the CH 4 yield (Figure 7b), while only marginal effects are recorded in CO yield (Figure 7c). However, as has been already mentioned in Section 3.2.1, the selectivity behavior of the Rh/CZ catalysts appears to be rather more complicated, especially over Rh(2.1nm)/CZ with the lowest Rh particles size (Figure 7d), for which the carbon balance (X CO2 = Y CH4 + Y CO ) did not close well, implying the formation of additional C-containing byproducts besides CH 4 and CO [72]. Overall, as shown in Figure 7, the methanation superiority over Rh/CZ catalysts in relation to Rh particle size follows the order Rh(5nm)/CZ > Rh(2.3nm)/CZ > Rh(2.1nm)/CZ; the reverse water-gas shift (rWGS) reaction appears, again, rather insensitive to Rh particle size.   The Arrhenius plots of the turnover frequency of methane formation (TOF CH4 ) on Rh/CZ catalysts with different particle sizes ( Figure 8) show a similar to the aforementioned methanation trend. Moreover, a slight decrease in the apparent activation energy upon decreasing Rh particle size is obtained: 83.6 kJ/mol < 85.8 kJ/mol < 87.5 kJ/mol over Rh(5nm)/CZ, Rh(2.3nm)/CZ, and Rh(2.1nm)/CZ catalysts, respectively, while the pre-exponential factor is marginally affected (Figure 8; Table 4).    The above effects are reasonably understood by the following considerations: lower in size dispersed Rh particles correspond to a larger amount of undercoordinated sites, leading to enhanced metal-support interactions. Therefore, as previously discussed, in regard to Figure 2 results, the Rh(5nm)/CZ fresh catalyst inhibiting effect on CO 2 methanation of the very high OSC CZ support, due to over-strengthening of the Rh-CO bond, becomes even worse in the case of smaller particles.

Rh/CZ Catalysts with Different Mean Rh Particle Sizes
That is, the lower the Rh particles size, the higher the undercoordinated sites and metal-support interactions responsible for reaction inhibition via enhanced CO-poisoning. Indeed, the inhibiting effect is more pronounced on Rh(2.1nm)/CZ and less so on Rh(2.6nm)/CZ, as compared to that on Rh(5nm)/CZ catalysts (Figure 8). Although quite small, the obtained increase in apparent activation energy by decreasing Rh particles size (Table 4) is consistent to the above interpretation. Vulnerability of the CO 2 methanation reaction to CO-poisoning has also been reported by Italiano et al. [22] on Ni/CeO 2 catalysts. In addition to the above, strongly bonded CO on Rh surface seems to promote side reaction paths, leading to the formation of a variety of C 2+ byproducts at the expense of methane formation (Figure 7b,d and Figure 8). It should be noted, however, that the occurrence of these competitive side reactions makes the system more complex and the full understanding of the behavior of the methanation reaction more difficult. Figure 9 depicts the impact of the support's nature (γ-Al 2 O 3 and CZ) and mean metal particle size on CO 2 methanation intrinsic activity (TOF) of supported Rh nanoparticles, in a comparative manner. The results clearly reveal that the structure sensitivity of CO 2 methanation over Rh is a combined effect of metal-support interactions and particle size: over high OSC capacity support (Rh/CZ catalysts), the reaction is favored by larger particles, in apparent contrast to that obtained over Rh/γ-Al 2 O 3 catalysts (with lack of OSC). The reported here, for the very first time, synergy between the particle size and the OSC characteristic of the support on CO 2 methanation structure sensitivity provides an effective method for promoting catalysts' methanation efficiency via fine-tuning of these two critical parameters (metal particle size and support OSC) during the design of novel effective CO 2 methanation catalyst formulations.
Regarding the most promising in methanation efficiency Rh/ACZ catalyst, investigation of the Rh particle size on its CO 2 methanation activity, similar to that performed for Rh/γ-Al 2 O 3 and Rh/CZ, was not conducted herein. This is because only a very limiting variation on Rh particle size was achieved for Rh/ACZ upon oxidative treatment at 750 and 850 • C (Supplementary Materials Figure  S1 and Table S1), making the resulted catalysts unsuitable for such an investigation. Nevertheless, optimization studies focused only on this catalyst, involving preparation of a new series characterized by a wide variation in both Rh particle sizes and OSC values of the ACZ support (the latter to be achievable by changing its CeO 2 content) are currently under work. Finally, the time-on-stream stability of Rh/γ-Al 2 O 3 , Rh/ACZ, and Rh/CZ catalysts was investigated in 6 h, in stability experiments, under constant feed and temperature conditions. Independently of the support used, the catalysts were very stable (Supplementary Materials Figure S3), confirming the well-known low sintering and coke formation propensity of Rh experienced in close to the present CH 4 , H 2 , and CO 2 containing environments at elevated temperatures [53]. It is expected, however, that materials with the ability to provide O δ− species onto metal particle surfaces (such as ACZ and CZ) add further to catalyst robustness in two ways: oxidizing deposited carbon, and thus preventing carbon accumulation [53], and preventing particles sintering through the mechanism described in Reference [65]. Although, according to the sacrificial promoter concept [66], O δ− species on the Rh surface are themselves reactive and are rapidly consumed by the oxidizable reactants present (e.g., H 2 ), they remain nevertheless effective because they are continuously replenished by labile O 2− species provided by the support (Equations R3 and R5). At the relatively low temperatures (200-450 • C) advantageous for the CO 2 methanation reaction, the lifetime of O δ− species is expected to be high enough, and, accordingly, their promotional and anti-sintering effects are significant. That is, the lower the Rh particles size, the higher the undercoordinated sites and metal-support interactions responsible for reaction inhibition via enhanced CO-poisoning. Indeed, the inhibiting effect is more pronounced on Rh(2.1nm)/CZ and less so on Rh(2.6nm)/CZ, as compared to that on Rh(5nm)/CZ catalysts (Figure 8). Although quite small, the obtained increase in apparent activation energy by decreasing Rh particles size (Table 4) is consistent to the above interpretation. Vulnerability of the CO2 methanation reaction to CO-poisoning has also been reported by Italiano et al. [22] on Ni/CeO2 catalysts. In addition to the above, strongly bonded CO on Rh surface seems to promote side reaction paths, leading to the formation of a variety of C 2+ byproducts at the expense of methane formation (Figures 7b,d and 8). It should be noted, however, that the occurrence of these competitive side reactions makes the system more complex and the full understanding of the behavior of the methanation reaction more difficult. Figure 9. Effect of Rh particle size on the turnover frequency of CO2 formation (TOFCH4) obtained over Rh/γ-Al2O3 and Rh/CZ catalysts at T= 280 °C. Experimental conditions as in Figures 5 and 7 for Rh/γ-Al2O3 and Rh/CZ, respectively, and the reaction to be operated at the differential mode (low conversions). Figure 9 depicts the impact of the support's nature (γ-Al2O3 and CZ) and mean metal particle size on CO2 methanation intrinsic activity (TOF) of supported Rh nanoparticles, in a comparative manner. The results clearly reveal that the structure sensitivity of CO2 methanation over Rh is a combined effect of metal-support interactions and particle size: over high OSC capacity support Figure 9. Effect of Rh particle size on the turnover frequency of CO 2 formation (TOF CH4 ) obtained over Rh/γ-Al 2 O 3 and Rh/CZ catalysts at T= 280 • C. Experimental conditions as in Figures 5 and 7 for Rh/γ-Al 2 O 3 and Rh/CZ, respectively, and the reaction to be operated at the differential mode (low conversions).

Supported Rh Catalysts
Rh nanoparticles were dispersed on the supports via the traditional wet impregnation method, using rhodium (III) nitrate solution (10% w/v Rh in 20-25 wt % HNO 3 ) purchased from Acros Organics as the metal precursor, after its dilution in water, in order to produce a 2 mg Rh/mL concentrated solution of Rh(NO 3 ) 3 . Appropriate amounts of each support were impregnated with this solution, under continuous stirring at 75 • C, until water evaporation to yield Rh-supported catalysts with 1.0 wt % Rh nominal loading. The resulted suspensions were dried at 110 • C for 12 h, calcined in air at 450 • C for 1 h (for the nitrate precursor decomposition), and then reduced at 400 • C under 50% H 2 /He flow for 2 h, followed by heating (20 • C/min) under 1% H 2 /He flow to 800 • C for 1 h, for the removal of any precursors' residuals and structure stabilization. Portions from the as-prepared Rh/γ-Al 2 O 3 , Rh/ACZ, and Rh/CZ fresh catalysts were further thermally treated as described below for Rh particle size modification.

Modification of Rh Particle Size
Changes in mean Rh particle size (agglomeration or redispersion) were achieved by further treatment of the fresh Rh/γ-Al 2 O 3 and Rh/CZ catalysts at high-temperature oxidative conditions via the methodology described in detail in Reference [65] (a concise summary of the method is given in the results and discussion section). Two treatment protocols were imposed. Protocol #1: heating in 10 NmL/min flow of 20% O 2 /He for 2 h at 750 • C, hereafter denoted as "treated at 750 • C" samples. Protocol #2: heating in 10 NmL/min flow of 20% O 2 /He for 2 h at 750 • C, followed by further heating in the same flow conditions for 2 additional hours, at 850 • C, hereafter denoted as "treated at 850 • C" samples. In this way, a set of seven Rh catalysts, supported on three different OSC supports and with different Rh particle sizes, was achieved that enabled us to a thoroughly investigate the combined effect of these two major material properties (OSC and catalyst particle size) on CO 2 methanation. The resulted catalysts are listed in Table 1, using the code Rh(Xnm)/Support, where X denotes mean Rh particle size, and Support is γ-Al 2 O 3 , ACZ, or CZ. Catalyst particle-size modification by this method enables a more reliable investigation of the structure sensitivity of the CO 2 methanation reaction in light of the advantage of constant metal loading in all the samples under comparison, thus avoiding possible complications created by changing the latter.

Materials Characterization
A thorough characterization of the materials (supports and catalysts) was performed by a variety of techniques and has been reported previously [53,65]. However, additional characterizations were also performed herein. Comprehensive details about the methodology followed are provided in the Supplementary Materials (Section S1); a concise summary follows. Total Rh contents were obtained by means of inductively coupled plasma-optical emission spectroscopy (ICP-OES) measurements. Textural, structural, and morphological characteristics of the synthesized catalysts were determined by Brunauer-Emmett-Teller and Barrett-Joyner-Halenda (BET-BJH) method from N 2 adsorption-desorption isotherms at −196 • C, high-resolution transmission electron microscopy (HRTEM), and powder X-ray diffraction (PXRD). The total oxygen storage capacities (OSC) of the samples were determined by utilizing hydrogen temperature-programmed reduction (H 2 -TPR) measurements conducted on pre-oxidized samples. Rh particle sizes (d Rh ) of the catalysts were determined via HRTEM images and isothermal hydrogen chemisorption experiments; the results were found to be in close agreement. Arithmetic mean values (d Rh ) obtained by the two methods for each catalyst were adopted herein.

Catalytic Performance Evaluation
Catalysts evaluation for the CO 2 methanation reaction (R1) was performed in a continuous flow experimental apparatus (Figure 10) consisting of a feed unit, the reactor, and an analysis unit utilizing on-line gas chromatography (SHIMADZU GC-2014, thermal conductivity detector (TCD), Ar carrier gas, equipped with a HayeSep D column), for the analysis of reactants and products. A feed composition of H 2 /CO 2 /Ar = 20%/5%/75% at 1 bar, which corresponds to the stoichiometric ratio (H 2 /CO 2 = 4/1) of the reaction (R1), was synthesized in the feed unit of the apparatus, utilizing mass flow meters (MKS-247) connected to cylinders containing the necessary gases, i.e., CO 2 (99.6 %), ultra-pure H 2 , and Ar, and fed to the reactor. The total feed flow rate was adjusted so as to maintain the effective mean contact time of the reactants with Rh active sites at the same value (τ(CO 2 ) = 1.26 s; τ(H 2 ) = 5.04 s). Since the various catalytic samples did not actually have an identical number of active sites per mass, the effective contact time is the key parameter that should be kept constant, in order to obtain rational comparison of their relative merits (at constant wGHSV, the comparison is not actually made on equal terms, due to the differences in the number of active sites that the reactants are exposed to). It is defined as the surface Rh atoms/(reactant molecules/s) [56,67] and estimated via Equation (1).
where w cat is the mass (g) of catalyst loaded in the reactor; X Rh is the rhodium content of the catalyst (g Rh /g cat ); M Rh is the molecular weight of rhodium (102.9 g/mol); F in (N mL/min) and [CO 2 ] in (v/v) are the total flow rate and CO 2 concentration in the reactor inlet, respectively; V mol is the molar volume of an ideal gas at room temperature and 1 atm pressure (24,450 N mL/mol); and D Rh is the rhodium dispersion associated with the mean Rh particles sizes (d Rh ), estimated via Equation (2).
where d Rh is the mean Rh particle size (nm), ρ Rh is the Rh metal density (12.4 g/N mL), a Rh is the area occupied by a surface Rh atom (7.58 Å 2 /atom), N AV is the Avogadro number (6.023·10 23 molecules/mol), and 10 23 is a unit conversion factor when the units of parameters in Equation (2) are used as indicated above. Based on the above definitions of the reaction output parameters, the following equation was used to check the carbon balance during kinetic experiments: Equation (4) was found to close well with a deviation of less than 3% for the vast majority of A 3 mm internal diameter tubular quartz, fixed bed, single-pass flow reactor was equipped in the apparatus ( Figure 10) for the catalytic performance data acquisition, which was loaded, in all cases, Catalysts 2020, 10, 944 20 of 25 with 50 mg of catalyst (grain size 180-250 µm), held between two quartz wool plugs; the catalyst temperature was measured by a centered K-type thermocouple ( Figure 10).
In order to keep τ(CO 2 ) at 1.26 s for all catalysts with different concentration of active sites per mass tested, the total feed flow rate (F in ) was accordingly varied between 20 and 100 NmL/min, corresponding to weight-basis gas hourly space velocity (wGHSV = F in /w cat ) variation into the interval 24,000 to 120,000 NmL/g cat h.
The following equations were used to calculate CO 2 conversion (X CO2 ), CH 4 and CO yields (Y CH4 , Y CO ), and corresponding selectivities (S CH4 , S CO ): where F in and F out are the total flow rates in the inlet and outlet of the reactor (NmL/min), respectively; the symbols in brackets are the concentrations (v/v) of the corresponding reactants and products. Based on the above definitions of the reaction output parameters, the following equation was used to check the carbon balance during kinetic experiments: Equation (4) was found to close well with a deviation of less than 3% for the vast majority of kinetic experiments, indicating that CH 4 and CO were the main CO 2 hydrogenation products over the catalysts studied and under the experimental conditions employed.
Attention was paid to both integral (high conversions; light-off performance) and differential (low conversions; intrinsic activity) operation of the reactor, for a better analysis and understanding of the catalytic behavior. Turnover frequency calculations of CH 4 formation, TOF CH4 (s −1 ), defined as molecules of CH 4 produced per surface Rh atom per second (Equation (5)), were based on data obtained at low CO 2 and H 2 conversions (ca. 5-15%; differential operation): where r CH4 (mol/g cat· s) is the intrinsic rate of CO 2 consumption obtained under differential reactor operation and calculated by using Equation (6): A combination of Equations (1), (3b), (5), and (6) finally provides the following:

Conclusions
A detailed investigation of the effect of the support, in regards to its oxygen storage capacity (OSC)/lability and catalyst particle size on important catalytic parameters (light-off characteristics, turnover activity, yield, and selectivity), relevant to CO 2 hydrogenation reaction, on supported Rh nanoparticles, was conducted. This enabled an in-depth understanding of the combined effect of these important catalyst properties on the titled reaction output characteristics. Different oxide supports, namely γ-Al 2 O 3 , ACZ, and CZ, with marginal, intermediate, and high oxygen ion lability, were used, over which the same loading (but different in crystallite size) of Rh nanoparticles was dispersed.
The results unambiguously show that OSC of the support is a significant catalyst property that can effectively determine the CO 2 methanation behavior of Rh. Specifically, supports with intermediate OSC values offer optimal CO 2 methanation performance. The effective double layer model of metal-support interactions, i.e., the spontaneously created O δ− layer on the surface of Rh particles via O 2− backspillover from supports with high oxygen lability and its concomitant effect on Rh-CO bond strength, fits well with the observations. This effect also accounts for the final competition result between the methanation and reverse water-gas shift reactions during CO 2 hydrogenation.
The structure sensitivity of the Rh-catalyzed CO 2 methanation reaction is confirmed to be a combined result of OSC of the support and mean Rh particle size, capable of being inversed on the same metal, depending on the oxygen lability characteristic of the support: CO 2 methanation efficiency is favored on smaller Rh particles when dispersed on support with lack of oxygen ion lability (Rh/γ-Al 2 O 3 catalyst), in full contrast to Rh particles dispersed on support with high oxygen ion lability (Rh/CZ), where the reaction is strongly favored on larger Rh particles.
Some new insights and a sense of awareness can be offered by these findings, which appear highly promising for rational design of CO 2 methanation catalysts; the simultaneous tailoring of support OSC and active phase particle size characteristics of the catalysts enables fine-tuning and optimization of their CO 2 methanation efficiency and selectivity, gaining merits and benefits on a process aiming to control greenhouse gas and renewable fuel production.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4344/10/8/944/s1. Figure S1: HRTEM images of the fresh, treated at 750 • C and treated at 850 • C Rh/γ-Al 2 O 3 (a, b, c), Rh/ACZ (d, e, f), and Ru/CZ (g, h, i) catalysts, respectively. Figure S2: Schematic representation of the effective-double layer account of chemical promotion and metal support interactions. Figure S3: First 6 h time-on-stream stability of fresh Rh/γ-Al 2 O 3 , Rh/ACZ and Rh/CZ catalysts at constant feed (5% CO 2 /20% H 2 /75% Ar at 1 bar) and temperature (T = 380 • C) conditions. Author Contributions: G.B., G.G., A.R., E.N., N.C. and P.Z. contributed to investigation, materials synthesis, catalytic data acquisition, and analysis; I.V.Y. contributed to the conceptualization, methodology, supervision of studies, results interpretation, visualization, writing-original draft, coordination and project administration, resources, and funding acquisition; M.K., D.G., N.C., M.G. and S.P. contributed to results interpretation, discussion, co-supervision of studies, reviewing and editing, resources; I.Y. edited and submitted the manuscript in the final form; All authors have read and agreed to the published version of the manuscript Funding: This research was co-funded by the European Union and Greek national funds through the Operational Program "Competitiveness, Entrepreneurship and Innovation", under the call "RESEARCH-CREATE-INNOVATE" (project code: T1E∆K-00782).

Conflicts of Interest:
The authors declare no conflict of interest.