Recent Advances in Heterogeneous Catalytic Hydrogenation of CO2 to Methane

With the accelerating industrialization, urbanization process, and continuously upgrading of consumption structures, the CO2 from combustion of coal, oil, natural gas, and other hydrocarbon fuels is unbelievably increased over the past decade. As an important carbon resource, CO2 gained more and more attention because of its converting properties to lower hydrocarbon, such as methane, methanol, and formic acid. Among them, CO2 methanation is considered to be an extremely efficient method due to its high CO2 conversion and CH4 selectivity. However, the CO2 methanation process requires high reaction temperatures (300–400°C), which limits the theoretical yield of methane. Thus, it is desirable to find a new strategy for the efficient conversion of CO2 to methane at relatively low reaction temperature, and the key issue is using the catalysts in the process. The advances in the noble metal catalysts, Ni-based catalysts, and Co-based catalysts, for catalytic hydrogenation CO2 to methane are reviewed in this paper, and the effects of the supports and the addition of second metal on CO2 methanation as well as the reaction mechanisms are focused.


Introduction
Over the past centuries, CO 2 has become the main carbon resource due to the decreases of limited resources such as coal, oil, and natural gas [1]. However, the CO 2 concentration in the atmosphere has consequently increased from ~280 ppm (preindustrial) to ~390 ppm in 2010 at a rate of ca. 1% per year [2], which arguably contributes to the "greenhouse effect," and increases the global temperatures and climate change. CO 2 emissions are still existing threat to humans; it is high time that effective measures should be taken to decrease the emission of CO 2 .
Hence, the carbon capture and sequestration (CCS) system is considered to be an efficient method for CO 2 utilization [3,4]. Nevertheless, the hydrogenation reaction is the most important chemical conversions of CO 2 ; moreover, which offers a good opportunity for sustainable development in the energy and environmental sectors. Indeed, the reaction process not only reduces the CO 2 amount in the atmosphere but also produces fuels and valuable chemicals [5].
As a promising fuel energy, methane, a simple hydrocarbon, has a wide range of applications in the industry and civil use, which also used to produce some downstream products, such as ethyne, hydrogen, and ammonia [6,7]; therefore, the strategy of CO 2 methanation is significantly meaningful. Undeniably, the resources of fossil fuels are diminishing and fuel prices have undergone strong fluctuation in recent years. Therefore, developing alternative fuels from nonfossil fuel sources and processes are highly desirable. The products from CO 2 hydrogenation, such as methane, hydrocarbons, methanol, and dimethyl ether, are excellent fuels in internal combustion engines, and are easily stored and transported, but the literature studies showed that the CO 2 conversion to methanol and dimethyl ether is still very low (~20%) and it is difficult to achieve higher conversion of CO 2 [8,9]. CO 2 methanation is a simple reaction, generating methane under atmospheric pressure with several advantages over other chemicals. Although the conversion was still very low, the CH 4 formation from CO 2 at low temperature has become an important breakthrough in the utilization of CO 2 [10]. CO 2 methanation is a significant catalytic hydrogenation process, as is shown in Eq. (1).
The methanation of CO 2 has a wide range of applications including the production of syngas and the formation of compressed natural gas [1]. A prototype CO 2 recycling plant to supply clean energy preventing global warming has been built in 1996 using these key materials and has been operating successfully [11]. Without doubt, CO 2 methanation is the key pathway for CO 2 recycling, which requires a catalyst to achieve acceptable rates and selectivities. And extensive studies have been conducted on metal-based catalytic systems in the hydrogenation of CO 2 to methane.
Noble metals (e.g., Ru, Rh, Pd) supported on oxide supports (e.g., TiO 2 , Al 2 O 3 , CeO 2 ) were the most effective catalysts for CO 2 methanation under relatively mild operating conditions [12][13][14]; however, the high cost of the catalysts limited their practical applications [15]. Therefore, to obtain a feasible and cost-effective catalytic process, nonnoble metal catalysts (e.g., Ni, Co) were focused by many scholars [16,17]. This review attempts to present the catalytic reactivity and reaction mechanism over the catalysts, particularly over the heterogeneous catalysts with an emphasis on the effects of supports and the second metal additives, as well as an overview regarding the challenges and opportunities for future research in the field.

Role of the support on catalyst activity
CO 2 methanation has been studied over a series of supported Ru and Rh catalysts, which were very active for CO 2 hydrogenation [13,14,20,21]. The supports, including Al 2 O 3 , TiO 2 , and CeO 2 for these active metals, have also been investigated. To clarify the influences of the supports on the catalytic behavior of ruthenium, a FT-IR study is used to obtain more insight into the reaction mechanism [21]. Based on the FT-IR spectra of CO and CO 2 adsorbed on the catalysts, the improvement in the CO 2 methanation activity was related to a higher positive polarization of ruthenium on the zeolite, which led to a weaker Ru−CO bond on the H-ZSM-5supported sample with a corresponding increase of the hydrogen surface coverage, which favors the transformation of the intermediate CO to methane, and which indicated that Ru/ ZSM-5 exhibits more CH 4 selectivity than Ru/SiO 2 [21].
The Ru dispersion was significantly influenced by the crystal phase structure of the TiO 2 supports [19]. Rutile-type TiO 2 (r-TiO 2 ) was a much better support than anatase-TiO 2 (a-TiO 2 ) in stabilizing of RuO 2 due to the interfacial lattice matching, resulting in a higher reactivity and stability in CO 2 methanation. Owing to the highly dispersed Ru catalyst with a narrow size distribution, r-TiO 2 was a promising support [12]. There was a strong interaction between RuO 2 and r-TiO 2 during the calcination process, which prohibited the aggregation of RuO 2 in the presence of the Ru-O-Ti bond. As represented in Figure 1, upon calcination at 300°C, the Ru/r-TiO 2 exhibited a much higher activity and thermal stability in CO 2 methanation than Ru/ a-TiO 2 . Moreover, the reaction rate of the Ru/r-TiO 2 was 2.4 times higher than that of the Ru/a-TiO 2 , which mainly originated from the different particle sizes of ruthenium [12].
The effects of reaction temperature on the CO 2 conversion over the Ru catalysts and (b) the specific rates of CO 2 conversion calculated at 225°C. The feed gas was 18 vol.% CO 2 + 72 vol.% H 2 + 10 vol.% N 2 , and the catalyst was each 0.040 g of Ru/TiO 2 diluted with 0.400 g of SiO 2 , the total space velocity was 75,000 mL·g cat −1 ·h −1 [12].
The Ru/TiO 2 catalysts were prepared via a spray reaction (SPR) [20,22], and the catalytic CO 2 hydrogenation activities of the SPR fine particles were much higher than those of impregnation catalysts [20]. The high activity of the SPR catalysts was attributed to the occurrence of new active sites at the metal-support perimeters without any strong metal-support interaction phenomenon. In addition, highly dispersed Ru nanoparticle-loaded TiO 2 was prepared using a "dry" modification method [18], which markedly enhances the performance of low-temperature methanation, achieving a 100% yield at 160°C. In addition, the methanation reaction over Ru/TiO 2 proceeded at temperatures as low as room temperature with a reaction rate of 0.04 mmol·min −1 ·g −1 .
Although Ru catalysts deposited on different supports, such as alumina, titanium, or silica, have been extensively studied, and the effect of the support on the catalytic properties of small Ru particles in CO 2 hydrogenation has not been fully recognized. Different supports (low and high surface area graphitized carbons, magnesia, alumina and a magnesium-aluminum spinel) were used in CO 2 methanation, and alumina was found to be the most advantageous material [23]. The catalytic properties of very small ruthenium particles are strongly affected by metalsupport interactions. In the case of Ru/C, the carbon support partly covers the metal surface, lowering the number of active sites (site blocking). A sequence of the surface-based activities (TOF): Ru/Al 2 O 3 > Ru/MgAl 2 O 4 > Ru/MgO > Ru/C is almost identical to that of electrondeficiencies of the metal, determined by the Lewis acidities of the supports [23].

Effect of metal loading
The most likely effects caused by increasing the loading amount are the growth of the particle size, e.g., the mean particle size of surface Rh species increased with the metal loading amount, which affected the reactivity [24]. From the study over Rh/γ-Al 2 O 3 , varying Rh amounts show Rh particle sizes of 3.6-15.4 nm, and a 100% methane selectivity was observed over the entire temperature range and Rh amounts, and the turnover frequency for CH 4 formation depended on the Rh particle size. Larger Rh particles exhibited a catalytic activity of up to four times higher than the smaller particles at 135-150°C, whereas at higher temperatures (200°C) the turnover frequencies are similar for all particle sizes [13].
The Rh loading amount can significantly change the product selectivity of CO 2 hydrogenation over Rh/SiO 2 [25], and the main products transformed from CO 2 to CH 4 with the loading amount of Rh, as shown in Figure 2. To the 1 wt% Rh/SiO 2 catalyst, the concentration of surface Rh particles was low, and the Rh species were surrounded by the hydroxyl groups of SiO 2 . For the 10 wt% Rh/SiO 2 , 5.8 times more surface Rh particles than that of 1 wt% Rh/SiO 2 were found with accordingly less surface hydroxyl groups of SiO 2 existed around Rh particles [25]. In the Ru/Al 2 O 3 catalysts with a Ru amount of 0.1-5.0%, the CH 4 selectivity in CO 2 methanation increased with the increase in the Ru loading amount [26]. In the 0.1% Ru/Al 2 O 3 catalyst, Ru is mostly present in the atomic dispersion, and the agglomeration of small metal particles (and atoms) in the 3D clusters was observed, indicating a decrease in CH 4 selectivity.

Effect of second metal
Actually, when the alkaline salts were added to Ru/Al 2 O 3 catalysts, a synergetic effect can be detected, including the electron donation of an alkaline promoter modified the local electron density of the Ru metal, the formation of alkaline chlorides to neutralize the residual chlorine ions, and the removal of the depositional inactive carbon, which was formed on the catalyst surface during CO 2 hydrogenation [22]. Tests of the Ba-and K-containing Rh/Al 2 O 3 and the pure Rh/Al 2 O 3 in 300-700°C revealed remarkable differences in the cata-lytic behavior (Figure 3). The Ba-containing and especially the pure Rh/Al 2 O 3 catalyst showed high selectivity to CH 4 below 500°C with a maximum CH 4 yield of 60% at 400°C; however, at higher temperatures, the CO formation became significant. K-containing Rh/ Al 2 O 3 converted CO 2 only to CO in 300-700°C and no CH 4 was found. A vastly different adsorption behavior of the Ba-and K-containing catalysts and a significant influence of these additives on the Rh(0)/Rh(I) ratio were revealed [27].

Effect of supports 2.2.1.1. Enhancement of catalytic performance
Choosing a suitable support is mostly according to its properties to activate CO 2 and the interaction between the metal and supports, which is a key parameter for the methanation reaction [28]. The structure and properties of the support do affect the dispersity of active metals and the stability, which enhance the activity of catalysts.
Currently, various materials are used as the supports for nickel catalysts, such as γ-Al 2 O 3 [29][30][31], SiO 2 [32,33], Ce x Zr 1−x O 2 [33][34][35][36], and TiO 2 [37]. Because the support has a significant influence on the morphology of the active phase, adsorption, and catalytic properties [38], Ni was supported on the mesostructured silica nanoparticles (MSNs), MCM-41, HY zeolite, SiO 2 , and γ-Al 2 O 3. And the CO 2 methanation activity followed in the order of Ni/MSN > Ni/MCM-41 > Ni/HY > Ni/SiO 2 > Ni/γ-Al 2 O 3 [32]. The high activity of Ni/MSN is due to the presence of both intraparticle and interparticle porosities, which led to a high concentration of basic sites. In addition, the defect sites or oxygen vacancies in MSNs were responsible for the formation of surface carbon species, while Ni sites dissociated hydrogen to form atomic hydrogen.
An encouraging result was found in the CO methanation reaction over the zeolite supports, and the same results also found in the Ru/Y and Ru/Al 2 O 3 catalysts [39], as well as the supporting Pd on the zeolites, and the catalytic activity on the supporter was in the order of HY > HZSM-5 > NaZSM-5 > NaY > SiO 2 [40]. Similarly, when CO 2 hydrogenation to methane was carried out over nickel species supported on a HNaUSY zeolite, interesting CO 2 conversions and CH 4 selectivities were achieved. CO 2 conversion increased with the Ni content from 2 to 14%, due to the higher amount of Ni 0 species after reduction [41]. Nickel particles were grafted onto SBA-15, and a chemical bond was formed between Ni and Si by O, and no bulk nickel oxides existed in the Ni-grafted SBA-15 [42]. Therefore, the Ni-grafted SBA-15 suited CO 2 methanation, resulting in the higher CO 2 conversion (TOF of 19.4 s −1 ) and methane selectivity (92%) than a NiO dispersed SBA-15. The status of catalytic systems for the synthesis of methane by CO 2 hydrogenation is summarized in

Nickel dispersion
As a highly active catalyst for CO 2 methanation, a highly uniform dispersed active species over the support is required; therefore, a high specific surface area support is needed. In general, the support usually plays a very important role in the interaction between the Ni and the support. The nickel compounds on different support surfaces result in different "metal-support effects" [30], which implies that catalysts would exhibit different performance toward activity and selectivity for a given process.
Ni/Al 2 O 3 with a high specific surface area showed an excellent controllability on the specific surface area of catalysts with the increase in the Ni amount, and increased the reducibility of the catalyst. However, a further increase in the Ni amount would cause a decrease in CO 2 conversion due to the bigger crystallite size and lower surface area of the catalyst [29,30]. Indeed, the CO 2 conversion and CH 4 yield are strongly dependent on the Ni amount and the calcination temperature. Compared with the no pretreatment catalysts, the prereduced 16% Ni catalyst obtained 100% CH 4 selectivity with no CO detected [47]. With a higher calcination temperature, the metal nickel is in the form of NiAl 2 O 4 , which is an inactive phase for methanation [47,48]. The existential state of Ni is usually affected by the support. Cubic metallic Ni particles are found mostly without carbon whiskers, and fast methanation occurs at the expense of the CO intermediate on the corners of nanoparticles interacting with Al 2 O 3 [43].
The Ni-based catalyst prepared by coprecipitation is active for CO 2 methanation as well. Coprecipitated Ni/Al 2 O 3 catalysts are found to be efficient promoters for CO 2 methanation, and Al 2 O 3 is active for CO 2 adsorption [49]. A Ni-Al hydrotalcite-derived catalyst (Ni-Al 2 O 3 -HT) was prepared by a coprecipitation method with a narrow Ni particle-size distribution and an average particle size of 4.0 nm, a large number of Ni nanoparticles were surrounded by amorphous alumina [31]. As for the Ni amount up to 78 wt%, the average crystalline size of Ni was only 4 nm with a narrow distribution in the range of 3-9 nm. Compared with the 78 wt % Ni/Al 2 O 3 catalyst using an impregnation method, the Ni-Al hydrotalcite-derived catalyst exhibited a much higher Ni dispersion than its impregnated counterpart, indicating that Ni-Al hydrotalcite is an ideal precursor for preparation of a well-dispersed Ni catalyst.
Recently, a surface defect-promoted Ni nanocatalyst with a high dispersion and high particle density embedded on a hierarchical Al 2 O 3 matrix exhibits excellent activity and stability simultaneously for CO 2 methanation. The abundant surface vacancy clusters serve as the active sites, accounting for the significantly enhanced low-temperature activity of the supported Ni nanoparticles [43]. Ni/H−Al 2 O 3 (400) clearly possesses a significantly enhanced low-temperature activity for CO 2 methanation. The CO 2 conversion exceeded 90% at 265°C and reached the maximal value of 99% at 300°C (Figure 4A). The methane production rate increased along with the Ni surface area, indicating a strong correlation between the activity and the Ni surface area. The TOF value as a function of Ni dispersion for the three samples ( Figure 4B) [43].
The different Ni loading amount over the Ni/TiO 2 catalyst strongly affects catalytic CO 2 methanation. When the Ni loading amount was increased to 10 wt%, the selectivity switched to favor the CH 4 formation. Ni nanoparticles (NPs) immobilized on a TiO 2 support were synthesized using a deposition-precipitation method followed by a calcination-reduction process, and the CO 2 conversion and CH 4 selectivity achieved 96 and 99% with a Ni loading of 15 wt% at 260°C [37]. Due to the good dispersion of Ni NPs with large unsaturation facilitates a high exposure of active sites, the formation of surface-dissociated hydrogen and the subse-quent hydrogenation removal of surface nickel carbonyl species was accelerated, accounting for the resulting enhanced low-temperature catalytic performance [37]. In the past few years, CeO 2 -ZrO 2 solid solution (Ce x Zr 1−x O 2 ), an active oxygen material, has been commonly used as a support for automotive three-way catalysts because of its high oxygen storage capacity (OSC), which is important in many reactions [50,51], and it also used as the support for CO 2 methanation. The Ni-based catalysts on Ce x Zr 1−x O 2 are greatly efficient in terms of activity and stability, which can be attributed to their high oxygen storage capacities and high Ni dispersion [34][35][36]. In CO 2 methanation, the Ni 2+ ion incorporation into the Ni-Ce x Zr 1−x O 2 (Ni-CZ) catalyst significantly enhances the specific catalytic activity of the CZ catalyst [44], and the global catalytic activities of CO 2 methanation on CZ catalysts depended on the surface for available metallic nickel, the composition of the support, and its modification by Ni 2+ doping. In addition, the Ce x Zr 1−x O 2 catalyst can be synthesized by a simple hydration process, which achieved the goal of Ce and Ni enriched on the surface [34]. Meanwhile, a new NH 3 reduction method for the preparation of Ni−Ce 0.12 Zr 0.88 O 2 lead to a higher active metal reducibility, smaller Ni 0 crystallite size, and higher metal dispersion compared to the H 2 -reduction method with 100% CO and 97% CO 2 conversions and ≥ 98% CH 4 selectivity at 250°C [36]. For NH 3 -treated samples, the metal dispersion is found to decrease with the increase in Ni amounts due to the formation of bulk Ni particles. However, all H 2 -treated samples showed a larger NiO particle size and a lower metal dispersion than the NH 3 -treated samples might owing to the H 2 -reduced sample exhibits an aggregation of smaller particles and/or metal sintering [36].
Nowadays, metal-organic frameworks (MOFs) have attracted much interest as catalysts and/or supporting materials for active metals or complexes in heterogeneous catalysts [52,53], e.g., a highly active catalyst Ni/MOF-5 showed unexpected activity at low temperature for CO 2 methanation [46]. For 10Ni/MOF-5, a very high specific surface area of 2961 m −2 ·g −1 and a large pore volume of 1.037 cm −3 ·g −1 led to a high dispersion of Ni of 41.8%, and the highly uniform dispersion of Ni in the framework of MOF-5 facilitates a high exposure of active sites, resulting the enhancement of the CO 2 conversion to 75.09% and CH 4 selectivity to 100% at 320°C. To further confirm the high dispersion of Ni on the MOF-5 support, the Ni dispersion on MOF-5 and SiO 2 was measured by the H 2 chemisorption. The Ni dispersion on the 10Ni/ MOF-5 catalyst was 41.8% as well as that on 10Ni/SiO 2 was 33.7%, as shown in Figure 5, which indicated that Ni was more highly dispersed on MOF-5 [46]. In conclusion, the Ni loading amount is dependent on the type of support used, and the Ni loading amount on the support will determine its crystallite size and dispersion on the surface of the support.

Catalyst stability
The stability of a catalyst is closely related to the structural destruction, coking, and metal sintering during CO 2 methanation [28,54]. The long-term catalytic stability and thermal stability of Ni/H−Al 2 O 3 was investigated, the CO 2 conversion decreases slowly in the first 180 h and then remains almost constant with a total decrease of 7% after 252 h. No obvious aggregation or sintering of Ni nanoparticles was observed for the Ni/H−Al 2 O 3 catalyst after 252 h upon streaming [43]. Moreover, the control of thermal sintering is critical for maintaining the activity, which requires a stable support and an effective method to prevent particle migration and coalescence [55]. The embedding of Ni nanoparticles onto the Al 2 O 3 matrix enhances the metal-support interaction, and prevents the sintering and/or the aggregation of the active nickel species, which shows that the Ni species was embedded in the hierarchical matrix by an in situ reduction approach, and the Ni species exhibit a high dispersion degree and high stability, guaranteeing their high activity during the long-term use.
The Ni/MOF-5 catalyst also shows the catalytic activity during 100 h of CO 2 methanation over 10Ni/MOF-5 at 280°C (Figure 6). The CO 2 conversion remained above 47.2% and CH 4 selectivity was almost 100% during the 80 h reaction. Obviously, the 10Ni/MOF-5 catalyst was quite stable [46]. However, on Ce x Zr 1−x O 2 support, the Ce-rich sample (5NiC4Z) showed the better stability from the CO 2 conversions (72.21-62.18%), whereas the CO 2 conversions were 51.63-36.42% and 37.64-23.19% over 5NiCZ and 5NiCZ4, respectively [34]. The higher reducibility of the Ce-rich supported highly-dispersed Ni catalyst was considered to be the important factors to ensure its long-term stability [34].  Figure 7, the stability of different Ni supported catalysts was studied, and the rate formation of CH 4 of Ni/MCM-41, Ni/HY, Ni/SiO 2 , and Ni/γ-Al 2 O 3 catalysts decreases slightly with time on stream increases; however, the rate formation of CH 4 on the Ni/MSN catalyst shows no obvious decrease [32]. In particular, the Ni/MCM-41 shows a minimum percent decrease of the CH 4 formation rate of 3.4%, whereas the Ni/HY, Ni/SiO 2 , and Ni/γ-Al 2 O 3 is 9.0, 10.6 and 26.6%, respectively. The presence of coke deposition on the active sites is known for the catalyst deactivation; however, no coke content was observed on the Ni/MSN catalyst from the TGA result and the highest coke content was observed on the Ni/Al 2 O 3 catalyst (9.1%), indicating that the Ni/MSN catalyst did not show any sign of deactivation for the methanation reaction up to 200 h of time-on-stream. Therefore, the Ni/MSN catalyst is resistant toward coke formation and presented good stabilities under the reaction conditions [32].

Effect of the second metal 2.2.2.1. Enhancement of catalytic performance
Ni-based catalysts are vulnerable to sintering and coking, which may lead to their deactivation. Hence, many efforts have been made to enhance the catalytic activity, including selection of appropriate supports and addition of catalytic promoters such as Ce, Zr, La, Mg, V, and Co [45,56,57]. The most noticeable effect due to the promotion with these metals is a considerable increase both in the CO 2 conversion and CH 4 selectivity under steady conditions. The catalytic performance of nickel-based catalysts supported on mesoporous nanocrystalline γ-Al 2 O 3 promoted with CeO 2 , MnO 2 , ZrO 2 , or La 2 O 3 was investigated, and the Ce promoter considerably increases the CO 2 conversion in the methanation reaction ( Table 3). The addition of the Ce promoter to Ni increased the dissociation and CO 2 hydrogenation, and weakened the C=O bond of CO 2 adsorbed on the Ni active sites. Compared with the unpromoted Ni/ Al 2 O 3 catalyst, the addition of Ce strengthen the interaction between Ce and Ni, resulting in better activity of the Ce-Ni/Al 2 O 3 catalyst [58]. Doping the Ni-zeolites catalysts with 3-15% of Ce would be much more enhanced the catalytic performance than the unpromoted catalysts [41]. Actually, the presence of CeO 2 after reduction might promote CO 2 activation into CO, the final catalyst properties being due to the synergetic effect between the metal active sites and the promoter. Reaction conditions: H 2 /CO 2 molar ratio = 3.5, GHSV = 9000 mL·g cat −1 ·h −1 and 350°C.

Catalysts
Some active metals, such as Co, Cu, and Fe, are also used to control the catalytic performance over the supported Ni catalyst, which behave an active aspect as the second metal. Compared with Co and Cu, iron is a suitable second metal for the Ni/ZrO 2 catalyst for low-temperature CO 2 methanation [59], which might be due to its strong electron-donating ability, and Fe 2+ can promote the reduction of nickel and zirconia. Interestingly, similar results are verified and evaluated the catalytic performance of mesoporous nickel-alumina xerogel catalysts (denote as NiAX) with different second metal (M = Fe, Zr, Ni, Y, and Mg) in a fixed bed reactor ( Table 4) [45]. However, the oxidized Co is more active toward the methane formation at low temperatures [59,60], and the Co addition can remarkably change the catalytic performance when active Ce x Zr 1−x O 2 are used as a support for the Ni catalysts [61]. In addition, a homogeneous alloy of Co and Ni can be formed after H 2 reduction and remain after use for reaction in Co-Ni bimetallic catalysts, which increase the metal dispersion in the catalyst, indicating a certain amount of Co addition can considerably improve the catalytic performance [61,62].  Table 4. Catalytic performance of 35Ni5MAX (M = Fe, Zr, Ni, Y, and Mg) catalysts for methane production from carbon dioxide and hydrogen obtained at 220°C after a 10 h-catalytic reaction [45].

Nickel reducibility
In general, the promotion of methanation catalysts with addition of second metals would enhance the nickel reducibility [63,64]. The improvements in the Co reducibility may occur without any effect on the Co dispersion for the Ni-Ce/USY catalysts [41]. While the effect of promotion with Ce on the Ni reducibility is particularly pronounced with the alumina-supported Ni catalysts [63]. Compared to the unpromoted Ni/Al 2 O 3 , the lower reduction temperature of NiO in Ni-CeO 2 /Al 2 O 3 samples implies that addition of CeO 2 decreased the reduction temperature by altering the interaction between Ni and Al 2 O 3 , and improved the catalyst reducibility [16,63,64]. CNTs-supported catalysts exhibited better catalytic performance than the traditional Al 2 O 3 -supported catalysts [16], which attributed to the outstanding reduction properties of the CNTs-supported catalysts, which provided much more active sites for CO 2 methanation. As shown in H 2 -TPR analysis (Figure 8), the accession of Ce could effectively promote the reduction of the nickel oxides, the high reduction peak temperature, corresponding to the highly dispersed nickel oxides in intimate contact with the exterior walls of the CNTs, decreased from 480 to 460°C for the 12Ni/CNT and 12Ni4.5Ce/CNT [16], which suggested easily reducible nickel species on the surface of the 12Ni4.5Ce/CNT catalyst, which may due to the interaction change between the metal oxides and CNTs by the addition of cerium. Recently, a new kind of γ-Al 2 O 3 −ZrO 2 −TiO 2 −CeO 2 composite oxide supported Ni-based catalysts was synthesized for CO 2 methanation [65]. The optimal catalytic activity of the composite oxide supported Ni-based catalysts was achieved because of the improvements in the reducibility. According to the H 2 -TPR profile for all the catalysts, the high temperature peak (weakly interacted with Al 2 O 3 , or called Ni rich phase) shifts downward for the composite oxide-supported Ni-based catalysts, suggesting a weaker interaction between NiO and the composite support. Furthermore, the reduction of the Ni rich phase would benefit the formation of large-sized Ni particles, which are active at low temperatures [66]. Therefore, increasing the fraction of Ni rich phase, i.e., NiO, the active species for the methanation reaction, would result in an increase in the CO 2 conversion at lower temperatures. Moreover, the H 2 consumed amount increased on the composite oxides support, confirming a higher reducibility of NiO on the composite oxides due to the weaker metal-support interaction [65].

Cobalt-based catalysts for low-temperature methanation of CO 2
Generally, the Co-based Fischer-Tropsch catalysts exhibit a superior catalytic performance with respect to low-temperature CO 2 methanation [17,67,68]. A higher CH 4 selectivity was observed in the Fischer-Tropsch synthesis when the Co catalysts were not completely reduced or when the catalysts contain smaller Co 3 O 4 particles [67]. When taking the coke oven gas as feed gas and using a nanosized Co 3 O 4 catalyst, CO was easily adsorbed onto the smaller nanosized Co 3 O 4 surface and react with H 2 , and the temperature at which CO completely converted to CH 4 was much lower than that using nanosized Co 3 O 4 with large particles [67].
In addition, the Ru-doped Co 3 O 4 catalyst with a relatively rough surface shows a lower light-off temperature than that of a Co catalyst [68]. The relatively rough surface morphology of Ru-doped Co 3 O 4 probably results from the larger ionic radius of Ru 3+ , which affects the dissolution-recrystallization process. Therefore, the final surface morphology of nanorods was disrupted with the addition crystalline defects. The correlation between the surface chemistry and the catalytic performances suggests that doping a noble metal to an oxide of an earth-abundant metal followed by reduction could create a chemically stable, cost-effective catalyst with a bimetallic surface, which has an equivalent or much better catalytic performance [68]. Usually, the catalytic activity affected by the catalyst composition and structure, e.g., when used the mesoporous Co/KIT-6 and Co/meso-SiO 2 in CO 2 methanation, the highly ordered bicontinuous mesoporous structure of the Co/KIT-6 catalyst exhibits higher methane selectivity than the Co/meso-SiO 2 catalyst, and the CO 2 conversion exceeds 48.9%, and the methane selectivity can be retained at 100% at 280°C [17].

Reaction mechanisms
According to the previous research, the reaction mechanism was difficult to establish mainly because of the different opinions on the intermediate and the methane formation process. Two feasible reaction mechanisms were proposed for CO 2 methanation in the past decades. The first one involves the CO 2 convert to CO prior to methanation, and the subsequent reaction follows the same mechanism as CO methanation [69]. Similar to the mechanism of CO 2 hydrogenation to CH 3 OH, someone considered CO was an intermediate [32], and the CO hydrogenation to methane also been focused [70,71]. The other mechanism involves the direct CO 2 hydrogenation to methane without forming CO as an intermediate [72]. However, the mechanism depends on different catalysis systems, which are still under investigation.
The atomic hydrogen dissociated from Ni sites in the MSN may facilitate the formation of methane, as shown in Figure 9. The oxygen vacancies will be formed when H 2 react with the surface oxygen along with the water generation, which activate additional CO 2 to fill the vacancies and produce CO. During the CO 2 methanation reaction, CO was also suggested as the alternative product, which was an intermediate, as shown in Eqs.
(2)-(9) [32]. Therefore, the higher CH 4 selectivity can be explained by the enhanced supply of adsorbed hydrogen to the activated adsorbed CO intermediate, which was the rate-determining step [73]. However, some researchers considered that the main mechanism for CO 2 methanation does not require CO as the reaction intermediate [28,74], which can be explained by the importance of weak basic sites the adsorption of CO 2 [28].
Density functional theory is helpful in understanding the mechanistic aspects of the reactions. Different mechanisms of CO 2 methanation on Ni(111) surfaces were investigated, and the energy barrier of 237.4 kJ mol −1 is acquired for the dissociation of CO into C and O species, which support that CO 2 is converted to CO, subsequently to carbon before hydrogenation [75].
As mentioned, the CO 2 adsorption is a crucial step for methanation. Indeed, CO 2 dissociation is the rate-limiting step. CO 2 dissociation over Rh-based catalysts is influenced by the CO coverage on the surface and the strength of the bond Rh−CO, and the hydrogen adsorption at the surface is competed with CO 2 adsorption. Due to the preferential adsorption of CO 2 and the accumulation of CO on the surface, hydrogen coverage on the rhodium catalyst is very small [76]. However, CO 2 adsorption on the medium basic sites of Ni/Ce 0.5 Zr 0.5 O 2 results in monodentate carbonates, and monodentate formate derived from monodentate carbonate on medium basic sites, which could be hydrogenated more quickly than bidentate formate derived from hydrogen carbonate. The medium basic sites were proposed to promote the formation of monodentate formate species, thus to enhance the activity [77]. Figure 9. A probable mechanism for Ni/MSN whereby spillover of atomic hydrogen from Ni interacts with C(a) species and sequentially hydrogenates carbon until the product methane desorbs [32].
In addition, at 383 K, a reaction mechanism was proposed for the carbon dioxide methanation reaction on 2% Ru/TiO 2 , which investigate the precursor existence for the adsorbed CO and reaction intermediate, and the side-product, formate was also found adsorbing on the support [78], which suggested the surface intermediate corresponding to the adsorbed formate on the metal-support interface, and the measured formate infrared bands are corresponding to the diffused formate species from the interface to the support. A pathway involving hydrogen carbonate is also presented for the formation of the interfacial formate, because the species is formed on the support during the reaction, and the transient response is consistent with the response of a CO precursor. The reaction mechanism that could account for all of these observations is presented in Figure 10 [78].
To make a better understanding of the adsorption of possible intermediates, the reaction mechanism and factors determining the product selectivity, DFT calculations were considered to be a suitable method to investigate the hydrogenation process of CO 2 and CO on the Ru(0001) surface [79]. For CO 2 hydrogenation, the HCOO intermediate are firstly formed from the adsorbed CO 2 hydrogenation, and subsequently produces an adsorbed CHO and O species. The active C and CH species then undergo stepwise hydrogenation to CH 2 , CH 3 and CH 4 , or the CHx species, and further transforms to longer carbon chains. From the calculation results, CH 3 hydrogenation is considered to be the rate determining step in the sequence of C hydrogenation on the Ru(0001) surface, and the lowest barrier channel of C-C coupling occurs via the CH + CH reaction [79]. In addition, the study based on DFT calculations on a Ru nanoparticle supported on the TiO 2 catalyst further confirms the stronger electron transfer from the Ru cluster to the TiO 2 (101) facet than to TiO 2 (001); the Ru species supported on the (101) plane possesses a relatively lower activation energy for the CO dissociation, resulting in the highly catalytic activity toward CO 2 methanation reaction [80]. Finally, the detailed mechanism was proposed for CO 2 methanation over metal-based MSNs [81]. As shown in Figure 11, CO 2 and H 2 were adsorbed and dissociated on the metal active sites to form CO, O, and H, followed by the migration of these atoms to the MSN surface. Subsequently, the CO dissociated from the active sites interacted with the MSN oxide surfaces to form the carbonyl, including bridged and linear carbonyl, and the H atom in the reaction facilitated the formation of bidentate formate. And the above three species were responsible for the methane formation, among them, the main route for the methane formation was due to the bidentate formate species, and the MSN support served as the sites for carbonyl species, which act as a precursor to methane formation [81].

Conclusions and perspectives
CO 2 has been promoted to an important carbon resource for conversion and utilization, and CO 2 hydrogenation is a feasible and powerful process, especially for methanation. However, CO 2 is chemically stable and thermodynamically unfavorable. To eliminate the limitations on the conversion and selectivity, various technical directions and specific research approaches on rational design of catalysts and exploration of reaction mechanisms have been presented.
Noble metal catalysts such as Ru, Rh and Pd are efficient for the formation of methane under relatively mild operating conditions, but the high cost as well as their limited availability restricts their practical applications. Therefore, researchers have paid increasing attention on the immobilization of homogenous catalysts to combine the efficient activity with the properties of separation and recyclability. Ni-and Co-based catalysts are, of course, more practical for industrial applications compared to noble metal catalysts. The catalysts with larger surface areas and higher metal dispersion can usually possess higher activity and selectivity, and longer stability in the hydrogenation of CO 2 . However, the Ni-based catalysts are more resistless to carbon formation compared with noble metal catalysts. Thus, one strategy has to be proposed for pursuing high-performance catalysts with abilities of low-temperature methanation and resisting carbon formation. In addition, understanding the fundamental mechanisms of CO 2 methanation and explore its relationship with catalyst active site structures using both theoretical calculations (molecular/electronic level modeling) and experimental approaches to tailor new catalyst structures are considerably needed.