Looking for an Optimal Composition of Nickel-Based Catalysts for CO 2 Methanation

: AbstractA detailed critical analysis of the scientiﬁc literature data concerning catalysts for CO 2 methanation based on nickel supported over oxides was performed. According to the obtained information, it seems that an ionic support is necessary to allow a good nickel dispersion to produce very small nickel metal particles. Such small metal particles result in being very active toward methanation, limiting the production of carbonaceous materials. The use of support and/or surface additives gives rise to medium surface basicity, allowing medium-strong adsorption of CO 2 , and it is also advisable to increase the reaction rate. A medium nickel loading would allow the free support geometric surface to be covered densely by small nickel metal particles without the production of larger Ni crystals. It is also advisable to work at temperatures where Ni(CO) 4 formation is not possible (e.g., >573 K). The promising properties of systems based on doped Ni/Al 2 O 3 , doped with basic and re-active oxides such as MnO x or/and CeO 2 , and those based on Ni/CeO 2 were underlined.


Introduction
The methanation of captured CO 2 using green hydrogen is considered a possible way to valorize and reuse it, also in view of the so-called "power to Gas" technologies [1,2]. The new process to be designed for CO 2 methanation [3] is strictly related to ones developed decades ago for Substitute Natural Gas (SNG) production from CO-rich syngases [4]. It possibly implies a CO 2 purification step followed by a catalytic reaction step. Depending on the performances and design of the catalytic reactor and on the final use of the produced methane, the following steps may be needed: condensation and removal of coproduced water, product separation from unreacted CO 2 and/or hydrogen (that can be recycled to the reactor), and separation from byproducts, such as CO and/or other hydrocarbons or oxygenated compounds. Thus, as always for industrial chemical plants, the key step for an efficient process is to develop high activity and high selectivity catalysts, just to reduce the onerousness of the separation and recycling steps and to maximize productivity. Although artificial intelligence-based methods are under development for discovering new and better catalysts for industrial chemical reactions [5], up to now, empirical methods based on human intelligence still represent the most reliable approach, in our opinion.

Thermodynamics of CO 2 Hydrogenation
The CO 2 methanation reaction, The production of higher hydrocarbons (paraffins and olefins), as it occurs in the socalled CO2-Fischer Tropsch process, through the following reactions, n CO2 + (3n + 1) H2 = CnH2n + 2 + 2n H2O n > 1 m CO2 + 3m H2 = CmH2m + 2m H2O m > 1 also represent competitive reactions, in principle. These reactions, which are also exothermic equilibrium reactions most favored at low temperature and high pressure, are  The production of higher hydrocarbons (paraffins and olefins), as it occurs in the so-called CO 2 -Fischer Tropsch process, through the following reactions, n CO 2 + (3n + 1) H 2 = C n H 2n + 2 + 2n H 2 O n > 1 m CO 2 + 3m H 2 = C m H 2m + 2m H 2 O m > 1 also represent competitive reactions, in principle. These reactions, which are also exothermic equilibrium reactions most favored at low temperature and high pressure, are less favored than methanation and may be realized in conditions where rWGS is still not favored. The higher hydrocarbon yield, forecasted at thermodynamic equilibrium at atmospheric pressure, is extremely low, and methane also remains the most favored hydrocarbon product at higher pressures [9]. According to Gao et al. [6] and our own thermodynamic calculations, carbon formation theoretically does not occur when the starting molar ratio is H 2 /CO 2 = 4, while it should occur at T < 873 K when H 2 /CO 2 = 2. In contrast, the calculations of Schmider et al. [10] report the formation of significant amounts of carbon by CO 2 hydrogenation with H 2 /CO 2 = 4 at atmospheric pressure, with a maximum amount at around 800 K, which almost disappears if reaction pressure is raised to 2 MPa.
Thus, the overall available thermodynamic data indicate that, in principle, the best conditions for an industrial CO 2 methanation process are to realize the reaction in the presence of a selective catalyst for methanation, with low or no activity for reverse water gas shift and higher hydrocarbon synthesis, with a moderate H 2 excess (H 2 /CO 2 > 4) working at T < 650 K or less at a moderately high pressure (1-2 MPa) to push conversion and selectivity to methane, limiting coproduction of CO and carbon deposition. In addition, the adoption of high space velocities will limit the significant production of higher hydrocarbons, achieving high CH 4 productivity.
To test and develop optimal catalysts, however, conditions in which possible limits become evident may give more information. To evidence and compare intrinsic catalytic activity and selectivity toward methane, as well as stability with respect to carbon deposition phenomena, studies realized at atmospheric pressure are more informative than those realized at high pressures [11], where the production of CO and carbon are essentially thermodynamically not possible.

Nickel Catalysts for CO 2 Hydrogenation Reactions
As is well known, transition metals are the most catalytically active materials for hydrogenation reactions, at least in sulfur-free environments [12]. As mentioned above, CO 2 hydrogenation can produce CO by reverse water gas shift, methane, and any other hydrocarbon and oxygenated organic compound. According to the literature, while Cuand Pd-based catalysts are very active for methanol and higher alcohol synthesis, and Fe, Co, and Ru catalysts are active for higher hydrocarbon synthesis, Ni and Ru are the most active for CO x methanation reactions. In fact, commercial catalysts are available for methanation in CO-rich syngases [12]: they are essentially based on Ru/Al 2 O 3 for low-temperature applications, mainly for the methanation of CO x residues in hydrogen and ammonia synthesis gas. Ni/Al 2 O 3 is used for the same application and for hightemperature applications for the production of SNG [13].
An enormous amount of research work was recently carried out concerning catalysts for CO 2 methanation. Several reviews collect such studies [14][15][16][17]. Most of these papers, however, when concerning catalyst composition, appear to be mere collections of data with poor criticism. According to the literature, the commercial catalysts for CO-rich syngas methanation are also active for pure CO 2 methanation, although the optimal composition for these quite different applications can be remarkably different. In fact, although CO methanation and CO 2 methanation are certainly related reactions, the two reactants CO and CO 2 have very different chemical and adsorption properties. CO adsorbs strongly, also at very low temperatures, on metal surfaces, including nickel [18], with which it can Energies 2023, 16, 5304 4 of 19 even react in mild conditions, producing the volatile carbonyl compound Ni(CO) 4 [19]. Instead, the interaction of CO with oxides, such as catalyst carriers, is essentially weak [20]. In contrast, CO 2 interaction with metal surfaces may be relatively weak [21] and may dissociatively occur at moderate temperatures, such as on Ni [22]. Instead, CO 2 adsorption on metal oxides occurs at room and even lower temperatures and may be very strong with basic oxides, producing a number of surface and subsurface species [23,24] and sometimes bulk carbonates [21]. The stronger reactivity of CO 2 with oxide carriers suggests that the carrier can have a more relevant role in the kinetics of CO 2 hydrogenation than of CO hydrogenation.
Taking into account that nickel is by far cheaper than ruthenium and other noble metals, and that the reaction temperature for CO 2 methanation may not necessarily be so low, a large part of the current research is centered on nickel catalysts [25,26]. The main approach is to increase the low-temperature activity of Ni-based catalysts to retain total selectivity to methane, limiting deactivation phenomena.

Unsupported Nickel Catalysts
Unsupported Ni (nano)particles can be produced in several ways, as summarized by Abboud et al. [27]. The preparation of ferromagnetic Ni particles [28] can help in the recovery of the catalyst by magnetic methods or by using induction heating systems. Thermal stability against sintering is a main point concerning unsupported metal particles. Lucchini et al. [29] reported relevant sintering and consequent deactivation of differently prepared unsupported nickel nanoparticles upon CO catalytic hydrogenation experiments in the 550-800 K range. CO 2 methanation was investigated on unsupported Ni metal particles produced by the reduction of NiO with H 2 by Garbarino et al. [30], Pandey et al. [31], Varvoutis et al. [32], and Kristiani and Takeishi [33], while Lee et al. tested Raney nickel [34]. Ni particles sized at 100-200 nm show moderate catalytic activity, definitely lower than for supported nickel [30,32]. Low Turn Over Frequency [33] but also low selectivities to methane with both coproduction of CO and a quite quick deactivation was associated with the production of encapsulating carbon and carbon nanotubes [30]; see Figure 2.

Unsupported Nickel Catalysts
Unsupported Ni (nano)particles can be produced in several ways, as summarized by Abboud et al. [27]. The preparation of ferromagnetic Ni particles [28] can help in the recovery of the catalyst by magnetic methods or by using induction heating systems. Thermal stability against sintering is a main point concerning unsupported metal particles. Lucchini et al. [29] reported relevant sintering and consequent deactivation of differently prepared unsupported nickel nanoparticles upon CO catalytic hydrogenation experiments in the 550-800 K range.
CO2 methanation was investigated on unsupported Ni metal particles produced by the reduction of NiO with H2 by Garbarino et al. [30], Pandey et al. [31], Varvoutis et al. [32], and Kristiani and Takeishi [33], while Lee et al. tested Raney nickel [34]. Ni particles sized at 100-200 nm show moderate catalytic activity, definitely lower than for supported nickel [30,32]. Low Turn Over Frequency [33] but also low selectivities to methane with both coproduction of CO and a quite quick deactivation was associated with the production of encapsulating carbon and carbon nanotubes [30]; see Figure 2. Similarly, Hao et al. [35] reported low activity and low selectivity to methane over "bare nickel". According to Pandey [31], the moderate activity of pure nickel is significantly increased by the addition of small amounts of iron. On Raney nickel, selectivity to methane is reported to be not far from 100% for the purest Ni sample, which is also the most active [34]. Very small (d < 8 nm) unsupported nickel nanoparticles produced by the reduction of NiCl2*6H2O with NaBH4 were found almost inactive in CO2 methanation by Riani et al. [36]. However, it must be taken into account that such a preparation method may leave B impurities on the metal samples [37], which can affect catalytic behavior. Lan et al. [38] reported that the increase in the (111) to (200) crystal planes, obtained by exposing the catalyst to a magnetic field during preparation, enormously increases the catalytic Similarly, Hao et al. [35] reported low activity and low selectivity to methane over "bare nickel". According to Pandey [31], the moderate activity of pure nickel is significantly increased by the addition of small amounts of iron. On Raney nickel, selectivity to methane is reported to be not far from 100% for the purest Ni sample, which is also the most active [34]. Very small (d < 8 nm) unsupported nickel nanoparticles produced by the reduction of NiCl 2 *6H 2 O with NaBH 4 were found almost inactive in CO 2 methanation by Riani et al. [36]. However, it must be taken into account that such a preparation method may leave B impurities on the metal samples [37], which can affect catalytic behavior. Lan et al. [38] reported that the increase in the (111) to (200) crystal planes, obtained by exposing the catalyst to a magnetic field during preparation, enormously increases the catalytic activity of unsupported nickel at 453 K, with total methane selectivity.

Nickel Supported on Simple Metal Oxide Carriers
Several papers report a comparison of the catalytic activity of nickel catalysts supported on different simple oxide carriers for CO 2 methanation, with partially agreeing data. According to Muroyama et al. [39], the CO 2 conversion for most catalysts drastically increases at 498-523 K and reaches a maximal value at 573-623 K, approaching thermodynamically allowed values for the given conditions. The order of CH 4 yield at 523 K was as follows: Ni/Y 2 O 3 > Ni/Sm 2 O 3 > Ni/ZrO 2 > Ni/CeO 2 > Ni/Al 2 O 3 > Ni/La 2 O 3. The catalytic activity can be partly explained by the basic property of the catalysts. However, the same authors reported different scales for selectivity, with Ni/Al 2 O 3 being the most selective and performant in methanation at 673 K. In at least partial agreement, Hatzisymeon et al. [40] reported the following sequence of activity: Ni/CeO 2 > Ni/Y 2 O 3− stabilized ZrO 2 > Ni/Al 2 O 3 > Ni/TiO 2 , all with 5 wt% Ni. These authors attribute the trend to the decreasing particle size of Ni metal particles. According to Ma et al. [41], the CO 2 methanation activity order for 5 wt% nickel catalysts on relatively low surface area supports is Ni/CeO 2 > Ni/Al 2 O 3 > Ni/TiO 2 > Ni/ZrO 2 . According to Gac et al. [42], the activity trend for 10-40 wt% Ni-supported catalysts is Ni/CeO 2 > Ni/ZrO 2 > Ni/Al 2 O 3 , all with very high methane selectivity. However, these authors underline that Al 2 O 3 -supported catalysts have the highest resistance against sintering. Lee et al. [43] reported the following activity trend: Ni/CeO 2 ≈ Ni/Y 2 O 3 > Ni/Al 2 O 3 > Ni/TiO 2 . These authors suggest that the reducibility of both metal and support at low temperatures, strong metal-support interaction, and small Ni particle size are important factors for low-temperature CO 2 methanation. Li et al. [44] reported the activity scale of Ni/CeO 2 > Ni/TiO 2 > Ni/SiO 2 , with additional lower selectivity to methane for the Ni/SiO 2 catalyst, attributed to the absence of strong metal-support interaction.

Ni/SiO 2 Catalysts
As it is well known, amorphous silicas are largely covalent supports [45] which are usually supposed to be essentially non-dispersive with respect to other more ionic materials. Depending on the preparation method and type of materials, amorphous silicas have high morphological stability towards sintering and loss of surface area up to 873 K or more [46,47]. Silica is active in reversible adsorption, including of CO 2 , with an important role of surface silanol groups as hydrogen bond donors [48], but without any Lewis acidity and basicity [49]. This is the reason for poor dispersing ability towards supported phases, as well as for poor adsorption of CO 2 without the formation of carbonate-like species.
Preparations of Ni/SiO 2 catalysts have been described by several authors. In general, it has been found that supporting nickel salts on amorphous silica followed by calcination gives rise to segregated NiO particles, although quite a small part of Ni 2+ can be highly dispersed on ion exchange sites, i.e., exchanging protons in silanol group nests [50][51][52]. After reduction, such Ni/SiO 2 catalysts usually mainly present large Ni metal particles with poor dispersion [53] and poor interaction with the support, although the copresence of dispersed unreduced nickel has also been reported [52]. Stability towards sintering of the Ni/SiO 2 catalyst is limited [27], although the use of meso/microporous silica supports and opportune preparation methods can improve it [54].
The CO 2 hydrogenation reaction over Ni/SiO 2 catalysts has been the object of several studies. According to Gac et al. [55], small particles (d = 15.8 nm) of nickel on silica can show very high catalytic activity and selectivity to methane at low temperatures, while selectivity to CO increases with increased particle size and reaction temperature above  [56] reported a different result, showing that low-loading (0.5 wt% Ni/SiO 2 ) catalysts essentially work as a rWGS catalyst while high-loading and high particle size samples (10% Ni/SiO 2 ) are effective methanation catalysts. Galhardo et al. [57] reported high selectivity to CO for 15 wt% Ni/SiO 2 at both low and high temperatures, while near 673 K selectivity to methane is slightly higher than that to CO. These authors report data suggesting that the formation of carbon deposits on Ni metal particles produces Ni carbide species favoring the rWGS activity and deactivating methanation. Li et al. [44] reported lower methane selectivity for Ni/SiO 2 with respect to other supported Ni catalysts. Chen et al. [58] investigated CO 2 hydrogenation on Ni nanoparticles supported on surfacemodified, cage-type mesoporous silica at 773 K, observing high selectivity to CO and low selectivity to methane. Riani et al. [52] reported that low-loading Ni/SiO 2 catalysts with particle sizes < 20 nm showed high selectivity to CO and moderate activity for the rWGS reaction, likely mainly due to nickel species dispersed in silica exchange sites, as evidenced by visible spectroscopy. The same authors reported that high-loading Ni/SiO 2 catalysts with particle size~100 nm show both methanation and rWGS activity, but there is evident short-term deactivation for methanation (see the decreased methane yield in the decreasing temperature steps at the right of Figure 3), attributed to large, segregated Ni metal particles, covered by a carbon veil. evidenced by visible spectroscopy. The same authors reported that high-loading Ni/SiO2 catalysts with particle size ~ 100 nm show both methanation and rWGS activity, but there is evident short-term deactivation for methanation (see the decreased methane yield in the decreasing temperature steps at the right of Figure 3), attributed to large, segregated Ni metal particles, covered by a carbon veil. Hao et al. [35] reported low activity for methanation and high activation energy (125 kJ/mol) for Ni/SiO2 when compared with Ni/CeO2 (Eact near 100 kJ/mol). Guo et al. [60] compared the methanation activity of different 10 wt% nickel catalysts supported on amorphous and mesoporous silica (SBA-15-type and MCM-41-type), showing that they have selectivity to methane approaching 100%, Ni/MCM-41 being definitely less active. In contrast, Bakariza et al. [61] found 15 wt% Ni/MCM-41 more active than Ni/SBA-15. Aziz et al. [62] reported high methanation activity and selectivity for 5% Ni/SiO2 materials, with better performances when using sol-gel prepared mesostructured silica nanoparticles. Yang et al. [63] reported near-equilibrium conversion with 92% methane selectivity on a Ni/MCM-41 silica catalyst, whose activity is improved by zirconium doping. Vogt et al. [64] reported high selectivity to methane for a Ni/SiO2 catalyst with 6 nm Ni particle size but lower selectivity, decreasing with time on stream, for methane for a Ni/SiO2 catalyst with 1.8 nm Ni particle size. Zhu et al. [65] emphasized the progressive deactivation of a 20 wt% Ni/SiO2 catalyst in a 30 h time on stream experiment. However, they also reported a significant increase in activity, selectivity to methane, and stability by further adding, in a second step, additional silica, alumina, or zirconia over Ni/SiO2. Guo and Lu [66] reported that co-impregnation of 1 wt% magnesium improves the activity and stability of 10% Ni/SiO2, but the addition of more Mg results in lower activity and selectivity to methane. Hao et al. [35] reported low activity for methanation and high activation energy (125 kJ/mol) for Ni/SiO 2 when compared with Ni/CeO 2 (E act near 100 kJ/mol). Guo et al. [60] compared the methanation activity of different 10 wt% nickel catalysts supported on amorphous and mesoporous silica (SBA-15-type and MCM-41-type), showing that they have selectivity to methane approaching 100%, Ni/MCM-41 being definitely less active. In contrast, Bakariza et al. [61] found 15 wt% Ni/MCM-41 more active than Ni/SBA-15. Aziz et al. [62] reported high methanation activity and selectivity for 5% Ni/SiO 2 materials, with better performances when using sol-gel prepared mesostructured silica nanoparticles. Yang et al. [63] reported near-equilibrium conversion with 92% methane selectivity on a Ni/MCM-41 silica catalyst, whose activity is improved by zirconium doping. Vogt et al. [64] reported high selectivity to methane for a Ni/SiO 2 catalyst with 6 nm Ni particle size but lower selectivity, decreasing with time on stream, for methane for a Ni/SiO 2 catalyst with 1.8 nm Ni particle size. Zhu et al. [65] emphasized the progressive deactivation of a 20 wt% Ni/SiO 2 catalyst in a 30 h time on stream experiment. However, they also reported a significant increase in activity, selectivity to methane, and stability by further adding, in a second step, additional silica, alumina, or zirconia over Ni/SiO 2 . Guo and Lu [66] reported that co-impregnation of 1 wt% magnesium improves the activity and stability of 10% Ni/SiO 2 , but the addition of more Mg results in lower activity and selectivity to methane.

Ni/Al 2 O 3 Catalysts
In contrast to silica, spinel-type aluminas are considered to be ionic solids with high surface, Lewis acidity and basicity, and a strong ability to disperse supported species [67,68]. Thus, Ni dispersion on alumina is very high. Impregnation of γ-Al 2 O 3 with Ni salts followed by calcination results in the formation of a NiAl 2 O 4 -like surface layer which is hardly reducible, forming, upon reduction, very small Ni metal clusters [69] which are very selective for CO 2 methanation [52,70], as evident from Figure 3. Instead, even at a high Ni loading, the formation of bulk NiAl 2 O 4 does not occur until at least 873 K, given how the reaction between NiO and γ-Al 2 O 3 is still slow and incomplete even at 1273 K [71]. However, the surface spinel layer stabilizes the Ni/γ-Al 2 O 3 catalysts up to high temperatures [72], preventing bulk transformation and loss of surface area.
Surface characterization studies performed by IR spectroscopy of adsorbed CO allowed for the distinguishing of three types of surface Ni metal sites depending on Ni loading on reduced Ni-γ-Al 2 O 3 . For a low Ni loading, nearly isolated metal sites are able to adsorb CO in the form of polycarbonyls; at medium loadings, small nickel metal particles adsorb CO on-top (terminal carbonyls) only, and at higher coverages, larger metal particles adsorb CO also as bridging species [73].
In their screening work, among oxide-supported nickel catalysts, Muroyama et al. [39] found Ni/Al 2 O 3 to be not very active at low temperatures (473-523 K) with respect to other oxide-supported Ni catalysts, but to be the most performant and selective to methane at 673 K. In an early study, Van Herwijnen et al. [74] reported a kinetic study of methanation of diluted CO 2 (1.4 vol%) in hydrogen, over 33.6 wt% NiO/γ-Al 2 O 3 at 473-503 K, with full conversion at sufficiently high contact times already at 500 K at atmospheric pressure. For γ-alumina supports with 180-220 m 2 /g, the best activity and selectivity are reported for catalysts with 15-25 wt% Ni [75,76], which roughly corresponds to the coverage of the support by a "geometric monolayer" of surface Ni aluminate species, while at a lower loading, both activity and selectivity decrease in favor of rWGS activity to CO. Otherwise, at an even higher loading, where larger but still strongly support-bonded Ni metal particles are formed, activity decreases, but deactivation is significant, while high selectivity is usually retained [77].
On the other hand, Ni-Al 2 O 3 catalysts prepared by coprecipitation, such as by coprecipitated Ni-Al hydrotalcite, were reported to be definitely more active and more selective than impregnated Ni/γ-Al 2 O 3 at medium-low loadings [78], but may be less active and selective at a near monolayer loading [75]. On the other hand, coprecipitated catalysts with very high Ni loadings (50-90 wt% Ni) were reported by Beierlein [77] to be definitely more active in methanation and stable than high-loading samples (>30 wt% Ni) produced by impregnation. These authors attribute this activity trend to the effect of higher nickel surface area. Indeed, when working with Ni-Al hydrotalcite, the Ni loading is usually very high. When working with a catalyst with a Ni/Al atomic ratio = 3, high selectivity to CO was found at low contact time, with methane selectivity increasing with increasing contact time [79].

Ni/CeO 2 Catalysts
Most authors of screening papers report ceria as the best oxide support for nickel catalysts used in CO 2 methanation. Hao et al. [35] reported excellent activity and near 100% methane selectivity over 7.5% Ni/CeO 2 with 25 m 2 /g surface area, although the Turn Over Frequency increases by decreasing particle size as a result of decreasing loading. Methanation activity is already significant at 498 K with near 100% selectivity. Similarly, Bian et al. [80] reported activity already in the 473-523 K range for Ni/CeO 2 with a near 100% methane selectivity, which somehow depended on ceria particle morphologies: Ni on ceria nanorods appeared to be more active than Ni on ceria nanocubes. Using ceria nanorods, Varvoutis et al. [32] showed high activity already in the 473-573 K range and equilibrium conversion above 623 K with total methane selectivity for samples with 8 to 33% Ni/CeO 2 (S BET 72 to 51 m 2 /g by increasing loading). On the other hand, low-loading Ni/CeO 2 , with 1-2% nickel, are reported to work as excellent catalysts for the competitive reverse water gas shift reaction to CO [81]. The role of the preparation method has also been emphasized by several authors: Ye et al. [82] reported much higher activity and selectivity of sol-gel prepared Ni/CeO 2 than from an analogous catalyst produced by impregnation. As a key for explaining the best performances of Ni/CeO 2 , the role of the partial reducibility of CeO 2 , with the detection of Ce 3+ ions, has been emphasized [83]. Indeed, among typical oxide carriers, ceria is the most reducible, and also has the ability to absorb hydrogen-forming bulk and surface hydride species [84]. In fact, it has been reported that the Ni/CeO 2 system is quite an unstable one, with the ability to dissolve Ni 2+ ions in the bulk at higher temperatures and longer times [85]. It is, in fact, known that NiO can dissolve in CeO 2 in significant amounts, forming a solid solution with a fluorite structure [86,87], at least for small particle size nanomaterials. A similar deactivation phenomenon with loss of metal surface area is reported, e.g., for Pd/CeO 2 for the Water Gas shift reaction [88].
In their review, Xie et al. [89] summarized recent attempts to further improve Ni/CeO 2 catalysts for CO 2 methanation by modifying the preparation method (constructing more efficient structures and producing more efficient morphologies) and adding additional components, such as alkali, alkali-metal, and rare earth oxides and by alloying Ni with other transition metals. According to these authors, from a practical industrial point of view, challenges remain in developing a durable and low-cost CeO 2 -based methanation catalyst with considerable low-temperature activity.

Ni/ZrO 2 Catalysts
As shown above, most screening studies report high methanation activity for Ni/ZrO 2 . Chen et al. [90] investigated the activity of 15 wt% Ni/ZrO 2 catalysts with predominant monoclinic or tetragonal zirconia support: the monoclinic sample showed lower initial conversion temperature (down to 358 K) and higher basicity than the tetragonal one, both being very selective to methane formation, but with a slight selectivity decrease above 623 K. A similar result was reported by Jia et al. [91] for a 10 wt% Ni/ZrO 2 monoclinic + tetragonal zirconia support. Tan et al. [92] used a monoclinic + tetragonal zirconia support with a surface area of 96.6 m 2 /g to prepare a series of Ni/ZrO 2 catalysts. While all catalysts showed selectivity to methane near 100%, the most active appeared to be that with a 6 wt% Ni loading. According to these authors, the intrinsic activity of the catalyst is related to the Ni dispersion and the size of metallic Ni; thus, excellent activity was obtained with highly dispersed particles. According to Vargas et al. [93], really highly dispersed 1 wt% Ni/m-ZrO 2 prepared by ultrasound-assisted synthesis is very active and selective in methanation of CO 2 . Ma et al. confirmed the high activity of Ni/m-ZrO 2 , more active than a catalyst based on cubic zirconia [94], attributing high activity to the high local electron density of nickel sites. Zhao et al. [95] compared the activities of different 15 wt% Ni/ZrO 2 catalysts produced by combustion with different combustion compounds with a sample produced by impregnation of a mainly monoclinic ZrO 2 support. The catalyst produced by the urea combustion method (resulting in being tetragonal) was reported to be the most active and stable, although, in the 573-773 K range, the impregnated sample was reported to be even more selective to methane. Concerns about the use of Ni/ZrO 2 catalytic materials is due to the potential of NiO to enter monoclinic, tetragonal, and cubic zirconia phases forming solid solutions [96,97] which influence the solid state chemistry of zirconia, apparently favoring the metastable tetragonal phase.

Nickel Catalysts Supported on Other Rare Earth Oxides
In a recent review, Spennati et al. [98] emphasized the role of rare earth oxides in the composition of Ni-based catalysts for CO 2 methanation. According to the screening work of Muroyama et al. [39], where pure support effect has been considered for 10 wt% Ni/oxide catalysts, Ni/Y 2 O 3 and, to a lesser extent, Ni/Sm 2 O 3 , show the highest methanation activities at 523 K, although they lose methane selectivity above 573 K in favor of CO. Thus, the methanation activity of Ni/Y 2 O 3 is closely similar to that of Ni/Al 2 O 3 and Ni/ZrO 2 at 623 K. Other screening studies confirm the high activity of Ni/Y 2 O 3, but no more than Ni/CeO 2 [43]. Li et al. [99] confirmed the very high methanation activity of 10 wt% Ni/Y 2 O 3, if prepared in the presence of large amounts of citric acid in particular. Ni/Sm 2 O 3 catalysts were also reported as interesting, with high low-temperature activity and almost total selectivity to methane, mainly when the sample is highly charged (63 wt% Ni) [100], with a slight deactivation attributed to carbonate formation.
Relatively high-loading Ni/La 2 O 3 catalysts [101] are reported to be selective in the methanation of CO 2, while low-loading Ni/La 2 O 3 appear to work as rWGS catalysts with high CO selectivity [102]. However, studies report that La 2 O 3 is unstable in reaction conditions, suffering from carbonation or hydration with the formation of oxycarbonate and hydroxide production [103], respectively. On the other hand, 10 wt% Ni supported on the oxycarbonate La 2 O 2 CO 3 has been reported to display enhanced activity and stability with respect to Ni/La 2 O 3 [104].

Other Ni/Oxide Catalysts
As commented before, most screening studies find a relatively low activity of Ni/TiO 2 with respect to other oxide-supported Ni catalysts. In their study, Unwiset et al. [105] found catalytic activity only above 573 K, increasing with the Ni loading, and high selectivity to methane for Ni/TiO 2 -anatase samples. The XRD study of the catalysts shows that part of the nickel was incorporated in the anatase framework in the solid solution. When looking at solid state stability, the Ni/TiO 2 -anatase system is considered to be quite an unstable one, since Ni 2+ is a promoter of the anatase to rutile phase transformation [106] that usually also occurs at low temperatures with a significant loss of surface area. For example, the TiO 2 phase transformation was reported to affect the performance of the Ni/TiO 2 catalyst for biomass fast pyrolysis volatiles [107]. On the other hand, Messou et al. [108] found Ni/rutile more active than Ni/anatase, although Ni over a mixed phase support was even more active, showing some kind of synergism between the two TiO 2 phases. A few studies also report on the methanation activity of Ni/MgO catalysts [109,110], where high activity and selectivity to methane can also be found. It must be mentioned, however, that the full reciprocal solubility of NiO and MgO (which are isostructural) may represent an element of instability of the system if prepared by impregnation, or of limited efficiency per nickel atom if prepared by coprecipitation.

Effect of Addition of Doping with a Second Oxide on Ni/γ-Al 2 O 3 and Ni/CeO 2
The data summarized above indicate that Ni/CeO 2 and Ni/γ-Al 2 O 3 are likely the most attractive systems for CO 2 methanation: alumina is usually the preferred support by catalyst producers because of its moderate costs, its well-known chemistry, and its wellproven stability. On the other hand, Ni-Al 2 O 3 catalysts are also already largely used in the industry for CO hydrogenation both to produce Synthetic Natural Gas and to remove CO and CO 2 from hydrogen and ammonia synthesis gas, as well as for several hydrogenations in the treatment of hydrocarbon streams [12]. Otherwise, Ni/CeO 2 seems to give more active catalysts, in particular at low temperatures. The use of CeO 2 is increasingly of interest in heterogeneous catalysis [111].

Effect of a Second Oxide Component on Ni/γ-Al 2 O 3
Silica is a well-known stabilizer of the γ-Al 2 O 3 against phase transition and surface area loss [49,112]. However, the addition of silica in small amounts to γ-Al 2 O 3 shows a slightly negative effect on CO 2 hydrogenation activity, although it may have a positive effect when applied in small amounts on the stability of the methanation activity [52]. Liang et al. reported on the effect of Na, K, Mg, Ca, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Zr, La, and Ce on the activity of Ni/γ-Al 2 O 3 [113], observing that Zn only promotes methanation activity at low temperature, while Cr, Fe, Mn, La, Ce promote stability in prolonged CO 2 methanation experiments. In contrast, Cai et al. [114] reported a marked promotion effect of zirconia on 12 wt% Ni/γ-Al 2 O 3 , in particular when the 3 wt% ZrO 2 /γ-Al 2 O 3 carrier was prepared by the impregnation-precipitation method. In their recent review, Spennati et al. [98] emphasized the positive role of doping Ni/Al 2 O 3 with rare earth oxide components. Ahmad et al. [115] reported the activity scale Pr > La > Gd > Ce > Eu for 12 wt% Ni/γ-Al 2 O 3 doped with 5 wt% rare earth metal. Guilera et al. [116] co-impregnated alumina with nickel and Ce, La, Sm, Y, and Zr compounds and found La and Ce gave the best catalytic performance, while Sm and Zr showed a lower positive effect. Looking at the optimal amount of Ceria, Kim et al. [117] report maximum activation for methanation with 15 wt% CeO 2 for 15 wt% Ni/Al 2 O 3 with 92 m 2 /g surface area. As for the Ni/La 2 O 3 -γ-Al 2 O 3 system, Garbarino et al. [59] reported 14 wt% La 2 O 3 as the best loading for 13.6 wt% Ni/γ-Al 2 O 3 (17 wt% as NiO). In Figure 3, the positive effect of La 2 O 3 addition to Ni/γ-Al 2 O 3 is evident. This effect was in part attributed to the increased strength of absorption of CO 2 in the form of carbonate species over the La 2 O 3 -Al 2 O 3 support, with respect to bicarbonate species on pure Al 2 O 3 . Riani et al. [118] reported later that the coaddition of silica and lanthana to Ni/γ-Al 2 O 3 can give rise to catalysts with even higher low-temperature methanation activity. A catalyst obtained by a precursor with a weight composition of 16.7% NiO, 37.0% La 2 O 3 , 4.6% SiO 2 , and 41.7% Al 2 O 3 was found to be very active at 523 K.
Beierlein et al. [77] confirmed the negative effect of sodium impurities in the methanation activity and selectivity of Ni/γ-Al 2 O 3 . Potassium addition on Ni/γ-Al 2 O 3 was reported by Gandara Loe et al. [119] to significantly increase CO selectivity, finally producing an excellent catalyst for rWGS at T > 873 K. As for alkali earth elements, Liang et al. [120] reported that Ba (1-5 wt% loading) doping of Ni/γ-Al 2 O 3 results in a remarkable selectivity increase, while Sr is less effective and Mg and Ca enhance rWGS reaction, thus reducing methane selectivity. However, Garbarino et al. [121] reported that Ni/Ca-aluminate industrial methane steam reforming catalyst (18 wt% NiO in the catalyst precursor) has excellent activity and selectivity to methane, which are both decreased by 1.8 wt% K 2 O doping. Garbarino et al. [122] found excellent low-temperature activity for Ca-doped 19 wt% Ni/γ-Al 2 O 3 , further enhanced by the addition of V 2 O 5 .
A very interesting point concerns the Ni/MgO-Al 2 O 3 system, which can be prepared by coprecipitation starting from Ni-Mg-Al hydrotalcite-type double hydroxides. Liu et al. reported that Ni/MgAl catalysts (36.9 wt% Ni, (Ni + Mg)/Al = 2) exhibit outstanding performance in CO 2 methanation, compared with Ni/γ-Al 2 O 3 and Ni/MgO [123], associated with a very small Ni metal particle size, allowing better low-temperature CO 2 methanation activity. Very high CO 2 methanation activity and selectivity, higher than that of Ni/γ-Al 2 O 3 , was recently reported by Fasolini et al. [124] for ex-hydrotalcite 27 wt% Ni/MgO-Al 2 O 3 .
Another element that has been reported as a good activator for methanation on Ni/γ-Al 2 O 3 is manganese. Zhao et al. [125] reported maximum activation of 15 wt% Ni/γ-Al 2 O 3 by adding 1.7 wt% Mn. These authors attributed the activation effect to the increased number of CO 2 adsorption sites, the improved Ni dispersion, and the weakened interactions between the nickel species and the support. In good agreement, Vrijburg et al. [126] reported significantly higher methanation rates and CH 4 selectivities for Mn-promoted Ni/γ-Al 2 O 3 . The optimal Mn/Ni atomic ratio was found to be 0.25, showing metallic Ni particles and small oxidic Mn 2+ species. According to these authors, Mn addition improves Ni dispersion and enhances Ni 2+ reducibility by weakening the interaction between the Ni-oxide precursor and the support. Wu et al. [127] reported on the excellent methanation activity of a catalyst with 20 wt% Ni/Al 2 O 3 -ZrO 2 doped with 6.7% Mn. All these studies conclude that Mn oxide species are dispersed on the support surface, favoring nickel dispersion.

Effect of a Second Oxide Component on Ni/CeO 2
A number of studies reported on the further promotion of Ni/CeO 2 for CO 2 methanation. The use of CeO 2 -ZrO 2 solid solutions is a common way to stabilize ceria. Pan et al. [128] found Ni/Ce 0.5 Zr 0.5 O 2 to definitely be more active than Ni/γ-Al 2 O 3 and attributed to the stronger basicity of the Ce 0.5 Zr 0.5 O 2 support, having more and stronger CO 2 adsorption sites than γ-Al 2 O 3 . Looking at different CeO 2 -ZrO 2 support compositions, Ullah et al. [129] found maximum activity for 5 wt% Ni catalysts when the support composition is Ce 0.7 Ni 0.3 O 2 . A number of studies reported the promoting effect of rare earth ions on the activity of Ni/CeO 2 , such as Eu 3+ [130], Sm 3+ [131], Y 3+ [132], the couples La 3+ -Sm 3+ , La 3+ -Pr 3+ , and La 3+ -Mg 2+ [133], etc. Looking at the effect of alkali earth metals, Liu et al. [134] found an increase in the methanation activity of Ni/CeO 2 by adding Ca 2+ and Mg 2+ , with a more limited effect for Sr 2+ and a deactivation effect for Ba 2+ .

Effect of Alloying
Several studies report bimetallic Ni-based catalysts being more active than pure nickel ones in CO 2 methanation. This subject has been reviewed recently by Tsiotsias et al. [26]. The addition of small amounts of noble metals (Pt, Ru, Rh, Pd, and Re) to Ni/Al 2 O 3 provides several advantages, such as excellent activity at low temperatures, high reducibility, and stability once oxidized [135], producing even new sites or by forming an alloy with nickel as evidenced for Ru and Pt or Pd, respectively. However, the use of very expensive noble metals to activate nickel may make sense only if they are added in extremely small amounts, due to their activity even at small loadings over oxide support. According to Pan et al. [136] the addition of 0.2 wt% Ru causes a 3-fold rate enhancement at low temperature for selective methanation of CO 2 relative to the monometallic 20 wt% Ni/CeO 2 and to 0.2 wt% Ru/CeO 2 .
The use of less expensive transition metal to increase Ni activity seems to be more interesting. As already cited, according to Pandey [31], the moderate activity of pure nickel is significantly increased by the addition of small amounts of iron. For example, the addition of 1 wt% cobalt was reported to significantly increase the low-temperature activity of 15 wt% Ni/MgO-Al 2 O 3 catalysts [137]. Similarly, the addition of 4.7 wt% iron significantly increases the low-temperature activity of a 20 wt% Ni/MgO-Al 2 O 3 catalyst [138]. According to Xu et al. [139], the additions of Fe, Co, and, to a lower extent, Mn on Ni/Ce 0 . 8 Zr 0.2 enhance activity, while copper addition decreases both activity and selectivity to CH 4 . Over low-loading Ni/La 2 O 3 , copper addition increases rWGS activity and selectivity to CO [102].

Discussion
The data summarized above show that, essentially, all Ni-containing catalysts found activity in CO 2 hydrogenation. In Table 1, the most relevant criteria for selecting optimal Ni-based CO 2 methanation catalysts are proposed. Catalysts where most of the nickel species remain unreduced, because they strongly interact with the oxide support, usually have predominant rWGS activity to CO, while those where sufficient nickel can be reduced to the metallic state have at least good initial activity to methanation. It seems that good and stable methanation activity is associated mainly with very small nickel metal particles associated with very high nickel dispersion, which can be stabilized mainly on ionic oxides. In fact, for large particle Ni catalysts, fast deactivation by carbon deposition is usually observed, which may also result in switching toward higher CO selectivity by rWGS activity, likely due to carbon or nickel carbide species. Methanation occurring on unsupported Ni particles and Ni/SiO 2 should fully occur at the nickel metal surface, where carbide or carbon intermediates are reasonably formed and active in methanation [140]. However, mechanisms via oxygenate intermediates (carboxylate, carbonate, formate, formyl, and methoxy) can also occur at nickel surfaces [141]. On ionic oxides, where CO 2 adsorbs and forms different types of carbonate and bicarbonate species [23], a more important role of the support in the methanation mechanism is plausible [142].
Very high low-temperature methanation activity is obtained on small Ni metal particles supported on semiconducting and slightly reducible metal oxides characterized by medium surface basicity, such as, in particular, for Ni/CeO 2 . On the other hand, it seems also quite evident that the addition of basic, semiconducting, and moderately reducible components to alumina (a non-conducting, non-reducible, and mainly acidic support) is sufficient to significantly increase the methanation activity at 473-523 K, up to values comparable to those obtained with (doped) Ni/CeO 2 , and with Ru/Al 2 O 3 as well. Such components to be added to alumina support include, e.g., cerium or manganese oxides.
The activating effect of such an intermediate level of basicity is usually attributed to the resulting ability of the support to adsorb CO 2 in the form of carbonate species that, thanks to the spillover of hydrogen from nickel metal particles, is at the origin of a oxygenate intermediate-based hydrogenation mechanism [143]. For example, the activating effect of the addition of La 2 O 3 to Ni/γ-Al 2 O 3 has been attributed in part to the increased strength in the adsorption of CO 2 [59]; as shown in Figure 4, the adsorbed species of CO 2 are fully modified.
On γ-Al 2 O 3, two slightly different forms of bicarbonates are observed (Figure 4, bands at 1644 cm −1 ; O-C-O asymmetric stretching, 1484, 1442 cm −1 , two different O-C-O symmetric stretchings, 1236 cm −1 , COH deformation mode), whose band decreased by the addition of La 2 O 3 until almost disappearing, while bands due to two forms of carbonates are formed (COx stretchings at 1680 and 1380 cm −1 , likely bridging carbonates, and at 1540 and 1435 cm −1 , likely monodentate carbonates). However, slight effects in modifying the electronic properties of Ni metal particles by increasing support basicity as well as semiconducting or redox species cannot be excluded [59].
As shown above, the addition of small amounts of another metal, e.g., cobalt or ruthenium, can also produce a significant increase in activity at low temperatures.
On the other hand, to develop industrial catalysts, besides (initial) activity, other factors must be considered and are even more relevant. In their work concerning hightemperature methanation of CO-rich syngases, Nguyen et al. [13] reported that catalysts must be stable against metal particles and support sintering, formation of carbon and gum residues, as well as the formation of nickel tetracarbonyl. Such a volatile compound can favor Ni particle growth through its formation and decomposition. The need to avoid Ni(CO) 4 formation drives the limiting of nickel loading and the use of sufficiently high sufficient to significantly increase the methanation activity at 473-523 K, up to values comparable to those obtained with (doped) Ni/CeO2, and with Ru/Al2O3 as well. Such components to be added to alumina support include, e.g., cerium or manganese oxides.
The activating effect of such an intermediate level of basicity is usually attributed to the resulting ability of the support to adsorb CO2 in the form of carbonate species that, thanks to the spillover of hydrogen from nickel metal particles, is at the origin of a oxygenate intermediate-based hydrogenation mechanism [143]. For example, the activating effect of the addition of La2O3 to Ni/γ-Al2O3 has been attributed in part to the increased strength in the adsorption of CO2 [59]; as shown in Figure 4, the adsorbed species of CO2 are fully modified. On γ-Al2O3, two slightly different forms of bicarbonates are observed (Figure 4, bands at 1644 cm −1 ; O-C-O asymmetric stretching, 1484, 1442 cm −1 , two different O-C-O symmetric stretchings, 1236 cm −1 , COH deformation mode), whose band decreased by the addition of La2O3 until almost disappearing, while bands due to two forms of carbonates are formed (COx stretchings at 1680 and 1380 cm −1 , likely bridging carbonates, and at 1540 and 1435 cm −1 , likely monodentate carbonates). However, slight effects in modifying the It seems that the limitation of the formation of carbon residue and gum can be obtained using catalysts allowing the production of small to very small nickel metal particles in reaction conditions. This is apparently obtained using high-dispersing ionic oxide supports and a moderate loading of nickel.
Another point is certainly the solid state stability of the support and its reactivity towards nickel species. In fact, most metal oxides allow solubility of Ni 2+ , forming mixed oxides. The penetration of nickel in the bulk of the support, with the help of the slight oxidizing properties of the reactant CO 2 and of the product H 2 O, would result in the consumption of the active phase (metallic nickel) and in the deterioration of the morphological properties of the support. From this point of view, γ-Al 2 O 3 seems to be optimal, taking into account that it tends to be covered by a NiAl 2 O 4 -like surface layer upon the preparation of the catalyst that protects the bulk from further Ni penetration. On the other hand, for decades, the use of Ni/Al 2 O 3 catalysts in several other hydrogenation processes in relevant industrial conditions has been a confirmation of the stability of this system. In fact, as reported recently by Spennati et al., commercial Ni catalysts for CO x methanation seem to be exclusively based on Ni/Al 2 O 3 [98]. This might be due to its cheapness and well-known chemistry of alumina, coupled with its commercial availability, stability in mild conditions, and versatility, which make it preferred by the industry, e.g., as hydrotreating catalysts, as reported years ago [144]. It must also be taken into consideration that even though coprecipitated Ni/Al 2 O 3 -based catalysts may be a little more active than impregnated ones at least at initial steps, the less complete use of nickel in coprecipitated samples and the likely simpler preparation of catalysts by co-impregnation of commercial alumina with nickel and dopants can have advantages (also from the environmental point of view) with respect to more complex coprecipitation procedures.

Conclusions
A detailed analysis of the scientific literature data concerning catalysts for CO 2 methanation allowed us to evidence the main features needed for such an application. It seems that the necessary features of the best-performing oxide-supported nickel catalyst for CO 2 methanation are as follows: 1.
The use of an ionic support that allows good nickel dispersion, resulting in very small nickel metal particles. Such small metal particles would be very active toward methanation and would not produce many carbon residues and gums;

2.
The use of a support and/or surface additives that produce a medium basicity, allowing a medium-strong adsorption of CO 2 ; 3.
A medium nickel loading, which allows the free support's geometric surface to be covered with optimal density by small nickel metal particles, without the production of larger Ni crystals and allows sufficient active sites for CO 2 adsorption on the support surface; 4.
Taking into account the already significant use of Ni/Al 2 O 3 systems for catalytic hydrogenations, its ascertained long-term stability, and the common use of γ-alumina as a catalyst support, the application of 15-20 wt% nickel over a typical γ-Al 2 O 3 (200 m 2 /g) with small amounts of slightly reducible and basic promoters, such as Ce 4+ and Mn n+ , or/and of simply mild basic components, such as Mg 2+ , and maybe small amounts of an alloying metal, such as cobalt or iron, can be recommended. The most likely alternatives are based on Ni/CeO 2 with additives, whose long-term stability, however, should be verified.