Review of CO₂ Reduction on Supported Metals (Alloys) and Single-Atom Catalysts (SACs) for the Use of Green Hydrogen in Power-to-Gas Concepts

Abstract

The valorization of carbon dioxide by diverting it into useful chemicals through reduction has recently attracted much interest due to the pertinent need to curb increasing global warming, which is mainly due to the huge increase of CO₂ emissions from domestic and industrial activities. This approach would have a double benefit when using the green hydrogen generated from the electrolysis of water with renewable electricity (solar and wind energy). Strategies for the chemical storage of green hydrogen involve the reduction of carbon dioxide to value-added products such as methane, syngas, methanol, and their derivatives. The reduction of CO₂ at ambient pressure to methane or carbon monoxide are rather facile processes that can be easily used to store renewable energy or generate an important starting material for chemical industry. While the methanation pathway can benefit from existing infrastructure of natural gas grids, the production of syngas could be also very essential to produce liquid fuels and olefins, which will also be in great demand in the future. In this review, we focus on the recent advances in the thermocatalytic reduction of CO₂ at ambient pressure to basically methane and syngas on the surface of supported metal nanoparticles, single-atom catalyst (SACs), and supported bimetallic alloys. Basically, we will concentrate on activity, selectivity, stability during reaction, support effects, metal-support interactions (MSIs), and on some recent approaches to control and switch the CO₂ reduction selectivity between methane and syngas. Finally, we will discuss challenges and requirements for the successful introduction of these processes in the cycle of renewable energies. All these aspects are discussed in the frame of sustainable use of renewable energies.

1. Introduction—CO₂ Reduction and Energy Perspectives

The catalytic reduction of carbon dioxide to useful chemical products is gaining exponentially increasing interest as it is considered one of the most important targets for the energy sector and at the same time is of high importance for obvious ecological reasons. CO₂ is related to global warming and resulting climate changes, which may lead to serious environmental and economic problems in the near future [1]. The symptoms of this climate problem, which have recently appeared in many areas across the world, have led to the quest to push global warming well below 2 °C, a level still higher than the average earth temperature of the pre-industrial era [2]. This aspiration needs either industrial activities to cease, or efficient technologies to be developed to capture CO₂ and then use it in a closed energy cycle [3,4,5]. Considering the enormous economic damage that would be caused by the suppression of industrial activities and the expected terrible socio-political consequences all over the world, the second option, which aims at strict mitigation of carbon emissions and their cyclic utilization is the only possible way forward.

The anticipated technical approach to solve this problem is based on: (i) harvesting carbon dioxide from air or at the outlet of emitters, which would need the development of suitable materials and technologies, (ii) the production of green hydrogen by water electrolysis using electricity generated from renewable energy resources, basically solar and wind energy, and finally, (iii) the reaction of green hydrogen with harvested CO₂ to produce either energy carriers (e.g., methanol, formic acid or methane) or other chemicals (see proposed energy cycle illustrated in Figure 1). This approach is not only ecologically favorable, but also central for new scientific concepts related to the sustainable use of renewable energy resources. Given the seasonality of electricity generation from renewable energy resources (i.e., fluctuations in energy supply due to change in weather, solar radiation, wind speed, etc.), the chemical storage of excess electricity, which is produced during periods of maximum electric power generation, in a stable and easily convertible energy carrier (e.g., as methanol, formic acid, or methane), would help to buffer the expected fluctuations in renewable energy supply.

For this purpose, different strategies were proposed to react hydrogen with carbon dioxide, either to produce energy carriers with higher energy density, or otherwise convert both molecules into a value-added compound. One of the top targets is the direct conversion of CO₂ to methanol, which serves as energy carrier in the first place (power-to-liquid). Methanol can be stored and transferred globally in a much easier way than hydrogen, which needs a huge amount of energy for liquefaction and more importantly requires extreme safety measures [6]. In this way, the liquid and easily transportable and storable methanol can be used to release stored hydrogen, or to produce other useful chemicals [7], or it can be directly used to generate electricity in the direct methanol fuel cells (DMFCs) [8]. The development of efficient catalysts for the synthesis of methanol from CO₂ reduction is still in the beginning and a lot of progress is still needed to achieve scalable, highly active, methanol selective, and commercially feasibly catalytic materials [9]. The other simple approach is the transformation of CO₂ to methane or syngas (i.e., power-to-gas concept).

Direct methanation of carbon dioxide, known as Sabatier reaction, is considered as a well-developed process that is in the course of being introduced into the renewable energy sector [10]. In this approach, the resulting methane, referred to as synthetic natural gas, can be stored and distributed and even transferred (exported/imported) globally using the already existing infrastructures (natural gas grid, transfer pipelines, liquefaction stations, and carrier ships) (see scheme in Figure 1). The methanation process is also very preferable from the technical point of view, as the operation conditions and safety measures are less complicated than the methanol production process. On the one hand it has been studied for over one hundred years, since it was first introduced by Paul Sabatier on a wide range of catalytic materials [11,12], which makes it close to a well-established technology for large scale production [13]. On the other hand, methanation of CO₂ can be operated at ambient pressure or slightly increased pressure with high activity and reasonable selectivity compared to methanol synthesis, which needs the use of much higher pressures to achieve enough selectivity and activity toward methanol yield. At this point, it is essential to mention the industrial implementation of the power-to-gas concept, originally in Germany. Preliminary studies, which focused on the development of so-called alpha plants, were carried out in Stuttgart. These findings then led to the possibility of building the Audi e-gas plant in Werlte. The electrical input from electrolysis here is 6 MW [14]. The hydrogen obtained from this step is compressed to 10 bar and brought to the Sabatier reaction with biogas-CO₂ at temperatures between 200 and 350 °C via staged dosage. A nickel catalyst is used, which promotes the conversion of carbon dioxide to over 90%. This development has been the seed of many other similar power-to-gas projects in several countries all over the world [15,16].

The reduction of CO₂ to syngas is considered as another important route with increasing importance over the recent years. This is related to the need for sustainable production of petrochemical commodities from renewable resources and anthropogenic carbon dioxide, especially with the continuous decrease of proven world reserves of crude oil. Here, it should be noted that syngas is typically produced from the reforming of natural gas or crude oil. In the not so far future, the syngas produced from CO₂ reduction will be able to be converted to a wide range of platform chemicals used in chemical industry, basically we refer to (i) the production of light olefins used in the polymer industries (ethylene, propylene and butene) [17], (ii) the commercial production of methanol [18], and (iii) liquid hydrocarbon fuels via Fischer Tropsch synthesis [19]. These products will continue to be indispensable for the chemical industry in the future, but also for air and maritime transport (see schematic in Figure 1). Based on these aspects, it seems that both reaction pathways, CO₂ methanation and RWGS reactions, are complementary for the sustainable use of renewable energy (i.e., green hydrogen) in the near future, even after the optimization of CO₂ reduction to methanol. It is thus highly desirable to find strategies for switching the selectivity of CO₂ reduction between these two processes to allow alternatives for both applications.

2. Thermodynamic and Kinetic Considerations

The product of CO₂ reduction depends decisively on the reaction conditions (temperature, pressure, and stoichiometry of the reactants) that affect the thermodynamic equilibrium. Considering a fixed pressure (i.e., here ambient pressure) and constant gas composition of CO₂/H₂ = 1:1, we can see in Figure 2 the effect of variation of the reaction temperature on the thermodynamic equilibrium composition. The direct pathways of CO₂ reduction at ambient pressure can proceed either via reverse water gas shift (RWGS: CO₂ + H₂ => CO + H₂O; ΔH = −41.1 kJ mol⁻¹) or via the Sabatier reaction/CO₂ methanation (CO₂ + 4 H₂ => CH₄ + 2H₂O + ΔH = −165 kJ mol⁻¹) [20,21]. Given the much higher exothermicity for CO₂ methanation compared to RWGS reaction, one can expect in the first glance, that RWGS would be more promoted at high temperatures and at the same time CO₂ methanation would be less pronounced. Looking at the equilibrium conversion of CO₂ to different products, it can be observed that the start of the decline of CO₂ methanation pathway commences at a temperature of 350 °C and continues to decrease with the increase of the temperature, which goes along with the decline in methane yield (T > 600 °C; methane yield < 30% (see Figure 2)).

The onset of CO formation via RWGS can be observed at a temperature of about 350 °C and continues to increase with temperature. In general, we see that methanation is thermodynamically more favorable at lower temperatures than CO formation. This primarily determines the working temperatures of the desired catalytic reaction, in which methanation (Sabatier pathway) or CO formation via RWGS should be operated [22]. These thermodynamic conditions should always precede a reaction design, as the catalyst can only change the chemical kinetics. However, considering the high importance of syngas production from CO₂ as outlined above, it is highly desirable to overcome kinetic limitations of the RWGS pathway to achieve high productivity at the lowest possible temperature. This can be achieved by the development of catalytic materials able to significantly lower the energy barrier for CO formation, prevent CO-poisoning, and at the same time lifting the energy barrier for CH₄ formation. This would mean that the RWGS pathway will prevail over the CO₂ reduction pathway. This issue will be discussed in more detail in Section 5.

3. Catalysts for CO₂ Reduction—Activity, Stability, and Selectivity of Different Supported Metals

3.1. Historical Background

The direct hydrogenation of CO₂ was first introduced by Paul Sabatier during his work in the beginning of the twentieth century on the development of a direct hydrogenation method on metal surfaces [23]. These early trials included the reduction/hydrogenation of a wide range of compounds including, nitriles, aromatics, unsaturated hydrocarbons, and carbon oxides, where gaseous hydrogen was basically used as reducing agent [12]. Focusing on the methanation of carbon oxides, Sabatier and Senderens found out that the contact of gaseous hydrogen with CO or CO₂ on the surface of nickel powder results in the formation of methane [11]. Different metals have been studied over time such as Ru, Rh, Ir, Co, Ni, Fe, and Pd, and were found active for the methanation of both CO and CO₂ [13]. In their early work on the synthesis of higher hydrocarbons from carbon oxides, Fischer and Tropsch compared in a systematic way the activity of different metals toward methanation activity and for the synthesis of higher hydrocarbons, known later as FT synthesis [24]. As a general consensus, it could be concluded that the metals active for methanation of CO are active for CO₂ methanation but are expected to be less active toward the FT synthesis of higher hydrocarbons [25]. For instance, the behavior of Ru and Ni catalysts on the one side and Fe and Co catalysts on the other side are clearly proving the correctness of this concept. Ru and Ni are by far the most active metals for methanation of CO and CO₂ at ambient pressure, but once transferred to the FT reaction regime they show poor selectivity in FT synthesis while Fe and Co shows the opposite behavior, i.e., promoting FT pathway and suppressing the CO₂ reduction to methane and/or CO [13,19]. Ru and Ni are, however, studied and used as promoters for FT catalysts.

Focusing on the widely studied CO₂ reduction under ambient pressure conditions, we will discuss and compare the behavior of different metals that are more active for the transformation of CO₂ to methane and syngas at a wide range of temperatures depending on the nature of catalyst and its support materials. In the following, we will contrast the advantages and disadvantages in possible applications for the most studied metals (Ni, Ru, and Rh) in addition to the recently emerging bimetallic catalysts based on different combinations of these metals together with other cheaper elements.

3.2. Nickel-Based Catalysts

Nickel as catalytically active element received an enormous amount of interest since the methanation process was developed by Sabatier and Senderens [11], not only because of its relatively high activity in CO₂ reduction at ambient pressure, but also because to its much lower price compared to other more active metals such as Ru, Rh, and Pt. Nickel has also been combined in different compositions with different oxide supports including the famous catalyst known as Raney nickel (nickel-aluminum alloy) [13], and nickel nanoparticles supported on a wide range of oxides such as Ni/Al₂O₃ [26,27], Ni/SiO₂ [28], Ni/TiO₂ [29,30], Ni/ZrO₂ [31], Ni/CeO₂ [32,33,34,35]. In this way, the use of reducible oxide supports was found to significantly improve the activity of nickel. In particular, compared to the standard Ni/Al₂O₃ catalysts, Ni/CeO₂ showed higher activity at low temperature, which allowed for higher selectivity toward methane (this parameter will be discussed in detail in Section 4). Recent trials to improve the performance of nickel catalysts in methanation and methane reforming involved the use of mixed oxide or ternary composites (Al₂O₃-ZrO₂ [36], WO₃–ZrO₂ [37], Ni-Mg-Al [38]) and also doping with transition metals such as Fe, Co, Mn [39].

The major challenge in the use of conventional nickel-based catalysts is basically their strong and continuous deactivation with time-on-stream. As an example, Ewald et al., demonstrated systematically that Ni/Al₂O₃ prepared with different loadings using co-precipitation and impregnation methods show continuous loss of activity at temperatures in the range from 250 to 350 °C [40]. Figure 3a shows the relative activity loss of such a catalyst in a narrow temperature range (250 to 350 °C). The deactivation rate of these catalysts was found to depend on the method of catalyst preparation. However, in general the loss of catalyst surface area along with the sintering of Ni nanoparticles during CO₂ reduction with time-on-stream were considered as the major reasons for the process. In connection with these findings, Munnik et al. [41], observed similar deactivation behavior for Ni/SiO₂, but during CO methanation in CO/H₂ (1:2) at 1 bar and at 230 °C (Figure 3b). These authors explained the continuous deactivation over time by the formation of large Ni nanoparticles. This is driven by the formation of Ni(CO)₄ species in the presence of CO and their decomposition on the support to form adatomic Ni or tiny Ni_x clusters, which migrate across the surface merging to the larger particles (ripening effect) and therefore leading to continuous loss of active sites (i.e., sintering via Ostwald ripening).

Considering this explanation, the following criteria should be taken into account for the efficient use of these catalysts, and hence prevent their continuous deactivation. First, the use of Ni catalysts at low temperatures <350 °C is not desirable if the catalyst is partially active for RWGS due to the formation of Ni(CO)₄, which would lead to catalyst deactivation in addition to its poisonous nature, which raises major safety issues. Second, Ni catalysts would be more desired for use at high temperature for the reduction of CO₂ to CO, where methane is also thermodynamically less preferred compared to CO formation, and at the same time Ni(CO)₄ formation is avoided. Some of these supported nickel catalysts such as Ni/SiO₂ [28], Ni/MgO [42], Ni/MgAl₂O₄ [43], Ni/Al₂O₃-CeO₂ catalysts are reported as excellent candidates for the RWGS over methanation and has been widely studied for this aspect [44]. Furthermore, nickel is known for its Janus-like catalytic behavior in the reductive conversion of CO₂ with hydrogen. The reaction preferably follows one of two possible mechanisms: (i) the redox mechanism forming CO_ad species on the catalyst surface, and (ii) an H₂ mediated cycle via HCOO_ad and COOH_ad to CO_ad adsorbates [45,46,47]. After these elementary steps, either desorption of CO (RWGS) or further hydrogenation to CH₄ (Sabatier reaction) takes place, which is dependent on the activation barriers of both the steps. Probably, the origin of this ambivalent behavior lies in the variation of the adsorption strength of the species formed on the catalyst surface with changes in the structure of the catalysts. In this regards, it is also known that the selectivity for methane increases with increasing the Ni particle size [48]. These mechanistic aspects are not yet fully understood and further studies are indispensable to tailor more stable and selective Ni catalysts for CO₂ reduction.

3.3. Ru-Based Catalysts

Supported Ru catalysts have been also one of the most studied catalysts for the reduction of CO₂ for over four decades [49,50,51,52,53,54]. In this regard, these catalysts have recently attracted intensive investigations in the recent few years by several research groups [55,56,57,58,59,60,61,62,63] as one of the promising options for power-to-gas applications. This high interest is supported by their impressively high activity at rather low temperatures [64,65]. In comparison with supported Ni catalysts prepared using similar methods and with comparable structural characteristics (support material, metal loading, treatment), Garbarino et al. showed that Ru/Al₂O₃ outperforms Ni/Al₂O₃ catalyst in activity and stability [26]. In terms of metal-based catalytic activity, supported Ru catalysts were found to be at least one order of magnitude more active [66]. The unsurpassed high activity in particular at low temperature of supported Ru catalysts is preferable not only for rationalizing the overall costs of the catalytic process, but also it gives preference for the methanation pathway when catalyst preparation and activation steps are well controlled.

Different from supported Ni catalysts, Ru nanoparticles (1–3 nm) supported on different metal oxides such as Al₂O₃ [67,68,69], TiO₂ [66,70], ZrO₂ [68,71], were found to be highly stable during time-on-stream measurements in CO₂ methanation. As an example, the comparison of the Ru mass normalized activity for CO₂ reduction at 190 °C/1 bar on different oxide supports (Al₂O₃, TiO₂, and ZrO₂) shows almost no/or very limited deactivation with time-on-stream (see Figure 4).

Note that the high stability of these catalysts during CO₂ methanation is independent of the initial catalyst pretreatment step, whether a reductive, or an oxidative gas atmosphere is applied for this purpose. We see that in both cases, the catalysts are highly stable (oxidative pretreatment -Figure 4a; reductive pretreatment-Figure 4b), although the activity is much more pronounced after reductive pretreatment. It is also clear that there is an initial activation phase for reducible oxides after the oxidative treatment (Figure 4a), which is related to the slow reduction of the support and the formation of O-vacancy defects, which are needed for the activation of CO₂ on the catalyst surface. Here, we need to underline that different from the behavior of supported Ru catalysts in CO₂ reduction, these catalysts (Ru/MO_x) show different behavior in the methanation of CO in CO/H₂ gas mixtures or in CO/CO₂/H₂ mixtures. In the presence of CO, modest deactivation can be observed, in particular for the reducible oxides such as Ru/TiO₂ catalysts [70,72,73,74]. This effect is also observable on Ru NPs (1–3 nm) supported on non-reducible oxides in the presence of water vapor in CO/CO₂/H₂ feed gases with limited CO partial pressures (0.01%CO) [75,76,77]. The reason for the deactivation of these catalysts in the presence of CO could be attributed to a combination of effects evolving simultaneously with time including (i) the change of the particle shape of Ru NPs from flat to hemispherical or spherical [72], (ii) sintering of small Ru NPs [72], and (iii) encapsulation of Ru NPs by reduced TiO_x species at high temperatures [66]. These effects lead individually or together to the loss of active Ru surface area during CO/CO₂ methanation with time on stream. More importantly, it was demonstrated that supported Ru catalysts (Ru/Al₂O₃) are also better catalysts than the corresponding Ni-based catalyst (Ni/Al₂O₃) under dynamic operation conditions involving the variation of reactor load [78,79]. This property is one of the prerequisites for the future decentralized use of CO₂ methanation reactors/stations. Therefore, a suitable stability of the catalyst under dynamic changes is mandatory to enable real applications.

Referring to the selectivity between methanation and reverse water gas shift reaction, except for supported Ru catalysts on TiO₂ with large specific surface area (> 200 m²g⁻¹), which show pronounced CO formation, most of these Ru/TiO₂ catalysts (SSA ≤ 125 m²g⁻¹) are active for methanation in the temperature range from 50 up to 350 °C [65,66,67,71,80], with selectivity almost of 100%. This aspect will be discussed in more detail in Section 5.2.

3.4. Rh-Based Catalysts

Like supported Ru catalysts, the thermo-catalytic reduction of CO₂ on oxide supported rhodium catalysts attracted significant attention due to their high activity at moderate to low reaction temperatures (<200 °C). These catalysts were intensively investigated by Solymosi and coworkers and were reported in several papers [81,82,83,84]. These authors showed that Rh is also active for the photocatalytic reduction of CO₂ in the presence of water vapor to different products other than CO and CH₄, such as formaldehyde, formic acid, methanol, which are viewed as candidates for hydrogen fixation in liquid carriers [85]. The thermal reduction of CO₂ by reaction with methane to CO + H_2, commonly known as dry reforming of methane_, on supported Rh catalysts has been studied by several groups on different supports including Rh supported on TiO₂ [86], SiO₂ [86,87], Al₂O₃ [86,88,89]. This reaction can serve as a possible pathway for the sustainable production of syngas and at the same time for valorizing the flaring methane at the natural gas wells. The approach can be useful in the periods of shortage of hydrogen production from renewable resources, where methane stored can be used for hydrogen or syngas by direct reaction with CO₂. Although supported Rh catalysts are known for their high methanation selectivity at low to moderate temperatures (<250 °C), recent results by Matsubu et al. demonstrated that the selectivity of CO₂ reduction can be steered toward CO by either atomic dispersion of Rh on metal oxide [90], and also through the modification of metal support interaction [91]. These results shall be discussed in more details in Section 5.

3.5. Bimetallic Alloy Catalysts

The combination of two or more metals in one alloy is intended on the one hand to create better electronic/structural properties, which are not reachable on supported single metals, and on the other hand is an approach to rationalize the use of one of two metals, which is usually a precious rare metal such as Ru, Pt, or Rh, by combination with less expensive metal such as Ni, Co, or Fe, e.g., the core-shell materials where an expensive metal is prepared as outer shell of a cheaper metal. The first example, which is widely studied, and is considered highly promising as cheap and stable option for CO₂ methanation, is NiFe alloys, which were first described by Andersson et al. [92].

In their study, these authors found out that the activity for CO hydrogenation depends decisively on the ratio of Ni:Fe in the alloy, with the maximum reaction rate achieved for an alloy composition of 1:1 (see Figure 5) [92]. These catalysts received recently increasing attention in experimental studies for CO₂ reduction at ambient pressure [93,94,95,96,97,98,99]. The amount of Fe in the NiFe alloy was found to affect the activity decisively and an optimum amount in the range from 20 to 25 mol% results in the highest catalytic activity at lower temperatures (200–400 °C) when compared to supported Ni catalysts [99,100]. In this regard, Meshkini et al. demonstrated that an alloy based on 80% Ni and 20% Fe supported on Al₂O₃ achieves about 100% CO₂ conversion to methane at 270 °C, which is not feasible on supported Ni catalysts [96]. The quantitative comparison of the methanation activity of the Ni alloy with different metals including Fe (Fe, Co, Zr, La, and Cu) indicated that FeNi alloy outperforms the catalytic activity and selectivity of the Ni alloys with other elements, with the highest CO₂ conversion rate and methanation selectivity close to 100% at 350 °C [101].

Possible explanation for the role of Fe was discussed by Huong et al., who used FTIR spectroscopy and DFT calculations for this purpose [102]. These results led to the conclusion that the overall activation barrier of CH₄ formation is significantly lowered on the alloy surface. In another study, Sherr et al. demonstrated that FeO_x clusters can be observed during reaction on top of the surface of FeNi nanoparticles, which play a vital role in the activation of CO₂ during reaction [103]. In addition to the higher activity of these catalysts, they are advantageously preferred during reaction on stream due to their enhanced stability. In a comparative study of the behavior of commercial Ni catalysts and the supported NiFe alloy (Ni₃Fe/Al₂O₃), Mutz et al. [104], reported much higher activity and time-on-stream stability, with much limited deactivation of the Ni₃Fe alloy (<5% loss of activity in 40 h at 358 °C). This highlights the vital role of alloying in solving one of the severe drawbacks of pure or supported nickel catalysts.

The second example is bimetallic RuNi alloy, which has been recently studied as a cheaper alternative for expensive Ru, offering a combination of characteristics of Ru and Ni catalysts [105,106]. As an example, Lange et al., showed that low Ru concentration-based RuNi alloy supported on ZrO₂ supports can achieve comparable activity to the pure Ru catalysts supported on ZrO_2, which are more active and stable during the CO₂ methanation than Ni/ZrO₂ catalysts [105]. The addition of Ru as promoter to the mesoporous NiAlO_x oxides promoted by CaO was also found to increase the methanation activity for CO₂ at lower temperatures with higher sintering resistance of Ni nanoparticles compared to the pure Ru free catalysts [105]. Although there is no significant enhancement of the performance of the resulting bimetallic catalysts, they are, however, much cheaper than the pure Ru counterparts [105]. These catalysts have been studied recently also for selective CO methanation [107,108]. Nevertheless, it is clear that these RuNi alloys have not received so far, enough attention for CO₂ reduction/methanation compared to the NiFe system. In general, the addition of small amounts of Ru to Ni in an alloy or as promoter in mesoporous NiAlO_x oxides results in the enhancement of the reducibility of Ni and therefore the adsorption and spillover of hydrogen, which increase their methanation activity and on-stream stability [107]. The alloying of Ru with Fe was also reported to yield an active catalyst for CO₂ reduction/hydrogenation [109,110], in addition to earlier reports about its use in CO methanation [111].

Proano et al. also studied the effect of alloying ruthenium or platinum with Ni to improve the catalytic properties of nickel catalysts for CO₂ methanation [112]. These alloys were combined with alkaline adsorbent for the capture of carbon dioxide and its subsequent methanation. Both metals, Ru and Pt, were found to promote the reduction of NiO formed during catalyst calcination. Ru resulted, however, in a more facile reduction of NiO than Pt and therefore alloying with Ru resulted in a higher activity to CO₂ methanation than with Pt. Note that the PtNi bimetallic alloys are proposed as active catalysts for the dry reforming of methane using CO₂ as oxidant [113], which may be considered for extensive studies for the direct reduction of CO₂. Overall, it can be concluded that RuNi and PtNi in general did not receive enough investigations and insofar are not the first choice for possible applications.

In general, we conclude that the alloying approach is very essential to efficiently enhance the catalytic properties of both Ni and Ru metals for the reduction of CO₂. The NiFe alloys in particular, which have been intensively investigated in recent years, seem very promising as a more active and stable alternative for commercial supported Ni catalysts for potential application. Nevertheless, a lot of progress is still needed to further lower the onset methanation temperature of these catalysts, which is a pertinent issue for future power-to-gas applications.

4. Support Effects on Different Metals

To discuss the influence of catalyst supports, we decided to focus on some selected examples both for the supported Ni and Ru catalysts, which attracted so far, the greatest attention for CO₂ reduction applications. It should be noted that the elucidation of the effect of this parameter (i.e., support effect) require the strict disentanglement from other influencing parameters like differences in metal particle size, specific surface area, metal loading, pretreatment, reaction conditions. In other words, the catalysts studied should be iso-structural. In this way, the differences in activity can be attributed solely to the chemical nature of catalyst support. We have summarized the TOF values in several studies for CO₂ reduction on different metals on different oxide supports together with their corresponding reaction conditions and structural aspects in

Table 1. Comparison of structural parameters of supported metal catalysts and their catalytic activity.

Catalyst	Metal Loading (wt.%)	BET/m2 g−1	Reaction Gas; Flow; GHSV	Metal Particle Size (nm)/Dispersion (%)	Selectivity (for CH4/for CO) ++	Rate or TOF (Temperature)	Ref.
Ru/Carbon	3	440	0.4% CO2/H2; 50 mL min−1	1.23/77	100/77	2.1 × 10−3 s−1 (240 °C) #	[114]
Ru/MgO	5	94	0.4% CO2/H2; 50 mL min−1	1.28/74	100/0	7.9 × 10−3 s−1 (240 °C) #	[114]
Ru/MgAl2O4	5	96	0.4% CO2/H2; 50 mL min−1	0.98/93	100/0	8.55 × 10−3 s−1 (240 °C) #	[114]
Ru/Al2O3	10	225	0.4% CO2/H2; 50 mL min−1	0.98/93	100/0	10.5 × 10−3 s−1 (240 °C) #	[114]
Ru/Al2O3	5	83	1% CO, 15% CO2, 50% H2; balance He; 200 mL min−1; 48,000 h−1	2.2/43	NR	110.7 μmolCO2 g−1 s−1 (330 °C) §	[115]
Ru/Al2O3	0.1	200	5% CO2, 15% H2; balance He; 60 mL min−1	NR/100	73/27	11 × 10−3 s−1 (300 °C)	[54]
Ru/Al2O3	2.2	63.3	15% CO2, 60% H2, 25% Ar 25%; 50 mL min−1	NR/15.2	100/0	0.69 × 10−3 s −1 (250 °C) #	[61]
Ru/Al2O3	2.3	127	15.5% CO2, 80.9% H2, balance N2; 41.6 mL min−1; 18,000 h−1	1.7/71	100/0	6.8 × 10−3 s−1 (190 °C) #	[67]
Ru/TiO2	5	42	1% CO, 15% CO2, 50% H2; balance He; 200 mL min−1; 48,000 h−1	4.5/21	NR	64.8 μmolCO2 gRu−1 s−1 (330 °C) §	[115]
Ru/TiO2	2.2	64–46	15.5% CO2, 80.9% H2, balance N2; 41.6 mL min−1; 18,000 h−1	1.6/66	100/0	26.2 × 10−3 s−1 (190 °C) #	[66]
Ru/CeO2	5	3.3	1% CO, 15% CO2, 50% H2; balance He; 200 mL min−1; 48,000 h−1	5.1/19	NR	5.21 μmolCO2 g−1 s−1 (330 °C) §	[115]
Ru/CeO2	2.3	34.2	15% CO2, 60% H2, 25% Ar 25%; 40 mL min−1	NR/35	100/0	0.71 × 10−3 s−1 (175 °C) #	[61]
Ru/ZrO2	2.1	114–64	15.5% CO2, 80.9% H2, balance N2; 41.6 mL min−1; 18,000 h−1	1.6/NR	100/0	23.3 × 10−3 s−1 (190 °C) #	[71]
Ru/SiO2	5	144	1% CO, 15% CO2, 50% H2; balance He; 200 mL min−1; 48,000 h−1	5.5/17	NR	27 μmolCO2 g−1 s−1 (330 °C) §	[115]
Ni/Al2O3	3	150	1% CO2, 4% H2, & balance N2; 3000–70,000 h−1	9.8/9.9	88/12	22 × 10−3 s−1 (277 °C) #	[116]
Ni/SiO2	3	200	1% CO2, 4% H2, balance N2; 3000–70,000 h−1	2.5/39	60/40	13 × 10−3 s−1 (277 °C) #	[116]
Ni/SiO2	5	8	20% CO2/H2; 16,000–66,000 h−1	5.3/19.1	100/0// 85/15	1.1 × 10−3 s−1 (200 °C)//2.5 × 10−3 s−1 (250 °C) #	[117]
Ni/TiO2	3	50	1% CO2, 4% H2, balance N2; 3000–70,000 h−1	13/7.5	99/1	47.3 × 10−3 s−1 (277 °C) #	[116]
Ni/TiO2	5	53	20% CO2/H2 (1:4); 16,000–66,000 h−1	9.4/10.8	100/0// 100/0	6.1 × 10−3 s−1 (200 °C)//11.5 × 10−3 s−1 (250 °C) #	[117]
Ni/CaO-Al2O3	23	170	3.75% CO2, 15% H2, He balance; 40 mL min−1; 15,000 h−1	8/9	100/0	19.6 × 10−3 s−1 (250 °C) #	[118]
Ni/CeO2	5	11	20% CO2/H2; 16,000–66,000 h−1	8.7/11.6	100/0// 100/0	24.4 × 10−3 s−1 (200 °C)//76.4 × 10−3 s−1 (250 °C)	[117]
Rh/Al2O3	5	100	20% CO2/H2; 3000–6000 h−1	NR/30.2	99/1	20.4 × 10−3 s−1 (200 °C) #	[119]
Rh/Al2O3	1	171	20% CO2/H2; 10 mL min−1; 6000 h−1	NR/18	93/7	220 × 10−3 s−1 (400 °C) #	[120]
Rh/SiO2	5	240	20% CO2/H2; 3000–6000 h−1	NR/22.8	99/1	4.4 × 10−3 s−1 (200 °C) #	[119]
Rh/TiO2	1	150	20% CO2/H2; 3000–6000 h−1	NR/22.3	99/1	150 × 10−3 s−1 (200 °C) #	[119]

^# refers to CH₄ formation; ^§: refers to CO₂ conversion rate; ⁺⁺: selectivity shown in the table are measured at the same temperature as TOF values; gas space velocity is not reported in all citations; NR: not reported.

. Thus, we want to help the reader to draw conclusions on this influencing parameter. For supported Ru catalysts, Kowalczyk et al. showed that Ru nanoparticles of similar average sizes/metal dispersion show significant variation of the CO₂ methanation activity in a 0.4% CO₂/H₂ gas mixture with the change of the nature of support [114]. Although Al₂O₃, MgO, Al₂MgO₄, and activated carbon (AC) are all non-reducible supports, there is clear differences in the Ru mass normalized reaction rate; activity measured at 240 °C increases in the order Ru/AC < Ru/MgO < Ru/MgAl₂O₄ < Ru/Al₂O₃ (see TOF value sin Table 1). This observation highlights that the chemical nature of the support, which does not differ much in reducibility, is enough to affect the catalytic activity.

Panagiotopolou et al. also studied the impact of the support material for Ru NPs on the activity for CO and CO₂ reduction (1% CO, 15% CO₂, 50%H₂, balance He) in the presence and in the absence of varying amounts of water vapor in the feed gases. By comparing similar metal dispersions (17–21%) for 5.0 wt.% Ru loading on different oxides, the activity for CO₂ reduction followed the order <Ru/CeO₂ < Ru/SiO₂ < Ru/TiO₂ < Ru/Al₂O₃ (see Table 1). These results show a higher activity of the Ru/Al₂O₃ catalyst in comparison with the reducible supports [115]. However, in this case, the catalysts were prepared with oxide supports having significant differences in the specific surface areas (<Ru/CeO₂: 3.3 m² g⁻¹; Ru/TiO₂ 42 m² g⁻¹; Ru/Al₂O₃: 83 m² g⁻¹; Ru/SiO₂: 144 m² g⁻¹). This parameter was recently shown to strongly affect the MSIs and as a consequence the activity of these catalysts for the methanation of carbon oxides [70,72].

Based on these findings, Behm and coworkers studied the effect of catalyst support by using iso-structural Ru/MO_x catalysts that were prepared by keeping the following parameters nearly similar for all catalysts: (i) specific surface area (110–120 m²g⁻¹), (ii) Ru loading (2.1 wt.% Ru loading), (iii) Ru NP size/Ru dispersion (1.5–1.8 nm/50–60% dispersion, (iv) catalyst pretreatment, and (v) reaction conditions [68]. For CO₂ methanation in 15% CO₂/H₂ gas mixture, the authors showed that the activity of Ru NPs supported on reducible oxides (TiO₂) [121] outperforms that for catalysts based on non-reducible (γ–Al₂O₃) [67] or weakly reducible supports (ZrO₂) [71]. The activity of these catalysts measured at 190 °C after temperature programmed reduction (TPR) in reaction gas between 190 and 350 °C followed the order Ru/TiO₂ (244 μmol g_Ru⁻¹ s⁻¹) [121] > Ru/ZrO₂ (140 μmol g_Ru⁻¹ s⁻¹) [71] > Ru/γ–Al₂O₃ (42 μmol g_Ru⁻¹ s⁻¹) [67]. In general, the reason for the enhancement of the activity of these catalysts after the reductive treatment is the modification of metal support interaction for ZrO₂ and TiO₂, and the change in the support basicity for γ–Al₂O₃, which helped to (i) enhance the electronic metal support interaction (EMSI) and charger transfer at Ru-TiO_2/Ru-ZrO₂ interface and (ii) increase the fraction of flatter Ru NPs on γ–Al₂O₃ support. These findings underline the pivotal role of the support material on the activity of Ru NPs for CO₂ reduction.

For supported nickel catalysts, the situation was about similar to those findings reported for supported Ru catalysts. Vance and Bartholomew showed that the activity for supported Ni catalysts, both in CO and CO₂ reduction, increases in the order of Ni/SiO₂ < Ni/Al₂O₃ < Ni/TiO₂ [116], i.e., with the increase of support reducibility (see Table 1). The selectivity of these catalysts for methane formation from CO₂ increased also in the same order. The authors attributed the observed increase in catalytic activity to an increase of the metal-support interactions (MSIs), which is highest for the Ni/TiO₂ catalysts. The origin of the enhancement of selectivity on these catalysts remains, however, an open question for future mechanistic studies.

In another study, Zawadzki and coworkers showed similar results for supported Ni catalysts, where activity decreased in the order Ni/CeO₂ > Ni/ZrO₂ > Ni/Al₂O₃ [122]. The authors employed H₂-temperature programmed reduction (H₂-TPR) measurements to evaluate the impact of support on the nature/reducibility of nickel nanoparticles, where a strong dependence on the support reducibility was correlated.

For detailed discussion about the parameters evolving with the variation of catalyst support and their impact on the CO₂ methanation on supported Ni catalysts, we refer to the review published by Shen et al. [123].

In contrast to findings reported on supported Ni catalysts, Pandey and Deo showed that the activity for CO₂ methanation on NiFe supported alloys is more pronounced for the non-reducible Al₂O₃ support compared to reducible oxides such as TiO₂ and Nb₂O₅. The activity toward CO₂ methanation at 250 °C follows the order NiFe/Al₂O₃ > NiFe/ZrO₂ > NiFe/TiO₂ > NiFe/SiO₂ > NiF/Nb₂O₅ (see Figure 6) [124]. Based on their results, all prepared catalysts showed comparable (i) specific surface area, (ii) degree of reduction, and (iii) crystallite sizes of the support or NiFe alloys. This led to the conclusion that the difference in activity is merely related to the nature of the support. The authors explained the enhancement of activity by improved CO₂ adsorption on NiFe/Al₂O₃ catalysts compared to other catalysts (NiFe/TiO₂, NiFe/SiO₂, NiF/Nb₂O₅). In comparison with previous observations on supported Ni and also supported Ru nanoparticles, it seems that there is special synergism between these NiFe alloys and the Al₂O₃ support leading to unexpected CO₂ methanation activity. The authors related this behavior to the improvement of CO₂ adsorption. We believe that other reasons should also be studied to explain this behavior such as the support basicity, concentration of OH groups, and adsorption strength of reaction products.

5. Nonconventional Approaches to Switch CO₂ Reduction Selectivity

In this section, we will discuss in detail some of the new approaches that have been recently reported by some groups for switching the selectivity of CO₂ reduction from methanation pathway to the CO formation via RWGS reaction. These involve the development of single atom catalysts (SCAs) and the modulation of the metal support interactions by recently emerging approaches.

5.1. Heterogeneous Single Atom Catalysts (SACs) Supported on Metal Oxides for CO₂ Reduction

The miniaturization of the size of active metal nanoparticles (NPs) in heterogeneous catalysts to their atomic limits, ultimately to the so-called single-atom catalysts (SACs), is emerging as an attractive approach for engineering new catalytic materials [125,126]. This is motivated on the one hand by the need to maximize the fraction of active sites on the surface of catalyst supports and therefore reduce the price of expensive metals used in catalytic applications. On the other hand, this approach is adopted to enhance the selectivity of metal active sites in specific reaction pathways. The latter aspect, in particular, is emerging recently as an approach to bridge heterogeneous catalysis with molecular homogeneous catalysis and is anticipated to help engineering unprecedented properties of heterogeneous catalysts (for further discussion on this aspect we refer to Ref. [127]). The use of SACs in the reduction of CO₂ emerged also recently as an attractive approach to engineer the unique properties of different metals such as Rh [90,91], Ru [59], and Pt [128], which by default are known as the most active catalysts for CO₂ hydrogenation/reduction reactions. Isolated nickel single atom catalysts were also prepared in stable forms and were active for the photo- and electrocatalytic reduction of CO₂ [129,130]. The stability of nickel single sites under thermo-catalytic reaction in CO/CO₂/H₂ gases is quite challenging due to the much higher tendency of these isolated sites to form Ni(CO)₄, at low temperatures. Therefore, much progress in the design of catalyst support is still needed for the architecture of such catalysts. Hence, we will focus on some key examples for Ru, Rh, and Pt where the atomic dispersion of the metals, i.e., as SACs result in the change of catalytic performance.

5.1.1. Rh-Single-Atom Catalysts

Recent work of Matsubu et al. on supported Rh/TiO₂ catalysts has shown that the dispersion of Rh nanoparticles/clusters to their atomic limits results in a drastic change of the selectivity of this element for the catalytic CO₂ reduction at ambient pressure and 200 °C from CH₄ to CO formation [90]. By leaching the Rh nanoparticles on a series of Rh/TiO₂ catalysts using H₂O₂/HCl mixture and correlating the activity of resulting catalysts with the fraction of Rh single sites on TiO₂, these authors found out a strong correlation between the activity toward CO formation with the fraction of isolated Rh single sites. In this study, the authors showed that the temperature onset of CO formation (CH₄ formation) is strongly affected by the ratio of CO₂:H₂ in the reaction gas mixture, where methane formation is promoted with the increase of the H₂/CO₂ ratio [90]. Further studies of Rh single sites on TiO₂ showed that the activity and selectivity of these atomic species are strongly dependent on the local surface environment, which is strongly dependent on the pretreatment (gas atmosphere/temperature) and the reaction gas mixture used in CO₂ reduction [131]. Moreover, the isolated rhodium single sites supported by porphyrin metal-organic framework showed high activity at room temperature during the photocatalytic reduction of CO₂ with high selectivity toward formate-like species [132]. Interestingly, these isolated single sites did not show any activity toward methane formation.

5.1.2. Ru-Single-Atom Catalysts

For supported Ru catalysts, it was recently also reported by Aitbekova and coworkers that Ru nanoparticles on CeO₂ can be dispersed to Ru single atoms upon exposure to an oxidative treatment at relatively low temperature (230 °C) [59]. Notably, upon transition from Ru nanoparticles to Ru single atoms the activity is switching from CO₂ methanation to CO formation via RWGS. Furthermore, the exposure to a sequence of low temperature oxidation treatment followed by a reduction step at 230 or 530 °C with much higher selectivity for CO could be seen compared to only exposure to oxidative treatment at 230 °C before the catalytic reaction. This behavior is strongly analogous to the behavior of Rh single sites supported on TiO₂, which are also selective to CO formation. Similar results were also observed when Al₂O₃ and TiO₂ were used as supports for Ru, although the response of TiO₂ to the pretreatments is more dependent on the temperature than for CeO₂ (see Figure 7).

The long-term stability of these species with time-on-stream is, however, questionable. The quick reduction of RuO_x species during reaction in CO/CO₂/H₂ feed gases have been studied with time-resolved XANES/EXAFS measurements for different supports including zeolites [76,77], Al₂O₃ [67,77], ZrO₂ [71], and TiO₂ [72,133]. Based on these findings, it has been established that the tiny RuO_x species on different oxides can be quickly reduced during the first few minutes of reaction and grow further to form larger NPs. The persistence of small fraction of oxidic Ru species cannot, however, be excluded. Therefore, a clear judgement of the stability of these single-atom catalysts needs elaborate deactivation experiments with time-on-stream at different temperatures. This would include also the examination of the impact of catalyst pretreatment as well as the catalyst preparation method and its structural characteristics, which are decisive for these structural changes.

In contrast to the findings of Aitbekova et al. [59], a recent study by Fan et al., depicted that atomically dispersed Ru into porous hexagonal boron nitride supports shows high selectivity toward methanation of CO₂ with extended time-on-stream stability. These catalysts are surprisingly more active than supported Ru nanoparticles [134]. The stability of these catalysts during reaction was attributed to the coordination of Ru by B and N atoms into defects sites of these 2D materials. It should be noted that the coordination with B and N also reduces the valency of Ru (Ru is more reducible), which presumably increases the local charge density needed to activate hydrogen. These observations underline the impact of the support material on the properties of SACs. Therefore, it cannot be accepted as a consensus that the SACs of the metals Rh or Ru should only be active for the RWGS.

Stable Ru-SAC catalysts could also be prepared using mesoporous silica supports. As an example, stable Ru-SACs were prepared by interaction of Ru³⁺ ions and CTA+ surfactant species inside the micro-cages of the mesoporous MCM-41 silica. This Ru SAC was found active and selective for the water-free reduction of CO₂ to formic acid by interaction with molecular hydrogen. Any formation of CO or CH₄ was not monitored leading to 100% selectivity toward formic acid [135]. The presence of surfactant into the micropores of MCM-41 was found to stabilize Ru³⁺ ions against singeing in the presence of hydrogen and has not made them inactive. Similarly, mononuclear Ru³⁺ sites coordinated with a N, P containing polymer were found active for CO₂ reduction to formats [136]. This study showed that the stabilization of the mononuclear Ru single sites by nitrogen-donor groups (phosphazene-N and pyridine-N) leads to an increase of the local electron density on these sites and therefore promoted H₂ dissociation and subsequently led to an enhanced catalytic activity.

5.1.3. Pt-Single-Atom Catalysts

For supported Pt catalysts, Wang et al. recently showed a comparison of the behavior of CeO₂-supported Pt nanoparticles and CeO₂-supported Pt single atoms for the thermocatalytic reduction of CO₂ at ambient pressure [128]. These authors prepared Pt SACs by the impregnation of a CeO₂ support but with limited concentration of Pt (0.05 wt.%) [128]. For comparison, 2 wt.% of Pt nanoparticles were loaded onto CeO₂. In both cases, the catalysts were post-treated by calcination in air at 450 °C. Catalytic performance of both catalysts was examined between 150 and 500 °C and results indicated that Pt-SACs can be entirely selective for CO formation. However, the supported Pt NPs require higher reaction temperature to show this feature since they start to switch selectivity from CH₄ to CO at temperatures ≥300 °C (see Figure 8a). The total reaction rate for CO₂ conversion normalized by the Pt mass is higher for the Pt-SCA compared to the Pt-NP catalyst (see Figure 8b). This is an impressive example of how SACs can be used to selectively control reactions with a significantly reduced amount of precious metal. In contrast, supported Pt single atoms by triazine covalent organic framework (COF) were found highly active to methanation under visible light illumination [137]. These results confirm again the concept that the activity of single atoms toward a specific product is strongly dependent on the support material, reaction conditions, and regime of catalysts used (e.g., photocatalysis versus thermocatalytic reactions). Interestingly, the activity of Pt-supported triazine COF catalysts is significantly lower than that detected on the Pt SAC catalyst [137]. These aspects should be examined more in depth in the future, in particular the effect of support material on the activity and selectivity toward a specific product. There seems to be a common pattern for different metals as discussed above, especially for the Pt and Ru SACs.

Overall, it is clear that most of the methanation active metals supported on typical metal oxides can be converted to RWGS-active catalyst by transforming the metal nanoparticles into atomically dispersed species on oxide supports. This cannot, however, be considered a general rule of thumb. Exceptions include metal single atoms supported by nonconventional supports such as carbon nitrides, and covalent organic frameworks, which were active toward CO₂ methanation rather than CO formation, in contrast to what had been observed for CeO₂ or TiO₂ supports. Deep characterization of these species under reaction conditions and clear understanding of the coordination environments of these supports are indispensable for easy control of their selectivity for CO₂ reduction.

5.2. Controlling the Metal-Support Interaction and Impact on CO₂ Reduction Selectivity

The activity of different catalytic materials toward CO₂ reduction are much influenced by the chemical nature of support, which in turn affects the metal-support-interactions (MSIs). This phenomena refers commonly to the classical strong metal-support-interaction effects (SMSI) reported in the late seventies and early eighties by Tauster et al. [138,139]. According to their model, SMSI was defined by the overgrowth of metal nanoparticles, in particular Pt and Ru, by partially reduced metal oxide species (i.e., partial encapsulation of metal NPs by reduced metal oxide support) at relatively high temperatures (> 300 °C). This results in the shrinkage of the accessible active metal surface area. This effect was reported recently to affect the activity of Ru/TiO₂ catalysts for both CO and CO₂ methanation [66]. Switching the selectivity for CO₂ reduction between CO formation via RWGS or to CH₄ formation via the Sabatier reaction is even more interesting aspect connected to changes in the MSIs. In the coming few paragraphs, we will show two recently reported key examples on the role of metal-support interactions in tuning the activity of supported metal nanoparticles for CO₂ reduction.

The first example was reported by Matsubu et al. for supported Rh catalysts [91]. These authors found out that the selectivity of CO₂ reduction at ambient pressure can be reversibly switched from methane-selective to CO-selective pathways by modulating the electronic properties of the catalysts (i.e., its tendency for O-vacancy formation). This approach was achieved by inducing a special kind of metal support interaction, conveniently referred to as “adsorbate-mediated strong metal-support interaction” (A-SMSI). It involves the deposition of reactant permeable CH_x adlayer on the Rh nanoparticles by treatment of the catalyst in 20% CO₂/2% H₂ gas mixture at 250 °C, which results in the decay of the selectivity for methanation down to <15% compared to fully reduced catalyst, which exhibits about 100% selectivity for methane. The change in selectivity is the outcome of decay of CH₄ formation rate and the concomitant increase of the rate of formation of CO, due to a change of the intrinsic activity of these catalysts toward CH₄ and CO formation. In comparison with the standard catalyst reduced in hydrogen, these A-SMSI effects resulted in a strong modification of the electronic/adsorption properties of Rh NPs, which was deciphered by observing a strong red-shift of the CO_ad-Rh signal as a consequence of inducing A-SMSI compared to hydrogen-reduced sample. These observations were monitored for Rh/TiO₂ and Ru/Ta₂O₅ supported catalysts, while other irreducible supports such as Rh/Al₂O₃ did not show similar results. Despite CeO₂ being a highly reducible oxide and having higher tendency of O-vacancy formation than other oxides such as TiO₂, the Rh/CeO₂ did not show similar behavior compared to Rh/TiO₂ and Ru/Nb₂O₅. This point opens the door for further studies to identify the dependence of these A-SMSI on the chemical nature of the support, and not only its reducibility. In total, we believe that this approach is a very agile route for modification of the reactivity and selectivity of CO₂ reduction catalysts and further studies on other metals are also needed to optimize its use for the control of the selectivity for CO₂ reduction.

The other example of modification of the selectivity of CO₂ reduction in CO/CO₂/H₂ and CO₂/H₂ gas mixtures (via modulation of MSI was reported recently by Abdel-Mageed et al. [70,121]). These authors found out that the systematic variation of the specific surface area of the TiO₂ supports of Ru/TiO₂ catalysts in the range from 20 to 230 m²g⁻¹, which correlate to a change in the TiO₂ particle size from about 24 to 9 nm, results in a strong variation of the activity toward methane and CO formation at ambient pressure in the temperature range between 190 and 350 °C, and therefore can be used to control the selectivity for CO₂ reduction.

For Ru/TiO₂ catalysts with specific surface < 120 m²g⁻¹, the catalysts showed almost 100% selectivity for methane in the temperature range from 190 to 350 °C [66], while catalysts with larger specific surface area (SSA; >200 m²g⁻¹) showed initially a limited methanation activity and no CO formation in the temperature range from 190 to 250 °C (see Figure 9a–d) [70,121]. At T ≥ 250 °C, the catalyst became inactive for methane formation and instead showed only CO formation (CO selectivity of 100%) with the reaction rate increasing continuously with rising the temperature up to 350 °C. These high SSA Ru/TiO₂ catalysts remained CO selective upon returning back to 190 °C after reaction at 350 °C (see Figure 9f), while the lower SSA Ru/TiO₂ catalysts showed only methane, but with higher reaction rate (see Figure 9e). These observations were explained by a strong modulation of the metal-support interactions with the change in the support TiO₂ particle size. With increasing the SSA (decreasing TiO₂ particle size), it could be observed that the concentration of O-vacancy defects increases significantly.

In their model, the authors claimed that a moderate reduction of the support enhances the CO₂ methanation activity to some extent, while over-reduction, which is associated with small TiO₂ crystallites (large SSA), changes the selectivity from 100% methanation to 100% CO formation via the RWGS reaction. Note that on the high SSA Ru/TiO₂ catalysts large fraction of isolated Ru³⁺ ions were detected, which may match in part with findings reported on the switch of selectivity from CH₄ to CO upon dispersion of Ru NPs into isolated Ru sites on CeO₂ supports as demonstrated in the findings by Aitbekova et al. [59].

One challenge to reconcile this postulate is the persistence of a large fraction of the Ru NPs on these Ru/TiO₂ catalysts after the high temperature TPR in addition to the different nature of CeO₂ used as support by Aitbekova et al., compared to TiO₂ oxides with large surface area. This apparent contradiction between these two studies would also be explained by the presence of Ru single sites in both cases in a different electronic structure (valency), which would definitely affect their activity.

In total, these two approaches (atomic dispersion of metals and modulation of MSI) allow for facile control of the selectivity for CO₂ reduction either to syngas or to methane. In several cases, in particular for the high surface area Ru/TiO₂ catalysts (BET > 200 m² g⁻¹), the catalyst can activate CO₂ to syngas at much lower reaction temperatures than those used on the commercial Ni catalysts. This is needed to save energy and also enhances the safety measures of the whole process.

6. Outlook, Challenges, and Requirements for Power-to-Gas Applications

As outlined in detail in the introduction, the ambient pressure CO₂ reduction can be envisaged as one of the rings in the chain of renewable energy, either for syngas production or for the production of synthetic natural gas. The implementation of this process would thus rely on production of highly active and selective catalysts on the one hand, and on the other hand high stability of these catalysts toward the fluctuation in the renewable energy supply, i.e., catalyst adaptability toward dynamic operations [78,79]. This kind of operation would include specifically the following incidents/circumstance that may affect the catalyst performance and its lifetime.

(i)

reactor abrupt shutdowns due to loss of power, which leads to a rapid reduction in reactor temperature, and thus leads to the condensation of water vapor, existing as a byproduct of either syngas or methanation pathways.

(ii)

hydrogen shortage during periods of low electricity production, which is intimately coupled with seasonal changes in temperature or speed of wind. This would lead to the decrease of the H₂/CO₂ ratios.

(iii)

considering that the anticipated supply of carbon dioxide would rely on the harvesting from air or from the outlet of industrial off-gases, any fluctuations in these two supplies would cause similar problems for the performance and stability of the catalyst. In this way, the reactor would operate instead at high H₂/CO₂ ratios.

(iv)

considering the collection of carbon dioxide from air the presence of some impurities of oxygen and water are highly possible in the CO₂ feed, which may also affect the stability of these catalysts.

(v)

any loss of selectivity toward either syngas or methane would indispensably complicate the practical applications, especially for small scale uses, e.g., for methanation reactors applied for domestic uses (e.g., house heating or fueling farms), the formation of syngas as a byproduct would impost extra costs for elimination and may also result in safety risks.

These aspects can occur especially in the decentralized small-scale plants, which are in the center of the use of the renewable energy cycle. Therefore, catalysts intended for these two processes should fulfil certain criteria and be optimized and investigated under conditions relevant to the anticipated use. We do not need only highly active catalyst, but more importantly they have to be strictly selective toward one of the products and highly stable under dynamic changes in the temperature, reaction gas composition (e.g., CO₂/H₂ ratios), and tolerant toward the leak of small amounts of oxygen and water condensation during reaction. Based on this overview and considering the strengths and weaknesses of the different catalytic systems discussed in previous sections, we would like to refer to some selected examples that would have technical potential for methane and syngas production.

6.1. Candidates for CO₂ to Methanation

Among the different catalytic systems described in previous sections, we see two important examples that can serve as excellent methanation catalysts in the gas-to-power applications. The first example is Ru nanoparticles supported on metal oxides, in particular the reducible oxides, which exhibit high selectivity toward methane and high activity at rather low reaction temperatures (<200 °C). These catalysts can be suitable for practical uses in small scale applications to convert hydrogen and carbon dioxide to methane, and they are highly stable under dynamic operation conditions as discussed before [78,79]. One of the challenges of using Ru/MO_x is the relatively high price of Ru. This would need in the long term the optimization of smart synthetic methods using limited amounts of Ru nanoclusters or alloying Ru with cheaper metals. The second example is the NiFe catalysts, which turned out recently as very promising selective CO₂ methanation catalysts. These catalysts showed very high stability and rather high activity at already lower temperature range (200–400 °C) than that used for the commercial NiAlO_x catalysts. Although it is still much less active than the supported Ru catalysts, the NiFe alloys are a much cheaper choice and the tradeoff between price, stability, and high selectivity makes them an excellent option, especially for large scale production of methane, e.g., in industrial complexes where huge amounts of CO₂ can be easily collected and diverted into methane. Extensive studies of stability under dynamic conditions including changes in reactor load and other parameters, which may affect their structural stability on stream, are still needed.

6.2. Candidates for CO₂ to Syngas

For the production of syngas, one would first refer to metal single atoms supported on oxides. Successful examples such as Pt1/CeO₂ [29], Ru1/CeO₂ [59], and Rh1/TiO₂ showed rather high selectivity toward CO at relatively moderate temperatures (≤250 °C). It should be noted that all these SAC catalysts are based on reducible oxide supports. The second candidate is the supported Ru on high specific surface area TiO₂ supports (>200 m²g⁻¹), which showed high activity and up to 100% selectivity toward CO formation at rather low temperatures (190 °C) after activation at moderate temperatures (190–350 °C). The long-term stability of these catalysts, especially under dynamic conditions remains unclear and further studies are needed before they are implemented in real applications.

7. Conclusions

In total, the increasing demand in energy, nowadays mainly produced from burning hydrocarbons and coal, which goes along with the enormous increase in CO₂ emissions, requires strict and well-defined solutions and technologies for the production, storage, distribution, and sustainable use of green hydrogen, entirely generated from renewable resources. The reduction of CO₂ serves as an immediate approach for the conversion of the green hydrogen, which can be used in the production of (i) methane that can be easily stored and distributed in the gas grid and (ii) syngas that can be used to produce a wide range of chemical commodities such as olefins and liquid fuels (via FT synthesis). This is feasible in the near future due to the well-established catalytic systems for these processes. This target needs, however, optimization of running conditions encountered in realistic operations involving dynamic changes of reactant supplies, which is coupled with the natural fluctuation in electric power generation from renewables. These aspects would need not only excellent catalytic materials, but also suitable engineering solutions.