Hydrogenation of Carbon Dioxide to Value-Added Chemicals by Heterogeneous Catalysis and Plasma Catalysis

Due to the increasing emission of carbon dioxide (CO2), greenhouse effects are becoming more and more severe, causing global climate change. The conversion and utilization of CO2 is one of the possible solutions to reduce CO2 concentrations. This can be accomplished, among other methods, by direct hydrogenation of CO2, producing value-added products. In this review, the progress of mainly the last five years in direct hydrogenation of CO2 to value-added chemicals (e.g., CO, CH4, CH3OH, DME, olefins, and higher hydrocarbons) by heterogeneous catalysis and plasma catalysis is summarized, and research priorities for CO2 hydrogenation are proposed.


Introduction
Climate changes are mostly induced by the greenhouse effect, and carbon dioxide (CO 2 ) accounts for a dominant proportion of this greenhouse effect. Indeed, one can barely ignore the connection between the emission of CO 2 and climate changes [1]. The CO 2 concentration in the atmosphere has actually climbed to 405 ppm in 2017, as shown in Figure 1a. As the global energy consumption is still mainly based on burning coal, oil, and natural gas, and this situation will last until the middle of the century (see Figure 1b), experts predict that the CO 2 concentration in the atmosphere will continue to rise to~570 ppm by the end of the century if no measures are taken [2]. Hence, it is urgent to control the CO 2 emissions by taking effective measures to capture and utilize CO 2 .
In principle, there are three strategies to reduce CO 2 emissions, i.e., reducing the amount of CO 2 produced, storage of CO 2 , and utilization of CO 2 [3]. Recently, some comprehensive reviews have discussed the technological state-of-the-art of carbon capture and storage (CCS) [4][5][6]. Direct air capture technology (DAC) draws people's attention to mitigate climate change by taking advantage of chemical sorbents (e.g., basic solvents, supported amine and ammonium materials, etc.) [4]. In addition, Bui et al. also considered the economic and political obstacles in terms of the large-scale deployment of CCS [6]. Significantly reducing the amount of CO 2 produced is unrealistic in view of the current energy structure dominated by fossil energies, as shown in Figure 1b. CO 2 storage seems to be a potential approach, but there are some challenges, such as efficiency of capture and sequestration of CO 2 , cutting down the operation costs for capture and separation of CO 2 , and the long-term stability of underground storage [7,8]. Utilization of CO 2 is a promising approach, since CO 2 is a cheap and One of the options to convert CO 2 , is catalytic hydrogenation, as illustrated in Figure 2. The production of H 2 , however, is also a vital problem to realize direct hydrogenation of CO 2 to oxygenates and hydrocarbons. Hydrogen production can be either based on renewable or non-renewable sources, such as electrical, thermal, photonic, and hybrid [10,11]. The main methods of hydrogen production include electrolysis, thermolysis, photo-electrolysis, and hybrid thermochemical cycles [10]. According to some evaluation criteria-such as global warming potential (GWP), social cost of carbon (SCC), acidification potential (AP), energy and exergy efficiencies, and production cost-the hybrid hydrogen production methods seems a promising route. On the other hand, the production cost evaluation shows that coal gasification ($0.92/kg H 2 ) and fossil fuel reforming ($0.75/kg H 2 ) are relatively low-cost methods compared to early R&D phase methods (e.g., photo-electrochemical: $10.36/kg H 2 from water dissociation). However, fossil fuel is non-renewable, being a major drawback. Therefore, reducing the cost of hydrogen production is an urgent and challenging issue, since we have to balance both the economy and sustainability. Various methods have been adopted-including photo-catalysis, electro-catalysis, heterogeneous catalysis, and plasma catalysis [12,13]-to realize hydrogenation of CO2. Dalle et al. systematically presented the activity of the first-row transition metal complexes (i.e., Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn) in CO2 reduction by electro-catalysis, photo-catalysis and photoelectroncatalysis [14]. Li et al. discussed the important roles of co-catalysts (e.g., biomimetic, metal-based, metal-free, and multifunctional) in selective photo-catalytic CO2 reduction [15]. Besides thermochemical approaches, Mota et al. made a detailed retrospective based on electrochemical and photo-chemical approaches for CO2 hydrogenation to oxygenates and hydrocarbons [16]. In this Various methods have been adopted-including photo-catalysis, electro-catalysis, heterogeneous catalysis, and plasma catalysis [12,13]-to realize hydrogenation of CO 2 . Dalle et al. systematically presented the activity of the first-row transition metal complexes (i.e., Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn) in CO 2 reduction by electro-catalysis, photo-catalysis and photoelectron-catalysis [14]. Li et al. discussed the important roles of co-catalysts (e.g., biomimetic, metal-based, metal-free, and multifunctional) in selective photo-catalytic CO 2 reduction [15]. Besides thermochemical approaches, Mota et al. made a detailed retrospective based on electrochemical and photo-chemical approaches for CO 2 hydrogenation to oxygenates and hydrocarbons [16]. In this review, we focus on the latter two methods. Heterogeneous catalysis has been widely studied, while plasma catalysis is still an emerging technology. Some excellent reviews were recently published for heterogeneous catalytic CO 2 hydrogenation [12,17,18]. Jadhav et al. focused on methanol production [17], Porosoff et al. on the synthesis of CO, CH 3 OH, and hydrocarbons [18], while Alvarez et al. discussed the greener preparation of formates (formic acid), CH 3 OH, and DME [12]. With regard to plasma catalysis for CO 2 hydrogenation, there are only a handful reports (as discussed below). Nevertheless, it exhibits great potential since plasma can operate at ambient temperature and atmospheric pressure.
In this review, we summarize the progress of mainly the last five years in CO 2 hydrogenation to value-added chemicals (e.g., CO, CH 4 , CH 3 OH, DME, olefins, and higher hydrocarbons) driven by both heterogeneous catalysis and plasma catalysis. The literature bibliography range is shown in Figure 3. Indeed, on the one hand, the insights obtained by heterogeneous catalysis can be useful for the further development of the emerging field of plasma catalysis. On the other hand, we also want to pinpoint the differences between heterogeneous and plasma catalysis, and thus the need for dedicated design of catalytic system tailored to the plasma environment.

Heterogeneous Catalysis
In heterogeneous catalysis, support materials, active metals, promoters, and the preparation methods of catalysts are major adjustable factors determining the catalytic activity. As far as the support materials are concerned, the catalytic performance is influenced by the metal-support interaction since it usually induces specific physicochemical properties for catalysis, e.g., active cluster, active oxygen vacancy, and acid-based property. Metal oxides and zeolites with special channel structures are usually selected. For the active metal components, research focuses on searching cheap and available metals to replace precious metals or to reduce the amount of precious metals in industrial heterogeneous catalysis. Promoters (structural-type and electron-type) can also significantly influence the catalytic activity by regulating the adsorption and desorption behavior of molecules (reactant, intermediate, and product) on the catalyst surface. Preparation methods usually determine the catalyst morphology including metal dispersion (particle size distribution), specific surface area, and channel structure, which also influences the catalytic performance [12]. In this section, we summarize recent progress of CO2 hydrogenation to CO, CH4, CH3OH, and some other products in terms of rational design of heterogeneous catalysts.

CO2 to CO
CO can be used as feedstock to produce liquid fuels and useful chemicals by Fischer-Tropsch (F-T) synthesis reaction. Therefore, the catalytic hydrogenation of CO2 to CO via the reverse watergas shift (RWGS) reaction, reaction (1), is promising to account for the shortage of energy.

Heterogeneous Catalysis
In heterogeneous catalysis, support materials, active metals, promoters, and the preparation methods of catalysts are major adjustable factors determining the catalytic activity. As far as the support materials are concerned, the catalytic performance is influenced by the metal-support interaction since it usually induces specific physicochemical properties for catalysis, e.g., active cluster, active oxygen vacancy, and acid-based property. Metal oxides and zeolites with special channel structures are usually selected. For the active metal components, research focuses on searching cheap and available metals to replace precious metals or to reduce the amount of precious metals in industrial heterogeneous catalysis. Promoters (structural-type and electron-type) can also significantly influence the catalytic activity by regulating the adsorption and desorption behavior of molecules (reactant, intermediate, and product) on the catalyst surface. Preparation methods usually determine the catalyst morphology including metal dispersion (particle size distribution), specific surface area, and channel structure, improve the life cycle of catalysts, non-noble metal carbides have been developed. Porosoff et al. synthesized a molybdenum carbide (Mo2C) catalyst for the RWGS reaction [20], yielding 8.7% CO2 conversion and 93.9% CO selectivity (at 573 K reaction temperature). The catalytic performance was much better than those of bimetallic catalysts (e.g., Pt-Co, Pt-Ni, Pd-Co, Pd-Ni) supported on CeO2 (~5% CO2 conversion and 83.3% CO selectivity at the same reaction temperature). The catalytic mechanism of Mo2C in the RWGS reaction was further studied by ambient pressure X-ray photoelectron spectroscopy (AP-XPS) and in situ X-ray absorption near edge spectroscopy (XANES), which demonstrate that Mo2C not only broke the C=O bond, but also dissociated hydrogen. Hence, Mo2C has a dual functional and is an ideal catalytic material for the RWGS reaction. Interestingly, the authors also found that by modifying the catalyst with Co, the catalytic activity of the Co-Mo2C catalyst was further improved (9.5% CO2 conversion, 99% CO selectivity at 573 K), probably attributed to the existence of the CoMoCyOz phase confirmed by X-ray Diffraction (XRD) (see Figure  4). The support plays an important role in CO2 hydrogenation to CO through the RWGS reaction. Kattel et al. prepared Pt, Pt/SiO2 and Pt/TiO2 materials as catalysts for the RWGS reaction [21], and showed that Pt nanoparticles alone cannot catalyze the RWGS reaction, while using SiO2 and TiO2 as support, the overall CO2 conversion can be significantly improved, pointing towards a synergy effect between Pt and the oxide support. To reveal the synergy effect, they combined density functional theory (DFT), kinetic Monte Carlo (KMC) simulations and other experimental measurements. They  The support plays an important role in CO 2 hydrogenation to CO through the RWGS reaction. Kattel et al. prepared Pt, Pt/SiO 2 and Pt/TiO 2 materials as catalysts for the RWGS reaction [21], and showed that Pt nanoparticles alone cannot catalyze the RWGS reaction, while using SiO 2 and TiO 2 as support, the overall CO 2 conversion can be significantly improved, pointing towards a synergy effect between Pt and the oxide support. To reveal the synergy effect, they combined density functional theory (DFT), kinetic Monte Carlo (KMC) simulations and other experimental measurements. They found that the hydrogenation of CO 2 to CO is promoted once CO 2 is stabilized by the Pt-oxide interface. In the case of Pt nanoparticles (NP) alone, the conversion was close to 0, since the ability of Pt NP in binding CO 2 is weak. When SiO 2 and defected TiO 2 with oxygen vacancies served as supports, however, the CO 2 conversion was enhanced (to 3.35% and 4.51%, respectively) on the Pt-oxide interface. The synergy effect between Pt and oxide supports in activating and hydrogenating CO 2 is shown in Figure 5, and possible reaction pathway [21] for the RWGS reaction is illustrated in Scheme 1.  (110). The black circle in (a) depicts the position of oxygen vacancy on TiO 2 (110). Note: Si: green, Ti: light blue, Pt: light gray, C: dark gray, O: red, and H: blue, reprinted with permission from [21]. Copyright Elsevier, 2016. (1) (a) Pt25/hydroxylated SiO2 (111), and (b) *CO2 species, (c) *CO species, (d) *HCO species, (e) *H2COH species, (f) *CH3OH species, (g) *CH2 species, and (h) *OH species adsorbed on Pt/SiO2 (111). The dashed lines show hydrogen bonds. (2) (a) Pt25/TiO2 (110) with oxygen vacancy, and side (top) and top (bottom) views of (b) *CO2 species, (c) *CO species, (d) *HCO species, (e) *H2COH species, (f) *CH3OH species, (g) *CH2 species, and (h) *OH species adsorbed on Pt/TiO2 (110). The black circle in (a) depicts the position of oxygen vacancy on TiO2 (110). Note: Si: green, Ti: light blue, Pt: light gray, C: dark gray, O: red, and H: blue, reprinted with permission from [21]. Copyright Elsevier, 2016.

Scheme 1.
Reaction pathways for the RWGS reaction, where '*X' represents species X adsorbed on a surface site, reproduced with permission from [21]. Copyright Elsevier, 2016.
Yan et al. investigated the effect of the Ru-Al2O3 interfaces on the catalytic activity of the RWGS reaction, and proposed Ru35/Al2O3 and Ru9/Al2O3 catalyst models to explain the experimental observations [22]. The product selectivity switched between CH4 and CO over the Ru/Al2O3 catalyst, i.e., monolayer Ru sites favored the production of CO, while Ru nano-clusters preferred the production of CH4, during CO2 reduction reaction. Confirmed by kinetic analysis, characterization of the surface structures and real-time monitoring of the active intermediate species, the product selectivity of CH4 and CO was regulated by the Ru sites and Ru-Al2O3 interfacial sites. Furthermore, based on the combination of theoretical calculations and isotope-exchange experimental results, the Scheme 1. Reaction pathways for the RWGS reaction, where '*X' represents species X adsorbed on a surface site, reproduced with permission from [21]. Copyright Elsevier, 2016.
Yan et al. investigated the effect of the Ru-Al 2 O 3 interfaces on the catalytic activity of the RWGS reaction, and proposed Ru 35 /Al 2 O 3 and Ru 9 /Al 2 O 3 catalyst models to explain the experimental observations [22]. The product selectivity switched between CH 4 and CO over the Ru/Al 2 O 3 catalyst, i.e., monolayer Ru sites favored the production of CO, while Ru nano-clusters preferred the production of CH 4 , during CO 2 reduction reaction. Confirmed by kinetic analysis, characterization of the surface structures and real-time monitoring of the active intermediate species, the product selectivity of CH 4 and CO was regulated by the Ru sites and Ru-Al 2 O 3 interfacial sites. Furthermore, based on the combination of theoretical calculations and isotope-exchange experimental results, the authors found that the O* species derived from the dissociative adsorption of CO 2 at interfacial Ru sites easily bridge with the Al sites from the γ-Al 2 O 3 support, and new Ru-O-Al bonds are formed via the oxygen-exchange process. The interfacial O species existing in Ru-O-Al bonds was responsible for the CO 2 activation via oxygen-exchange with the O atoms of CO 2 . Therefore, this experimental work is a good inspiration to further explore the influence of metal-support interfaces for the effective activation of CO 2 .
Besides the widely used impregnation method, Yan et al. also reported a doping-segregation method for the preparation of Rh-doped SrTiO 3 [23].
Precursors with a molar ratio of Sr:Ti:Rh = 1.10:0.98:0.02. First, TiO 2 was suspended in deionized water, and then Sr(OH) 2 ·8H 2 O and Rh(NO 3 ) 3 were introduced. Subsequently, the sample was poured in a stain steel acid digestion vessel, which was kept at 473 K for 24-48 h. Finally, the reaction product was dried at 343 K overnight after it was centrifuged, and washed via deionized water. As confirmed by in situ X-Ray-Diffraction (XRD) and X-ray Absorption Fine Structure (XAFS) measurements, the Rh-doped SrTiO 3 catalysts produce sub-nanometer Rh clusters, which are highly active for the conversion of CO 2 compared to the supported Rh/SrTiO 3 prepared by wetness impregnation. The better catalytic performance (7.9% CO 2 conversion and 95% CO selectivity at 573 K) could be ascribed to the cooperative effect between sub-nanometer Rh clusters and the reconstructed SrTiO 3 which is active for dissociation of H 2 and is favorable for adsorption/activation of CO 2 . Therefore, the novel approach, doping-segregation method, maybe a novel strategy to tune the size of active metals and the physicochemical properties of supports for rational design of catalysts for the RWGS reaction.
Dai et al. studied CeO 2 catalysts which were prepared by the hard-template method (Ce-HT), the typical complex method (Ce-CA), and the typical precipitation method (Ce-PC) for the RWGS reaction [24]. The experimental results show that catalysts prepared by Ce-CA, Ce-HT, and Ce-PC methods exhibit a 100% CO selectivity and the CO 2 conversions were 9.3, 15.9, and 12.7% respectively at 853 K. Obviously, the hard-template (Ce-HT) method is beneficial for the RWGS reaction in view of the catalytic performance. The Ce-HT method comprises the following steps: (1) Ce(NO 3 ) 3 ·6H 2 O was dissolved in ethanol, and KIT-6 mesoporous silica was added; (2) The mixture was stirred until a dry power was obtained; (3) The powder was calcined; (4) The obtained samples were treated with NaOH to remove the template. To reveal the relationship between the preparation method and the catalytic activity, the authors carried out XRD, Transmission Electron Microscopy (TEM) and Brunauer-Emmett-Teller (BET) characterization, and the characterization results show that CeO 2 catalysts which were prepared by the Ce-HT method have a porous structure and a high specific surface area, while CeO 2 catalysts prepared by the other methods (Ce-CA, Ce-PC) have an agglomerated structure (Ce-CA) and overlapped bulk structure (Ce-PC) with low porosity. Moreover, in the CeO 2 catalysts, oxygen vacancies as active sites were formed by H 2 reduction at 673 K, which were confirmed by in situ X-ray photoelectron spectroscopy (XPS) and H 2 -temperature-programmed reduction (H 2 -TPR). Therefore, the improvement of the catalytic activity for the RWGS reaction can be ascribed to the change of catalyst structure and oxygen vacancies. Except for the above monometallic catalysts, some bimetallic catalysts-i.e., Pt-Co, Fe-Mo and Ni-Mo [25][26][27]-were also prepared and tested in the RWGS reaction. In general, the preparation method of bimetallic catalysts generally can be divided into two categories: (1) the bi-metal was first prepared and then supported on the carrier [25]; (2) the bi-metal as formed in the preparation progress [26,27]. Compared with pure Co catalyst, Pt-Co catalyst showed a better catalytic activity (mainly CO, close to 100%) for the RWGS reaction. Ambient pressure X-ray photoelectron spectroscopy (AP-XPS) and environmental transmission electron microscopy (ETEM) showed that Pt migrates on the catalyst surface, and Pt aids the reduction of Co to its metallic state under appropriate reaction conditions confirmed by near edge X-ray absorption fine structure (NEXAFS) spectroscopy [25]. Abolfazl et al. investigated Mo/Al 2 O 3 and Ni-Mo/Al 2 O 3 catalysts prepared by the impregnation method used for the RWGS reaction. The experimental results showed that Ni-Mo/Al 2 O 3 catalyst is a promising catalyst (35% CO 2 conversion, i.e., close to the 38% equilibrium conversion) compared with Mo/Al 2 O 3 catalyst (15% CO 2 conversion). As demonstrated by XPS, the electronic effect, which transfers electrons from Ni to Mo and leads to an electron-deficient state of the Ni species, is beneficial for CO 2 adsorption, and thus improves the CO 2 conversion. Furthermore, the XRD and H 2 -TPR profiles indicated that the Ni-O-Mo structure crystallizes into the NiMoO 4 phase which can improves the adsorption and dissociation of H 2 on the Ni-Mo/Al 2 O 3 catalyst [27]. Similarly, they also found that Fe-Mo/Al 2 O 3 catalyst synthesized by the impregnation method is an efficient catalyst with high CO yield, almost no by-products and relatively stable (60 h) for the RWGS reaction. The enhancement of the catalytic activity may be attributed to better Fe dispersion with the addition of Mo and smaller particle size of active Fe species, which was confirmed by the BET method and scanning electron microscopy (SEM) [26].
Overall, to some extent, the reaction activity and stability of bimetallic catalysts for the direct hydrogenation of CO 2 to CO are excellent compared to mono-metallic catalysts. The reduction of the active metal, the formation of alloy metal, the dispersion and particle size of the catalyst are possible factors for the improvement of the catalytic performance in the RWGS reaction. Details of the conversion and product selectivity, along with the reaction conditions of several representative RWGS catalytic systems, are compared in Table 1. Obviously, precious metal catalysts (e.g., Pt, Rh, Ni) are still advantageous for the formation of CO compared to non-noble metal catalysts (e.g., Fe, Co, Mo), and upon increasing the gas hourly space velocity (GHSV), the selectivity of CO slightly decreases to some extent. Methane, a high value carbon source, is used to produce syngas via steam reforming, and subsequently the syngas is usually converted into chemicals and/or fuels through F-T synthesis. The direct hydrogenation of CO 2 to CH 4 (also called CO 2 methanation), reaction (2), is a feasible approach, if the production technology of H 2 becomes widespread and at low-cost.
Ni-based [29], Co-based [30], and Ru-based [31] catalysts have been used in CO 2 methanation, and Ni-based catalysts are considered to be the most effective and the lowest cost alternative. However, coke formation is a serious problem for Ni-based catalytic systems [2]. Therefore, researchers are trying to seek appropriate promoters to improve the activity and stability of Ni-based catalytic systems for CO 2 methanation. Yuan et al. investigated the effect of Re on the catalytic activity of Ni-based catalysts in CO 2 methanation [32]. Based on DFT calculations, they found that, attributed to the strong affinity of Re to O (see Figure 6), the presence of Re markedly lowered the energy barrier of C-O bond cleavage, which benefits the activation of CO 2 . Moreover, CH 4 selectivity can be enhanced owe to the presence of Re, which was confirmed by analysis of surface coverage of the adsorbed species on Ni (111) and Re@Ni (111). In addition, micro-kinetic analysis showed that, in addition to CO* and H*, a suitable amount of O ad atoms were present on Re@Ni (111), and thus a possible reaction network of CO 2 methanation was proposed as shown in Scheme 2, in which a red line represents the preferable steps in each pathway.   Besides Re, La is also an excellent promoter of Ni-based catalysts for CO2 methanation. Quindimil et al. [29] applied Ni-La2O3/Na-BETA catalysts for CO2 methanation. They found that the presence of La2O3 created more CO2 adsorption sites and more hydrogenation sites, mainly attributed to the improved surface basicity and Ni dispersion by the co-catalyst effect of La. Under the optimized reaction conditions, 65% CO2 conversion and nearly 100% CH4 selectivity were achieved over a Ni-10%La2O3/Na-BETA catalyst with a good stability for more than 24 h at 593 K.
CeO2, TiO2, and SiO2 have been used as the supports of methanation catalysts [33]. Reactions over Ni/CeO2 catalyst performed full selectivity to CH4 with higher TOF (up to forty-fold) compared to TiO2 and SiO2 supported Ni nanoparticles, and in CO2 methanation, the catalytic stability of    Besides Re, La is also an excellent promoter of Ni-based catalysts for CO2 methanation. Quindimil et al. [29] applied Ni-La2O3/Na-BETA catalysts for CO2 methanation. They found that the presence of La2O3 created more CO2 adsorption sites and more hydrogenation sites, mainly attributed to the improved surface basicity and Ni dispersion by the co-catalyst effect of La. Under the optimized reaction conditions, 65% CO2 conversion and nearly 100% CH4 selectivity were achieved over a Ni-10%La2O3/Na-BETA catalyst with a good stability for more than 24 h at 593 K.
CeO2, TiO2, and SiO2 have been used as the supports of methanation catalysts [33]. Reactions over Ni/CeO2 catalyst performed full selectivity to CH4 with higher TOF (up to forty-fold) compared to TiO2 and SiO2 supported Ni nanoparticles, and in CO2 methanation, the catalytic stability of Scheme 2. Reaction network for CO 2 methanation. The preferable steps in each pathway are marked with a red line, reproduced with permission from [32]. Copyright Elsevier, 2018.
Besides Re, La is also an excellent promoter of Ni-based catalysts for CO 2 methanation. Quindimil et al. [29] applied Ni-La 2 O 3 /Na-BETA catalysts for CO 2 methanation. They found that the presence of La 2 O 3 created more CO 2 adsorption sites and more hydrogenation sites, mainly attributed to the improved surface basicity and Ni dispersion by the co-catalyst effect of La. Under the optimized Catalysts 2019, 9, 275 9 of 37 reaction conditions, 65% CO 2 conversion and nearly 100% CH 4 selectivity were achieved over a Ni-10%La 2 O 3 /Na-BETA catalyst with a good stability for more than 24 h at 593 K. CeO 2 , TiO 2 , and SiO 2 have been used as the supports of methanation catalysts [33]. Reactions over Ni/CeO 2 catalyst performed full selectivity to CH 4 with higher TOF (up to forty-fold) compared to TiO 2 and SiO 2 supported Ni nanoparticles, and in CO 2 methanation, the catalytic stability of Ni/CeO 2 catalyst lasted for 50 h at 523 K. HRTEM analysis indicated that different supports induced distinctive crystal structure. Ni/CeO 2 catalyst presented hexagonal Ni nanocrystallites, while TiO 2 and SiO 2 favored the formation of pseudo-spherical Ni nanoparticles. Demonstrated by TPR, XPS, and UV Raman analysis, characterization results revealed partial reduction of the CeO 2 surface, and the partial reduction of the CeO 2 surface contributed to the generation of oxygen vacancies, which is beneficial for the formation of a strong metal-support interaction (SMSI) between Ni and CeO 2 , while no SMSI was observed over Ni/SiO 2 and Ni-TiO 2 catalyst. Furthermore, pulse reaction by temporal analysis products (TAP) demonstrated the capacity of CO 2 adsorption following the order: Ni/SiO 2 < Ni/TiO 2 < Ni/CeO 2 . Therefore, metal particle morphology and surface oxygen vacancies were used to anchor/stabilize Ni nanoparticles, and SMSI of Ni/CeO 2 catalyst contributed to the remarkable catalytic activity for CO 2 methanation.
Lin et al. investigated the influence of TiO 2 phase structure on the degree of dispersion of Ru nanoparticles [31]. Experiments showed that Ru/r-TiO 2 (rutile-type TiO 2 ) catalysts have a fast rate of CO 2 conversion, more than twice as fast as Ru/a-TiO 2 (anatase-type TiO 2 ) for CO 2 methanation. Meanwhile, compared to Ru/a-TiO 2 catalysts, Ru/r-TiO 2 catalysts exhibited a much higher thermal stability. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images showed that r-TiO 2 supported Ru nanoparticles displayed a narrower particle size distribution (1.1 ± 0.2 nm) compared with a-TiO 2 supported Ru nanoparticles (4.0 ± 2.4 nm). As confirmed by XRD measurements and H 2 -TPR experiments, a strong interaction existed in RuO 2 and r-TiO 2 contributed to the formation of the Ru-O-Ti bond. Experimental tests, revealed that the strong interaction between RuO 2 and r-TiO 2 , not only promotes the highly dispersion of Ru nanoparticles, but also prevents nanoparticles' aggregation, which is responsible for the enhancement of the catalytic activity and thermal stability.
Furthermore, the influence of Al 2 O 3 , ZrO 2 , SiO 2 , KIT-6, and GO supports on Ni-based catalysts [30,34,35] has also been reported recently ( Table 2). The specific area of support, the metal-support interaction and particle size of active sites are still mainly adjustable factors. Interestingly, Ni-SiO 2 /GO-Ni-foam catalyst which was synthesized via intercalation of graphene oxide (GO) exhibited excellent activity compared with Ni-SiO 2 /Ni-foam catalyst (see Figure 7) at 743 K for CO 2 methanation. Furthermore, the formation of nickel silicates on GO is responsible for the uniform dispersion of Ni active sites, which is favorable for inhibition of catalyst sintering.   Besides the traditional methods (co-impregnation method and deposition-precipitation method), some novel preparing methods, such as the sequential impregnation and a change of the preparation progress (e.g., the metal incorporation order, the reduction temperature, and the selection of different types of precursors), have been recently reported for preparing methanation catalysts.
Romero-Sáez et al. synthesized Ni-ZrO 2 catalysts supported on CNTs via the sequential and co-impregnation methods for CO 2 methanation [36]. The catalyst prepared by co-impregnation was apparently less active and selective to CH 4 , compared with the catalyst synthesized by the sequential impregnation method. The preparation approach of the sequential impregnation method is as follows: (1) the appropriate amount of ZrO(NO 3 ) 2 xH 2 O was dissolved in acetone; (2) the CNTs were added to the solution; (3) removal of the solvent, drying and heat treatment at 623 K; (4) in a rotatory evaporator, the same procedure for Ni(NO 3 ) 2 impregnation, followed drying and heat treatment was applied. As characterized by TEM analysis, NiO nanoparticles surrounded by ZrO 2 in core-shell structures were formed via co-impregnation method.The existence of core-shell structure reduced reactant access to Ni and Ni-ZrO 2 interface. However, when the catalyst was prepared via a sequential impregnation method, NiO nanoparticles were available and deposited either on the surface or next to the ZrO 2 nanoparticles, which improved the extent of the Ni-ZrO 2 interface. The ratio of Ni-O-Zr exposed species thus increases in the sequential impregnation method, and the interaction between H atoms (produced upon H 2 dissociation on the Ni surface) and the CO 2 molecule (activated by ZrO 2 ), can also be enhanced. The schematic representation is shown in Figure 8. Besides the traditional methods (co-impregnation method and deposition-precipitation method), some novel preparing methods, such as the sequential impregnation and a change of the preparation progress (e.g., the metal incorporation order, the reduction temperature, and the selection of different types of precursors), have been recently reported for preparing methanation catalysts.
Romero-Sáez et al. synthesized Ni-ZrO2 catalysts supported on CNTs via the sequential and coimpregnation methods for CO2 methanation [36]. The catalyst prepared by co-impregnation was apparently less active and selective to CH4, compared with the catalyst synthesized by the sequential impregnation method. The preparation approach of the sequential impregnation method is as follows: (1) the appropriate amount of ZrO(NO3)2 xH2O was dissolved in acetone; (2) the CNTs were added to the solution; (3) removal of the solvent, drying and heat treatment at 623 K; (4) in a rotatory evaporator, the same procedure for Ni(NO3)2 impregnation, followed drying and heat treatment was applied. As characterized by TEM analysis, NiO nanoparticles surrounded by ZrO2 in core-shell structures were formed via co-impregnation method.The existence of core-shell structure reduced reactant access to Ni and Ni-ZrO2 interface. However, when the catalyst was prepared via a sequential impregnation method, NiO nanoparticles were available and deposited either on the surface or next to the ZrO2 nanoparticles, which improved the extent of the Ni-ZrO2 interface. The ratio of Ni-O-Zr exposed species thus increases in the sequential impregnation method, and the interaction between H atoms (produced upon H2 dissociation on the Ni surface) and the CO2 molecule (activated by ZrO2), can also be enhanced. The schematic representation is shown in Figure 8. Interestingly, Bacariza et al. investigated the influence of the metal incorporation order in the preparation of Ni-Ce/Y(USY) zeolite catalysts [37]. In their experiments, three ways were used for the preparation of catalysts: Ni before Ce (Ce/Ni), Ce before Ni (Ni/Ce), and co-impregnation (Ni-Ce). Experimental tests showed that the catalytic activity follows the order: Ce/Ni ≈ Ni/Ce < Ni-Ce. TEM and H-TPR characterizations demonstrated that the Ni 0 average size decreases to approximately 2.5 nm, and in terms of Ni/Ce or Ni-Ce catalysts, stronger interactions between Ni and Ce species are established. However, the CO2 adsorption capacity is smaller for Ni/Ce catalyst. In contrast, even though larger Ni 0 particles (13.3 nm) are formed for Ce/Ni catalyst, CO2 adsorption capacity can be enhanced. Finally, Ce-Ni was found as the best preparation method for CO2 methanation using Nibased catalysts supported on CeO2.
In addition, it has been reported that the reduction temperature and a variety of different precursors (e.g., nitrate, chlorate, and oxalate) affect the number of active centers for CO2 methanation [30,38]. As discussed above, the preparation methods showed beneficial influences on the catalytic Interestingly, Bacariza et al. investigated the influence of the metal incorporation order in the preparation of Ni-Ce/Y(USY) zeolite catalysts [37]. In their experiments, three ways were used for the preparation of catalysts: Ni before Ce (Ce/Ni), Ce before Ni (Ni/Ce), and co-impregnation (Ni-Ce). Experimental tests showed that the catalytic activity follows the order: Ce/Ni ≈ Ni/Ce < Ni-Ce. TEM and H-TPR characterizations demonstrated that the Ni 0 average size decreases to approximately 2.5 nm, and in terms of Ni/Ce or Ni-Ce catalysts, stronger interactions between Ni and Ce species are established. However, the CO 2 adsorption capacity is smaller for Ni/Ce catalyst. In contrast, even though larger Ni 0 particles (13.3 nm) are formed for Ce/Ni catalyst, CO 2 adsorption capacity can be enhanced. Finally, Ce-Ni was found as the best preparation method for CO 2 methanation using Ni-based catalysts supported on CeO 2 .
In addition, it has been reported that the reduction temperature and a variety of different precursors (e.g., nitrate, chlorate, and oxalate) affect the number of active centers for CO 2 methanation [30,38]. As discussed above, the preparation methods showed beneficial influences on the catalytic performance, suggesting that reasonable adjustment of the preparation process is a strategy to optimize the experiments for the conversion of CO 2 to CH 4 . Details of CO 2 conversion and product selectivity, along with the reaction conditions of several representative catalytic systems, are compared in Table 2. From Table 2, we can see that Ni-based catalysts are the main catalytic systems for the conversion of CO 2 to CH 4 . Furthermore, the optimal temperature for CH 4 production is 300-400 • C.

CO 2 to CH 3 OH
Methanol is not only an important industrial raw material, but also a stable hydrogen storage compound, which is convenient for transport and reserve. Recently, the concept of "methanol economy" advocated by the Nobel Laureate George Olah revealed the importance of methanol in energy structure [41]. Hence, a large amount of heterogeneous catalysts for the direct hydrogenation of CO 2 to CH 3 OH, reaction (3), have emerged in recent years. Herein, we summarize some catalytic systems for the direct hydrogenation of CO 2 to CH 3 OH (see Figure 9).
Catalysts 2019, 9, x FOR PEER REVIEW 12 of 39 for the conversion of CO2 to CH4. Furthermore, the optimal temperature for CH4 production is 300-400 °C.

CO2 to CH3OH
Methanol is not only an important industrial raw material, but also a stable hydrogen storage compound, which is convenient for transport and reserve. Recently, the concept of "methanol economy" advocated by the Nobel Laureate George Olah revealed the importance of methanol in energy structure [41]. Hence, a large amount of heterogeneous catalysts for the direct hydrogenation of CO2 to CH3OH, reaction (3), have emerged in recent years. Herein, we summarize some catalytic systems for the direct hydrogenation of CO2 to CH3OH (see Figure 9). Cu-ZnO-based catalysts, developed by ICI (Imperial Chemical Industries) in the 1960s, are still widely used in CH3OH synthesis from syngas (CO and H2), and Cu 0 is considered to be the active site. Li et al. reported the synthesis of Cu/ZnO catalysts using a unique method, i.e., facile solid-phase grinding, using mixture of oxalic acid, copper nitrate, and zinc nitrate as raw materials [42]. Appropriate amount of these compounds were physically mixed, and then manually ground for 0.5 h. Subsequently, at 393 K, the obtained precursor was dried for 12 h, and at 623 K, calcined for 3 h in N2 flow, followed by passivation in 1% O2/N2 flow for 5 h. In contrast, by the following steps: (1) at 623 K, calcining the dried precursor at for 3 h in air; (2) at 503 K, reducing the oxide for 10 h in 5% H2/N2 flow; and (3) at room temperature, passivating the catalyst in 1% O2/N2 flow for 5 h, the catalyst can be obtained via H2 reduction. As demonstrated by XRD, H2-TPR and thermal-gravity-differential thermal analysis (TG-DTA), in the facile solid-phase grinding process, the decomposition of oxalate complexes and the reduction of CuO took place simultaneously when the samples were calcined in N2, which prevented the growth of active Cu 0 species and the aggregation of catalyst particles. In contrast, in the conventional H2 reduction process, the growth of active species or the aggregation of catalyst particles were inevitable. Notably, Cu/ZnO catalysts prepared by a facile solid-phase grinding method achieved 29.2% CO2 conversion (at 523 K and 3 MPa), which is even higher than the thermodynamic equilibrium conversion. Indeed, the equilibrium conversion of CO2 at 473 K and 3 MPa is 25.78% according to the Benedict-Webb-Rubin equation [43], and it is even lower at 523 K, since CO2 hydrogenation to CH3OH is an exothermic reaction. Generally, tandem reaction can break the limit of thermodynamic on reaction result. Inspiring from which, the above unusual experimental Cu-ZnO-based catalysts, developed by ICI (Imperial Chemical Industries) in the 1960s, are still widely used in CH 3 OH synthesis from syngas (CO and H 2 ), and Cu 0 is considered to be the active site. Li et al. reported the synthesis of Cu/ZnO catalysts using a unique method, i.e., facile solid-phase grinding, using mixture of oxalic acid, copper nitrate, and zinc nitrate as raw materials [42]. Appropriate amount of these compounds were physically mixed, and then manually ground for 0.5 h. Subsequently, at 393 K, the obtained precursor was dried for 12 h, and at 623 K, calcined for 3 h in N 2 flow, followed by passivation in 1% O 2 /N 2 flow for 5 h. In contrast, by the following steps: (1) at 623 K, calcining the dried precursor at for 3 h in air; (2) at 503 K, reducing the oxide for 10 h in 5% H 2 /N 2 flow; and (3) at room temperature, passivating the catalyst in 1% O 2 /N 2 flow for 5 h, the catalyst can be obtained via H 2 reduction. As demonstrated by XRD, H 2 -TPR and thermal-gravity-differential thermal analysis (TG-DTA), in the facile solid-phase grinding process, the decomposition of oxalate complexes and the reduction of CuO took place simultaneously when the samples were calcined in N 2 , which prevented the growth of active Cu 0 species and the aggregation of catalyst particles. In contrast, in the conventional H 2 reduction process, the growth of active species or the aggregation of catalyst particles were inevitable. Notably, Cu/ZnO catalysts prepared by a facile solid-phase grinding method achieved 29.2% CO 2 conversion (at 523 K and 3 MPa), which is even higher than the thermodynamic equilibrium conversion. Indeed, the equilibrium conversion of CO 2 at 473 K and 3 MPa is 25.78% according to the Benedict-Webb-Rubin equation [43], and it is even lower at 523 K, since CO 2 hydrogenation to CH 3 OH is an exothermic reaction. Generally, tandem reaction can break the limit of thermodynamic on reaction result. Inspiring from which, the above unusual experimental results (higher than thermal equilibrium value) could be most likely achieved through tandem reactions, in which every stepwise reaction was accelerated by different metal sites. Therefore, the simple and solvent-free method based on solid-phase grinding, opened a new approach to synthesize bimetallic or multimetallic catalysts without further reduction.
To improve the catalytic activity of Cu/ZnO/Al 2 O 3 catalysts, ZnO was replaced by g-C 3 N 4 -ZnO hybrid material, as reported by Deng et al. [44]. The experimental results showed that, at 12 bar and 523 K, the methanol space time yield (STY) reached 5.
which is superior to the methanol yield (5. under the same reaction pressure. As confirmed by time-resolved photoluminescence (TRPL) and electronic spin resonance (ESR), the electron-richness of ZnO was enhanced via the formation of type-II hetero-junction between g-C 3 N 4 and ZnO. In addition, in the TPR curve, enhanced SMSI was observed, and the SMSI between electron-rich ZnO and Cu could boost the catalytic performance in CH 3 OH production. The study provided a viable and economic method to modify traditional catalysts for improving CH 3 OH production. For CuO-ZnO-ZrO 2 catalysts, doping graphene oxide (GO) is a good strategy for improving the activity of CO 2 to CH 3 OH [45]. The highest methanol selectivity reached up to 75.88% (473 K, 20 bar) at 1 wt % GO content, while under the same conditions, the methanol selectivity was found to be 68% over the GO-free catalyst. Furthermore, CuO-ZnO-ZrO 2 -GO catalyst exhibited an almost constant space-time yield (STY) of methanol after 96 h time-on stream experiment. As demonstrated by H 2 -TPD and CO 2 -TPD, the CO 2 and H 2 adsorption capacity is enhanced over the CuO-ZnO-ZrO 2 -GO catalysts. Additionally, to explain the improvement of catalytic activity, the authors proposed that GO nano-sheet can serve as a bridge between mixed metal oxides, which strengthens a hydrogen spillover (see Figure 10). That is, H species migrating from the copper surface to the carbon species are adsorbed on the isolated metal oxide particles. simple and solvent-free method based on solid-phase grinding, opened a new approach to synthesize bimetallic or multimetallic catalysts without further reduction.
To improve the catalytic activity of Cu/ZnO/Al2O3 catalysts, ZnO was replaced by g-C3N4-ZnO hybrid material, as reported by Deng et al. [44]. The experimental results showed that, at 12 bar and 523 K, the methanol space time yield (STY) reached 5.73 mmol h −1 gCu −1 for Cu-g-C3N4-ZnO/Al2O3, which is superior to the methanol yield (5.45 mmol h −1 gCu −1 ) of industrial catalyst (Cu-ZnO-Al2O3) under the same reaction pressure. As confirmed by time-resolved photoluminescence (TRPL) and electronic spin resonance (ESR), the electron-richness of ZnO was enhanced via the formation of type-II hetero-junction between g-C3N4 and ZnO. In addition, in the TPR curve, enhanced SMSI was observed, and the SMSI between electron-rich ZnO and Cu could boost the catalytic performance in CH3OH production. The study provided a viable and economic method to modify traditional catalysts for improving CH3OH production.
For CuO-ZnO-ZrO2 catalysts, doping graphene oxide (GO) is a good strategy for improving the activity of CO2 to CH3OH [45]. The highest methanol selectivity reached up to 75.88% (473 K, 20 bar) at 1 wt % GO content, while under the same conditions, the methanol selectivity was found to be 68% over the GO-free catalyst. Furthermore, CuO-ZnO-ZrO2-GO catalyst exhibited an almost constant space-time yield (STY) of methanol after 96 h time-on stream experiment. As demonstrated by H2-TPD and CO2-TPD, the CO2 and H2 adsorption capacity is enhanced over the CuO-ZnO-ZrO2-GO catalysts. Additionally, to explain the improvement of catalytic activity, the authors proposed that GO nano-sheet can serve as a bridge between mixed metal oxides, which strengthens a hydrogen spillover (see Figure 10). That is, H species migrating from the copper surface to the carbon species are adsorbed on the isolated metal oxide particles. It is also reported that modification by small amounts (2 and 5 at %) of WO3 can improve the CO2 conversion and CH3OH selectivity of the CuO-ZnO-ZrO2 catalyst [46]. The optimal catalytic performance can be attributed to the specific surface area of metallic Cu, basic sites, and the reducibility of catalysts. However, if we want to effectively tune the catalytic reaction, a better understanding of the reaction mechanisms is essential. Up to now, the possible reaction pathways for It is also reported that modification by small amounts (2 and 5 at %) of WO 3 can improve the CO 2 conversion and CH 3 OH selectivity of the CuO-ZnO-ZrO 2 catalyst [46]. The optimal catalytic performance can be attributed to the specific surface area of metallic Cu, basic sites, and the reducibility of catalysts. However, if we want to effectively tune the catalytic reaction, a better understanding of the reaction mechanisms is essential. Up to now, the possible reaction pathways for the hydrogenation of CO 2 to methanol over Cu-ZnO-based catalysts are shown in Scheme 3.
Catalysts 2019, 9, x FOR PEER REVIEW 14 of 39 Scheme 3. Possible reaction pathways for CO2 hydrogenation to methanol, where '*X' represents species X adsorbed on a surface site, reprinted with permission from [21]. Copyright Elsevier, 2016.
Besides Cu-based catalysts, In2O3 catalyst with surface oxygen vacancies has attracted more and more attention from researchers. Using density functional theory (DFT) calculations, Ye et al. investigated a In2O3 catalyst for the hydrogenation of CO2 to CH3OH [47]. On a perfect In2O3 (110) surface, six possible surface oxygen vacancies (Ov1 toOv6) were investigated (see Figure 11a), and the D4 surface with the Ov4 defective site was found to be most beneficial for CO2 activation and further hydrogenation. Potential energy profiles of CO2 hydrogenation and protonation on the D4 defective In2O3 (110) surfaces are shown in Figure 11b. In addition, the simulation results showed that the formation of CH3OH replenishes the oxygen vacancy sites, while H2 contributes to generate the vacancies, and this cycle between perfect and defective states of the surface is responsible for the formation of CH3OH from CO2 hydrogenation. To demonstrate this hypothesis, In2O3 catalyst was used for the hydrogenation of CO2 to CH3OH in practice, which showed the superior catalytic activity (7.1% CO2 conversion, 39.7% CH3OH selectivity at 603 K). This experimental result is better than for many other reported catalytic systems, which generally show low selectivity of CH3OH at 603 K. Confirmed by thermo-gravimetric analysis (TGA), XRD, and HR-TEM, the authors found that In2O3 catalyst had satisfactory thermal and structural stability for CO2 conversion to CH3OH below 773 K [48]. Furthermore, to reveal the effect of oxygen vacancies, Martin et al. synthesized bulk In2O3 catalyst, and at a wide range of reaction conditions, the CH3OH selectivity could be tuned up to 100%. XRD, H2-TPR, CO2-TPD, XPS, operando diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS), and electron paramagnetic resonance (EPR) characterization confirmed that oxygen vacancies on the In2O3 catalyst are active sites for the reduction of CO2. Additionally, to further improve the stability of In2O3 catalyst for the production of CH3OH, the authors investigated a variety of supports (i.e., ZrO2, TiO2, ZnO2, SiO2, Al2O3, C, SnO2, MgO). ZrO2 showed the best catalytic performance since it prevented the sintering of the In2O3 phase, which was demonstrated by the enduring stability of In2O3/ZrO2 catalyst over 1000 h [49]. Overall, In2O3 is a potential catalyst for CO2 hydrogenation to CH3OH with high selectivity. Scheme 3. Possible reaction pathways for CO 2 hydrogenation to methanol, where '*X' represents species X adsorbed on a surface site, reprinted with permission from [21]. Copyright Elsevier, 2016.
Besides Cu-based catalysts, In 2 O 3 catalyst with surface oxygen vacancies has attracted more and more attention from researchers. Using density functional theory (DFT) calculations, Ye et al. investigated a In 2 O 3 catalyst for the hydrogenation of CO 2 to CH 3 OH [47]. On a perfect In 2 O 3 (110) surface, six possible surface oxygen vacancies (O v1 toO v6 ) were investigated (see Figure 11a), and the D4 surface with the O v4 defective site was found to be most beneficial for CO 2 activation and further hydrogenation. Potential energy profiles of CO 2 hydrogenation and protonation on the D4 defective In 2 O 3 (110) surfaces are shown in Figure 11b. In addition, the simulation results showed that the formation of CH 3 OH replenishes the oxygen vacancy sites, while H 2 contributes to generate the vacancies, and this cycle between perfect and defective states of the surface is responsible for the formation of CH 3 OH from CO 2 hydrogenation. To demonstrate this hypothesis, In 2 O 3 catalyst was used for the hydrogenation of CO 2 to CH 3 OH in practice, which showed the superior catalytic activity (7.1% CO 2 conversion, 39.7% CH 3 OH selectivity at 603 K). This experimental result is better than for many other reported catalytic systems, which generally show low selectivity of CH 3 OH at 603 K. Confirmed by thermo-gravimetric analysis (TGA), XRD, and HR-TEM, the authors found that In 2 O 3 catalyst had satisfactory thermal and structural stability for CO 2 conversion to CH 3 OH below 773 K [48]. Furthermore, to reveal the effect of oxygen vacancies, Martin et al. synthesized bulk In 2 O 3 catalyst, and at a wide range of reaction conditions, the CH 3 OH selectivity could be tuned up to 100%. XRD, H 2 -TPR, CO 2 -TPD, XPS, operando diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS), and electron paramagnetic resonance (EPR) characterization confirmed that oxygen vacancies on the In 2 O 3 catalyst are active sites for the reduction of CO 2 . Additionally, to further improve the stability of In 2 O 3 catalyst for the production of CH 3 OH, the authors investigated a variety of supports (i.e., ZrO 2 , TiO 2 , ZnO 2 , SiO 2 , Al 2 O 3 , C, SnO 2 , MgO). ZrO 2 showed the best catalytic performance since it prevented the sintering of the In 2 O 3 phase, which was demonstrated by the enduring stability of In 2 O 3 /ZrO 2 catalyst over 1000 h [49]. Overall, In 2 O 3 is a potential catalyst for CO 2 hydrogenation to CH 3 OH with high selectivity. Interestingly, Wang et al. synthesized a series of x% ZnO-ZrO2 solid solution catalysts (x% represents the molar ratio of Zn) for CO2 direct hydrogenation to CH3OH [50]. Under the specified reaction conditions (5.0 MPa, H2/CO2 = 3:1 to 4:1, 593 K to 588 K), ZnO-ZrO2 catalyst can achieve 86-91% methanol selectivity with single-pass CO2 conversion more than 10%, better than the results Interestingly, Wang et al. synthesized a series of x% ZnO-ZrO 2 solid solution catalysts (x% represents the molar ratio of Zn) for CO 2 direct hydrogenation to CH 3 OH [50]. Under the specified reaction conditions (5.0 MPa, H 2 /CO 2 = 3:1 to 4:1, 593 K to 588 K), ZnO-ZrO 2 catalyst can achieve 86-91% methanol selectivity with single-pass CO 2 conversion more than 10%, better than the results reported by other researchers. Moreover, in the presence of 50 ppm SO 2 or H 2 S in the reaction stream, no deactivation was observed. Experimental results confirmed that ZrO 2 and ZnO alone showed little activity in methanol synthesis, while the catalytic performance was significantly enhanced and CO 2 conversion reached the maximum value when the Zn/(Zn + Zr) molar percentage is close to 13%. Demonstrated by CO 2 -TPD, most of the CO 2 adsorbed on the Zr sites of ZnO-ZrO 2 catalyst, and the H 2 -D 2 exchange experiment indicated that ZnO had much higher activity than ZrO 2 . Therefore, in ZnO-ZrO 2 solid solution catalyst, the synergetic effect between the Zn and Zr sites markedly promotes the activation of H 2 and CO 2 . To explain the reaction mechanism on the solid solution catalyst (ZnO-ZrO 2 ), in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) and density functional theory (DFT) calculations were conducted. It was concluded that CO 2 direct hydrogenation to CH 3 OH is dominated via the formate pathway, and CH 3 OH was formed by H 3 CO* protonation on the surface of ZnO-ZrO 2 catalyst.
Although Cu-ZnO-based catalysts are highly selective for CO 2 conversion to methanol, there are still some serious problems during industrial operation, such as deactivation of active sites at high temperature and agglomeration of catalytic particles under industrial reaction conditions. To overcome these issues, researchers designed a series of bimetallic catalysts for the direct hydrogenation of CO 2 to CH 3 OH. Jiang et al. prepared a series of Pd-Cu bimetallic catalysts supported on SiO 2 with a wide range of total metal loading (i.e., 2.4-18.7 wt %) and evaluated the catalytic performance for the reduction of CO 2 to CH 3 OH. With a decrease in metal loadings, the CO 2 conversion dropped stepwise, namely from 6.6 to 3.7%. However, the CH 3 OH STY was slightly enhanced by 31% from 0.16 to 0.21 mmol mol −1 s −1 [51]. To unclose the reaction mechanisms, the authors carried out in situ diffuse reflectance FT-IR (DRIFTS) measurements on Pd(0.34)-Cu/SiO 2 catalyst [52]. The resulting spectra identified that the dominant species on a bimetallic surface were formate and carbonyl species on a bimetallic surface, which were dependent on the catalyst composition. Therefore, they proposed that on Pd-Cu catalysts, the surface coverage of formate species was correlated to the methanol promotion, implying its vital role in CH 3 OH synthesis. Bahruji et al. prepared PdZn/TiO 2 bimetallic catalysts by a solvent-free chemical vapor impregnation method [53]. According to the order in which the metals were impregnated, the catalysts could be classified as 2 Pd-1 Zn-TiO 2 , 2 Zn-1 Pd-TiO 2 , and PdZn/TiO 2 (1 and 2 represent the order of sequential metal impregnations). The experimental results showed that PdZn/TiO 2 catalyst achieved optimal catalytic performance, 10.1% CO 2 conversion and 40% CH 3 OH selectivity. The formation of ZnTiO 3 and PdZn nanoparticles was confirmed via XPS, XRD, and TEM. Combining the experimental results, the authors proposed that PdZn nanoparticles were beneficial for methanol formation and catalytic stability. Although it is hard to assign the formation of the PdZn alloy to the differences between the preparation methods, an interaction between Pd(acac) 2 and Zn(acac) 2 precursors might be responsible for the production of the PdZn active site in terms of PdZn/TiO 2 catalyst. Additionally, Yin et al. reported the preparation of Pd@zeolitic imidazolate framework-8 (ZIF-8) catalyst for the conversion of CO 2 to CH 3 OH [54]. The optimal methanol yield can reach 0.65 g g cat −1 h −1 over a PdZn catalyst. As demonstrated by XPS, XRD, TEM and electron paramagnetic resonance (EPR), after H 2 reduction, PdZn alloy particles formed, and abundant oxygen defects existed on the ZnO surface. The experimental results revealed that the active site is a PdZn alloy rather than metallic Pd, in terms of CH 3 OH formation. Numerous studies suggested that promoters have an indispensable role in the catalytic performance of CuZn catalysts [55], although the mechanism of action still remains unclear. Pd-doped CuZn catalysts were prepared to evaluate the surface modification effect. An interesting observation was made from the volcano-shaped relationship between CH 3 OH STY and Pd loading, which implied that an appropriate amount of Pd loading is beneficial for methanol synthesis. The chemisorption analyses further revealed a strong interaction between Pd and Cu surface, where a hydrogen spillover has taken place, generating more activated Cu sites. Hence, the CH 3 OH STY and the methanol turnover frequency (TOF) improved a lot with Pd loading of 1 wt % [55]. In addition, a variety of bimetallic catalysts (e.g., Cu-Fe, Cu-Co, Cu-Ni, Co-Ga, Ni-Ga) have been evaluated by many researchers [56][57][58][59]. A possible catalytic reaction mechanism of CO 2 hydrogenation over Cu−Ni/CeO 2 −NT is shown in Figure 12. Details of the CO 2 conversion and product selectivity, along with reaction conditions of several representative catalytic systems are compared in Table 3. In terms of hydrogenation of CO 2 to CH 3 OH, In-based and Pd-based catalysts have gradually drawn researchers' attention, besides traditional Cu-based catalysts. Additionally, many researchers demonstrated that 250 • C is the optimal temperature to produce CH 3 OH with the high selectivity. According to the current experimental results, increasing the reaction pressure is also a good strategy to improve CO 2 conversion, but a high pressure means a high cost of operation and equipment. Cu−Ni/CeO2−NT is shown in Figure 12. Details of the CO2 conversion and product selectivity, along with reaction conditions of several representative catalytic systems are compared in Table 3. In terms of hydrogenation of CO2 to CH3OH, In-based and Pd-based catalysts have gradually drawn researchers' attention, besides traditional Cu-based catalysts. Additionally, many researchers demonstrated that 250 °C is the optimal temperature to produce CH3OH with the high selectivity. According to the current experimental results, increasing the reaction pressure is also a good strategy to improve CO2 conversion, but a high pressure means a high cost of operation and equipment.

CO 2 to Other Products
Besides CO, CH 4 , and CH 3 OH, dimethyl ether (DME), light olefins [71][72][73][74], alcohol [75], isoparaffins [76], and aromatics [77,78] have also been produced by CO 2 hydrogenation. Clearly, in terms of CO 2 conversion, the coupling of C-C bond to produce higher hydrocarbons and oxygenates is a technological barrier. However, taking advantage of tandem catalysis to realize one-step synthesis of hydrocarbons via hydrogenation of CO 2 is feasible. The extensive studies can be mainly categorized into two categories: methanol-mediated and non-methanol-mediated reactions [79,80]. In the methanol-mediated approach, DME is usually produced as main product, while for the non-methanol-mediated approach, alkenes and alkanes are generally produced as main products. The possible reaction mechanism for CO 2 hydrogenation to DME and light olefins is shown in Scheme 4.

CO2 to Other Products
Besides CO, CH4, and CH3OH, dimethyl ether (DME), light olefins [71][72][73][74], alcohol [75], isoparaffins [76], and aromatics [77,78] have also been produced by CO2 hydrogenation. Clearly, in terms of CO2 conversion, the coupling of C-C bond to produce higher hydrocarbons and oxygenates is a technological barrier. However, taking advantage of tandem catalysis to realize one-step synthesis of hydrocarbons via hydrogenation of CO2 is feasible. The extensive studies can be mainly categorized into two categories: methanol-mediated and non-methanol-mediated reactions [79,80]. In the methanol-mediated approach, DME is usually produced as main product, while for the nonmethanol-mediated approach, alkenes and alkanes are generally produced as main products. The possible reaction mechanism for CO2 hydrogenation to DME and light olefins is shown in Scheme 4. Cu-based catalysts are in practice used for methanol synthesis, and HZSM-5 zeolites are widely employed for methanol dehydration due to its solid acid catalysis. Therefore, a variety of studies have been reported with regard to the bi-functional catalysts consisting of Cu and HZSM-5 used for the direct hydrogenation of CO2.
Zhang et al. reported a Cu-ZrO2/HZSM-5 catalyst promoted by Pd/CNT for direct synthesis of DME from CO2/H2 [79]. Although a minor amount of the Pd-decorated CNTs into the CuZr/HZSM-5 catalytic system caused little change in the activation energy for CO2 conversion compared with the Scheme 4. Possible reaction pathways for CO 2 hydrogenation to DME and light olefins. Note: ---H, ---O derived from H-ZSM5 zeolite, reprinted with permission from [21]. Copyright Elsevier, 2016.
Cu-based catalysts are in practice used for methanol synthesis, and HZSM-5 zeolites are widely employed for methanol dehydration due to its solid acid catalysis. Therefore, a variety of studies have been reported with regard to the bi-functional catalysts consisting of Cu and HZSM-5 used for the direct hydrogenation of CO 2 .
Zhang et al. reported a Cu-ZrO 2 /HZSM-5 catalyst promoted by Pd/CNT for direct synthesis of DME from CO 2 /H 2 [79]. Although a minor amount of the Pd-decorated CNTs into the CuZr/HZSM-5 catalytic system caused little change in the activation energy for CO 2 conversion compared with the CuZr/HZSM-5 catalyst, the former created a micro-environment including higher concentration of active H species and adsorbed CO 2 species on the surface of the catalyst system. Owing to the increase of hydrogenation reactions, under reaction conditions of 5 MPa and 523 K, the specific rate of CO 2 hydrogenation-conversion was 1.22 times higher than the pure CuZr/HZSM-5 catalyst. Furthermore, Zhang et al. reported a series of V-modified CuO-ZnO-ZrO 2 /HZSM-5 catalysts which were prepared via an oxalate co-precipitation method for the synthesis of DME from CO 2 /H 2 [81]. The catalytic performances of the catalysts were strongly dependent on the content of V, which was confirmed by XRD, N 2 O chemisorption, and X-ray photoelectron spectroscopy (XPS). Similarly, the influence of the promoter-i.e., La and W-has also been examined. The optimal catalytic activity was obtained (43.8% CO 2 conversion and 71.2% DME selectivity) when the amount of La was 2 wt %. In contrast, the Cu-ZnO-ZrO 2 catalyst admixed with WO x obtained 18.9% CO 2 conversion and 15.3% DME selectivity. Obviously, La is a better promoter compared with W, and this can be explained by the fact that the reducibility and dispersion of bi-functional catalysts (CuO-ZnO-Al 2 O 3 -La 2 O 3 /HZSM-5) were largely dependent on the modification of La, while the hybrid catalyst (Cu-ZnO-ZrO 2 -WO x /Al 2 O 3 ) modified by W strongly adsorbed water molecules, resulting in a lower catalytic performance. Moreover, as shown in Figure 13, the STY of DME of different WO x /Al 2 O 3 catalysts (i.e., on supports with different pore sizes) as a function of W surface density shows a volcanic curve relation. Details of CO 2 conversion and DME selectivity reported recently are shown in Table 4 [82,83]. The above studies indicate that the current catalysts used for DME preparation from CO 2 /H 2 are still dominated by Cu-based-HZSM-5 catalysts. Some other zeolite catalysts (SBA-15, SAPO-5, SUZ-4) with similar acid properties (acid content and acid strength) as HZSM-5 zeolite may be a direction of future research. CuZr/HZSM-5 catalyst, the former created a micro-environment including higher concentration of active H species and adsorbed CO2 species on the surface of the catalyst system. Owing to the increase of hydrogenation reactions, under reaction conditions of 5 MPa and 523 K, the specific rate of CO2 hydrogenation-conversion was 1.22 times higher than the pure CuZr/HZSM-5 catalyst. Furthermore, Zhang et al. reported a series of V-modified CuO-ZnO-ZrO2/HZSM-5 catalysts which were prepared via an oxalate co-precipitation method for the synthesis of DME from CO2/H2 [81]. The catalytic performances of the catalysts were strongly dependent on the content of V, which was confirmed by XRD, N2O chemisorption, and X-ray photoelectron spectroscopy (XPS). Similarly, the influence of the promoter-i.e., La and W-has also been examined. The optimal catalytic activity was obtained (43.8% CO2 conversion and 71.2% DME selectivity) when the amount of La was 2 wt %. In contrast, the Cu-ZnO-ZrO2 catalyst admixed with WOx obtained 18.9% CO2 conversion and 15.3% DME selectivity. Obviously, La is a better promoter compared with W, and this can be explained by the fact that the reducibility and dispersion of bi-functional catalysts (CuO-ZnO-Al2O3-La2O3/HZSM-5) were largely dependent on the modification of La, while the hybrid catalyst (Cu-ZnO-ZrO2-WOx/Al2O3) modified by W strongly adsorbed water molecules, resulting in a lower catalytic performance. Moreover, as shown in Figure 13, the STY of DME of different WOx/Al2O3 catalysts (i.e., on supports with different pore sizes) as a function of W surface density shows a volcanic curve relation. Details of CO2 conversion and DME selectivity reported recently are shown in  Among the large-scale technologies with industrial potential, the conversion of CO2 to DME is promising and relatively mature [84]. Korea Gas Corporation (KOGAS) has developed a DME plant, with CO2 as a raw material. The schematic diagram of the KOGAS tri-reforming process is shown in Figure 14 [85,86]. Kansai Electric Power Co. and Mitsubishi Heavy Industries have realized a benchscale (100 cm 3 catalyst loading) experiment for DME synthesis [87,88]. However, during the reaction process, water formation decreased the yield of DME [89][90][91]. Catizzone et al. and Bonura et al. Figure 13. STY of DME of different WO x /Al 2 O 3 catalysts, with different support pore size as a function of W surface density. AS, AM, and AL refer to the mean pore size of alumina supports with small, medium and large pores, respectively, reproduced with permission from [83]. Copyright Elsevier, 2018.
Among the large-scale technologies with industrial potential, the conversion of CO 2 to DME is promising and relatively mature [84]. Korea Gas Corporation (KOGAS) has developed a DME plant, with CO 2 as a raw material. The schematic diagram of the KOGAS tri-reforming process is shown in Figure 14 [85,86]. Kansai Electric Power Co. and Mitsubishi Heavy Industries have realized a bench-scale (100 cm 3 catalyst loading) experiment for DME synthesis [87,88]. However, during the reaction process, water formation decreased the yield of DME [89][90][91]. Catizzone et al. and Bonura et al. evidenced that zeolites or ferrierite could effectively mitigate the influence of water and avoid catalyst sintering [89,90]. In a fixed bed reactor, kinetic modeling confirmed the negative effect of water (formed during the reaction process) on DME production, which is consistent with the experimental results. Therefore, Falco et al. suggested that hydrophilic membranes could be promising for industrial production of DME [92]. Recently, Fang et al. advocated CO 2 capture and conversion by using a membrane reactor system, where a high-temperature mixed electronic and carbonate-ion conductor (MECC) membrane was used for CO 2 capture and a solid oxide electrolysis cell (SOEC) was used for CO 2 reduction [93]. Furthermore, based on membrane technology, Sofia et al. carried out a techno-economic analysis project of power and hydrogen co-production via an integrated gasification combined cycle (IGCC) plant with CO 2 capture [94]. The development of membrane technology would be an advantageous factor for the capture and utilization of CO 2 .
Catalysts 2019, 9, x FOR PEER REVIEW 20 of 39 evidenced that zeolites or ferrierite could effectively mitigate the influence of water and avoid catalyst sintering [89,90]. In a fixed bed reactor, kinetic modeling confirmed the negative effect of water (formed during the reaction process) on DME production, which is consistent with the experimental results. Therefore, Falco et al. suggested that hydrophilic membranes could be promising for industrial production of DME [92]. Recently, Fang et al. advocated CO2 capture and conversion by using a membrane reactor system, where a high-temperature mixed electronic and carbonate-ion conductor (MECC) membrane was used for CO2 capture and a solid oxide electrolysis cell (SOEC) was used for CO2 reduction [93]. Furthermore, based on membrane technology, Sofia et al. carried out a techno-economic analysis project of power and hydrogen co-production via an integrated gasification combined cycle (IGCC) plant with CO2 capture [94]. The development of membrane technology would be an advantageous factor for the capture and utilization of CO2. Li et al. synthesized a novel ZnZrO/SAPO tandem catalyst, combined by a ZnO-ZrO2 solid solution and a Zn-modified SAPO-34 zeolite, which achieved 80-90% olefin selectivity among the hydrocarbon products [95]. Based on the surface reaction kinetics, they proposed that the tandem reaction process proceeded as follows: (1) generation of CHxO species on ZnZrO via CO2 reduction; (2) olefins production from the derived CHxO species which migrate/transfer onto SAPO zeolite pore structure. Moreover, experimental results confirmed that the excellent selectivity can be ascribed to the effective synergy between ZnZrO and SAPO for the tandem catalyst. In addition, this catalyst (ZnZrO and SAPO-34) shows an excellent stability toward thermal and sulfur treatments, indicating the potential value for industrial application.
Recently, CO2 was converted to aromatics with a selectivity up to 73%, at 14% CO2 conversion over ZnZrO/HZSM-5 catalyst [77]. Demonstrated by operando infrared (IR) characterization, Li et al. proposed that CHxO, as an intermediate species, transformed from the ZnZrO surface into the pore structure of HZSM-5, which is responsible for C-C bond formation for aromatics production. The reaction scheme is shown in Figure 15. Interestingly, the presence of H2O and CO2 markedly suppressed the generation of polycyclic aromatics, consequently, enhanced the stability (100 h in the reaction stream) of the tandem catalyst, which showed potential industrial application prospects. Similarly, Ni et al. synthesized a tandem catalyst of ZnAlOx and HZSM-5, which yields 73.9% aromatics selectivity and 0.4% CH4 selectivity among the carbon products without CO [78]. Confirmed by XRD, SEM, TEM, and element distribution analysis, ZnAlOx was formed and uniformly dispersed in the tandem catalyst. Furthermore, demonstrated by 2,6-di-tert-butyl-pyridine Li et al. synthesized a novel ZnZrO/SAPO tandem catalyst, combined by a ZnO-ZrO 2 solid solution and a Zn-modified SAPO-34 zeolite, which achieved 80-90% olefin selectivity among the hydrocarbon products [95]. Based on the surface reaction kinetics, they proposed that the tandem reaction process proceeded as follows: (1) generation of CH x O species on ZnZrO via CO 2 reduction; (2) olefins production from the derived CH x O species which migrate/transfer onto SAPO zeolite pore structure. Moreover, experimental results confirmed that the excellent selectivity can be ascribed to the effective synergy between ZnZrO and SAPO for the tandem catalyst. In addition, this catalyst (ZnZrO and SAPO-34) shows an excellent stability toward thermal and sulfur treatments, indicating the potential value for industrial application.
Recently, CO 2 was converted to aromatics with a selectivity up to 73%, at 14% CO 2 conversion over ZnZrO/HZSM-5 catalyst [77]. Demonstrated by operando infrared (IR) characterization, Li et al. proposed that CH x O, as an intermediate species, transformed from the ZnZrO surface into the pore structure of HZSM-5, which is responsible for C-C bond formation for aromatics production. The reaction scheme is shown in Figure 15. Interestingly, the presence of H 2 O and CO 2 markedly suppressed the generation of polycyclic aromatics, consequently, enhanced the stability (100 h in the reaction stream) of the tandem catalyst, which showed potential industrial application prospects. Similarly, Ni et al. synthesized a tandem catalyst of ZnAlO x and HZSM-5, which yields 73.9% aromatics selectivity and 0.4% CH 4 selectivity among the carbon products without CO [78]. Confirmed by XRD, SEM, TEM, and element distribution analysis, ZnAlOx was formed and uniformly dispersed in the tandem catalyst. Furthermore, demonstrated by 2,6-di-tert-butyl-pyridine absorption (DTBPy-FTIR), the external Brønsted acid of HZSM-5 can be shielded by ZnAlO x , which is beneficial to aromatization. According to the operando diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) study, they proposed the following possible reaction mechanism: MeOH and DME, produced by hydrogenation of formate species, are transmitted to HZSM-5 and then converted into olefins and finally aromatics. absorption (DTBPy-FTIR), the external Brønsted acid of HZSM-5 can be shielded by ZnAlOx, which is beneficial to aromatization. According to the operando diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) study, they proposed the following possible reaction mechanism: MeOH and DME, produced by hydrogenation of formate species, are transmitted to HZSM-5 and then converted into olefins and finally aromatics. Figure 15. Reaction scheme of CO2 hydrogenation to aromatics, reproduced with permission from [77]. Copyright Elsevier, 2019.
Non-methanol-mediated reactions, i.e., the direct hydrogenation of CO2 to light olefins is even more significant than CH3OH synthesis from CO2 hydrogenation, since a large proportion of methanol is used for the synthesis of olefins in industry, through the methanol-to-olefins (MTO) reaction by using SAPO-34 zeolite catalysts. CO2 to lower olefins can be realized by coupling of the RWGS reaction and F-T synthesis, as shown in Schemes 1 and 3.
Iron-based catalysts has been considered as an excellent option for the synthesis of light olefins from CO2/H2, mainly because of their excellent activity, high selectivity, and low price. Using honeycomb-structure graphene (HSG) as the support and K as a promoter, Wu et al. prepared an iron-based catalyst [80]. They found out that the confinement effect of the porous HSG was beneficial for the sintering of the Fe active sites, and within 120 h stability test, no significant deactivation occurred. In addition, as confirmed by CO2 temperature programmed desorption (CO2-TPD) and H2-TPD, they found out that potassium as a promoter effectively enhanced the chemisorption and activation of the reactants CO2 and H2. Moreover, as revealed by 57 Fe Mossbauer absorption spectroscopy, for the Fe/HSG catalyst, there were two doublets with isomer shift (IS) of 0.96 mm s −1 and quadrupole splitting (QS) of 0.28 mm s −1 and IS of 1.21 mm s −1 and QS of 1.82 mm s −1 , which corresponded to the Fe (II) species in low-and high-coordination environments. Instead, for the FeK1.5/HSG catalyst, only one doublet with IS of 0.31 mm s −1 and QS of 1.05 mm s −1 ascribed to the Fe (III) species, which implies that K is capable of stabilizing high valence-state iron (Fe (III)) during CO2 hydrogenation to light olefins (CO2-FTO). The above mentioned three factors contributed to the Non-methanol-mediated reactions, i.e., the direct hydrogenation of CO 2 to light olefins is even more significant than CH 3 OH synthesis from CO 2 hydrogenation, since a large proportion of methanol is used for the synthesis of olefins in industry, through the methanol-to-olefins (MTO) reaction by using SAPO-34 zeolite catalysts. CO 2 to lower olefins can be realized by coupling of the RWGS reaction and F-T synthesis, as shown in Schemes 1 and 3.
Iron-based catalysts has been considered as an excellent option for the synthesis of light olefins from CO 2 /H 2 , mainly because of their excellent activity, high selectivity, and low price. Using honeycomb-structure graphene (HSG) as the support and K as a promoter, Wu et al. prepared an iron-based catalyst [80]. They found out that the confinement effect of the porous HSG was beneficial for the sintering of the Fe active sites, and within 120 h stability test, no significant deactivation occurred. In addition, as confirmed by CO 2 temperature programmed desorption (CO 2 -TPD) and H 2 -TPD, they found out that potassium as a promoter effectively enhanced the chemisorption and activation of the reactants CO 2 and H 2 . Moreover, as revealed by 57 Fe Mossbauer absorption spectroscopy, for the Fe/HSG catalyst, there were two doublets with isomer shift (IS) of 0.96 mm s −1 and quadrupole splitting (QS) of 0.28 mm s −1 and IS of 1.21 mm s −1 and QS of 1.82 mm s −1 , which corresponded to the Fe (II) species in low-and high-coordination environments. Instead, for the FeK1.5/HSG catalyst, only one doublet with IS of 0.31 mm s −1 and QS of 1.05 mm s −1 ascribed to the Fe (III) species, which implies that K is capable of stabilizing high valence-state iron (Fe (III)) during CO 2 hydrogenation to light olefins (CO 2 -FTO). The above mentioned three factors contributed to the excellent catalytic performance, i.e., 56% olefins selectivity and a 120-h stability testing experiment (as shown in Figure 16). excellent catalytic performance, i.e., 56% olefins selectivity and a 120-h stability testing experiment (as shown in Figure 16). Zhang et al. used an impregnation method to prepare Fe-Zn-K catalysts, in which K was used as a promoter [71]. The experimental results showed that the Fe-Zn-K catalyst with H2/CO reduction showed the best catalytic activity, with 51.03% CO2 conversion and 53.58% C2-C4 olefins selectivity, at the reaction conditions of 593 K and 0.5 MPa. XRD, H2-TPR, and XPS characterization results revealed that, in the Fe-Zn-K catalyst, ZnFe2O4 spinel phase and ZnO phase were formed. Among them, ZnFe2O4 spinel phase strengthens the interaction between iron and zinc, and changed the reduction and CO2 adsorption behaviors. In addition, the H2-TPR profiles show that the catalyst modified by K contributed to a slight shift of the initial reduction peak to higher temperature, indicating the reduction of Fe2O3 and formation of Fe phase was inhibited by K modification.
You et al. investigated the catalytic activity of non-supported Fe catalysts (bulk Fe catalysts) modified by alkali metal ions (i.e., Li, Na, K, Rb, Cs) for the conversion of CO2 to light olefins [72]. By calcining ammonium ferric citrate, non-supported Fe catalyst was prepared. Compared with Fe catalyst without modification (5.6% CO2 conversion, 0% olefins), the modification of the Fe catalyst with an alkali metal ion markedly enhanced the catalytic activity for CO2 conversion to light olefins. For Fe catalyst modified by K and Rb, the conversion of CO2 and the olefin selectivity (based on only the hydrocarbon compounds, without CO) increased to about 40% and 50%, respectively, and the yield of light (C2-C4) olefins can reach 10-12%. Further investigation via XRD showed that over the alkali-metal-ion-modified Fe catalysts, Fe5C2 was formed, while in the unmodified catalyst, iron carbide species were not observed after the reaction. Accordingly, they proposed that in the presence of an alkali metal ion, the generation of iron carbide species was one possible reason for the enhanced catalytic activity. Therefore, modification by alkali metal ions remains a good strategy to tune the product distribution.
Additionally, Wei et al. reported a highly efficient, stable and multifunctional Na-Fe3O4/HZSM-5 catalytic system [107], for direct hydrogenation of CO2 to gasoline-range (C5-C11), in which the selectivity of C5-C11 hydrocarbons reached 78% (based on total hydrocarbons), while under industrial conditions only reached 4% methane selectivity at a 22% CO2 conversion. Characterization by high resolution transmission electron microscopy (HRTEM), X-ray diffraction (XRD), and Mossbauer spectroscopy showed that two different types of iron phase-i.e., Fe3O4 and x-Fe5C2-were discerned Zhang et al. used an impregnation method to prepare Fe-Zn-K catalysts, in which K was used as a promoter [71]. The experimental results showed that the Fe-Zn-K catalyst with H 2 /CO reduction showed the best catalytic activity, with 51.03% CO 2 conversion and 53.58% C 2 -C 4 olefins selectivity, at the reaction conditions of 593 K and 0.5 MPa. XRD, H 2 -TPR, and XPS characterization results revealed that, in the Fe-Zn-K catalyst, ZnFe 2 O 4 spinel phase and ZnO phase were formed. Among them, ZnFe 2 O 4 spinel phase strengthens the interaction between iron and zinc, and changed the reduction and CO 2 adsorption behaviors. In addition, the H 2 -TPR profiles show that the catalyst modified by K contributed to a slight shift of the initial reduction peak to higher temperature, indicating the reduction of Fe 2 O 3 and formation of Fe phase was inhibited by K modification.
You et al. investigated the catalytic activity of non-supported Fe catalysts (bulk Fe catalysts) modified by alkali metal ions (i.e., Li, Na, K, Rb, Cs) for the conversion of CO 2 to light olefins [72]. By calcining ammonium ferric citrate, non-supported Fe catalyst was prepared. Compared with Fe catalyst without modification (5.6% CO 2 conversion, 0% olefins), the modification of the Fe catalyst with an alkali metal ion markedly enhanced the catalytic activity for CO 2 conversion to light olefins. For Fe catalyst modified by K and Rb, the conversion of CO 2 and the olefin selectivity (based on only the hydrocarbon compounds, without CO) increased to about 40% and 50%, respectively, and the yield of light (C 2 -C 4 ) olefins can reach 10-12%. Further investigation via XRD showed that over the alkali-metal-ion-modified Fe catalysts, Fe 5 C 2 was formed, while in the unmodified catalyst, iron carbide species were not observed after the reaction. Accordingly, they proposed that in the presence of an alkali metal ion, the generation of iron carbide species was one possible reason for the enhanced catalytic activity. Therefore, modification by alkali metal ions remains a good strategy to tune the product distribution.
Additionally, Wei et al. reported a highly efficient, stable and multifunctional Na-Fe 3 O 4 /HZSM-5 catalytic system [76], for direct hydrogenation of CO 2 to gasoline-range (C 5 -C 11 ), in which the selectivity of C 5 -C 11 hydrocarbons reached 78% (based on total hydrocarbons), while under industrial conditions only reached 4% methane selectivity at a 22% CO 2 conversion. Characterization by high resolution transmission electron microscopy (HRTEM), X-ray diffraction (XRD), and Mossbauer spectroscopy showed that two different types of iron phase-i.e., Fe 3 O 4 and x-Fe 5 C 2 -were discerned in the spent Na-Fe 3 O 4 catalyst, which cooperatively catalyzed a tandem reaction (RWGS and FT).
The possible reaction scheme for CO 2 hydrogenation to gasoline-range hydrocarbons is shown in Figure 17. Furthermore, acid sites existing in the HZSM-5 zeolite were favorable for acid-catalyzed reactions (oligomerization, isomerization, and aromatization). Notably, the multifunctional catalyst exhibited a significant stability for 1000 h on stream, showing the potential as promising industrial catalyst material for CO 2 conversion to liquid fuels. Similarly, Gao et al. investigated a bi-functional catalytic system composed of In 2 O 3 and HZSM-5, which can achieve 78.6% C 5+ selectivity with only 1% CH 4 at a 13.1% CO 2 conversion [73]. As demonstrated by Ye et al. [47], CO 2 and H 2 can be activated in the oxygen vacancies on the In 2 O 3 surfaces, and catalyzed CH 3 OH formation. Subsequently, C−C coupling occurred inside the zeolite pores structure (HZSM-5) to synthesize gasoline-range hydrocarbons with a relatively high octane number (C 5+ ). in the spent Na-Fe3O4 catalyst, which cooperatively catalyzed a tandem reaction (RWGS and FT). The possible reaction scheme for CO2 hydrogenation to gasoline-range hydrocarbons is shown in Figure  17. Furthermore, acid sites existing in the HZSM-5 zeolite were favorable for acid-catalyzed reactions (oligomerization, isomerization, and aromatization). Notably, the multifunctional catalyst exhibited a significant stability for 1000 h on stream, showing the potential as promising industrial catalyst material for CO2 conversion to liquid fuels. Similarly, Gao et al. investigated a bi-functional catalytic system composed of In2O3 and HZSM-5, which can achieve 78.6% C5+ selectivity with only 1% CH4 at a 13.1% CO2 conversion [73]. As demonstrated by Ye et al. [47], CO2 and H2 can be activated in the oxygen vacancies on the In2O3 surfaces, and catalyzed CH3OH formation. Subsequently, C−C coupling occurred inside the zeolite pores structure (HZSM-5) to synthesize gasoline-range hydrocarbons with a relatively high octane number (C5+). Details of the conversion and selectivity for the hydrogenation of CO2 into other products, along with the reaction conditions of several representative catalytic systems, are compared in Table 4. Febased catalysts are beneficial for the production of olefins, while Cu-based + HZSM-5 is still a dominant catalytic system for CO2 conversion to DME. Similarly, 250 °C is the optimal temperature for the production of DME based on the current experimental results, and 300-400 °C, which is close to industrial production conditions, seems to be better for the direct conversion of CO2 to olefins. On the other hand, the selectivity of DME can be maintained upon increasing GHSV (>10,000 mL gcat −1 h −1 ), but the high selectivity of olefins generally relies on a relatively low GHSV (<5000 mL gcat −1 h −1 ). Details of the conversion and selectivity for the hydrogenation of CO 2 into other products, along with the reaction conditions of several representative catalytic systems, are compared in Table 4. Fe-based catalysts are beneficial for the production of olefins, while Cu-based + HZSM-5 is still a dominant catalytic system for CO 2 conversion to DME. Similarly, 250 • C is the optimal temperature for the production of DME based on the current experimental results, and 300-400 • C, which is close to industrial production conditions, seems to be better for the direct conversion of CO 2 to olefins. On the other hand, the selectivity of DME can be maintained upon increasing GHSV (>10,000 mL g cat −1 h −1 ), but the high selectivity of olefins generally relies on a relatively low GHSV (<5000 mL g cat −1 h −1 ). Table 4. Catalytic activity of several catalysts for CO 2 hydrogenation into other products (e.g., DME, olefins, alcohol, isoparaffins, gasoline, aromatics), in terms of CO 2 conversion and product selectivity, along with the reaction conditions.

Opportunities of Heterogeneous Catalysis for CO 2 Conversion
The relationship between products selectivity and CO 2 conversion by heterogeneous catalytic hydrogenation is shown in Figure 18. It is obvious that, in the RWGS reaction, although the CO selectivity reaches up to nearly 100%, the CO 2 conversion is low. In the methanation reaction of CO 2 , the CH 4 selectivity is high enough, and the CO 2 conversion also exceeds 50%. With regard to the synthesis of CH 3 OH, DME, and light olefins, the relationship between conversion and selectivity is less clear. In the RWGS reaction, CO is mainly produced over Fe-, Co-, Mo-, and Pt-based catalysts (Table  1), while CO2 methanation is typically carried out on Co-and Ni-based catalysts ( Table 2). Although CO is a main component of syngas and CH4 is an important energy resource, neither is the best product for CO2 conversion from a relatively economic point of view. Furthermore, the conversion and utilization of CO and CH4 is also an important research field, which also implies that CO and CH4 are not the optimal choice as end-products of CO2 conversion. Therefore, the coupling reaction is a potential direction for CO2 conversion to some higher value products, i.e., CH3OH and light olefins.
Cu-ZnO-based catalyst is still the main catalytic material for direct conversion of CO2 to CH3OH In the RWGS reaction, CO is mainly produced over Fe-, Co-, Mo-, and Pt-based catalysts (Table 1), while CO 2 methanation is typically carried out on Co-and Ni-based catalysts ( Table 2). Although CO is a main component of syngas and CH 4 is an important energy resource, neither is the best product for CO 2 conversion from a relatively economic point of view. Furthermore, the conversion and utilization of CO and CH 4 is also an important research field, which also implies that CO and CH 4 are not the optimal choice as end-products of CO 2 conversion. Therefore, the coupling reaction is a potential direction for CO 2 conversion to some higher value products, i.e., CH 3 OH and light olefins.
Cu-ZnO-based catalyst is still the main catalytic material for direct conversion of CO 2 to CH 3 OH under industrial reaction condition, while highly innovative methodologies are awaited to achieve low-pressure and low-temperature CH 3 OH synthesis processes. Considering the industrial application value of methanol, exploring novel catalytic systems and designing rational reactors to improve the catalytic activity should be targeted in the future. DME, up to now, is synthesized via Cu-based and H-ZSM5 catalyst (Table 4), and this process essentially remains a two-step tandem reaction. Hence, the bottleneck of DME production could be the deactivation of CH 3 OH dehydration catalyst, due to water poisoning. The formation of coke is also the major reason of catalyst deactivation, especially in a relatively long-term operation process. To maintain high activity of catalyst for the production of DME, water-and coke-resistant catalysts need to be developed for CH 3 OH dehydration.
As shown in Table 4, the direct hydrogenation of CO 2 to light olefins is mainly catalyzed by Fe-based catalysts. However, the starting point of current research work is mainly the coupling of RWGS, F-T and/or methanol-to-olefins (MTO) reactions. Therefore, suitable catalytic materials are sought for the coupling of C-C bonds, which is an effective strategy for improving the catalytic activity.

Plasma Catalysis
Plasma, the 'fourth state of matter', consists of electrons, neutral species (i.e., molecules, radicals, and excited species) and ions. Plasma can be in so-called thermal equilibrium or not, based on which it is subdivided into 'thermal plasma' and 'non-thermal plasma' (NTP) [13,112,113]. In non-thermal plasma, the gas temperature remains near room temperature, while electrons temperature is extremely high, usually in the range of 1-10 eV (~10,000-100,000 K). The latter is enough to activate stable gas molecules into reactive species (e.g., radicals, excited atoms, molecules, and ions). These reactive species, especially the radicals, can trigger reactions at low temperature. That is, NTP offers a unique approach to enable thermodynamically unfavorable chemical reactions to proceed at low temperature by breaking thermodynamic limits. Nevertheless, the control of selectivity of desired products in plasma is extremely difficult, since the reactions in plasma are mainly triggered through nonselective collision between active species (radicals, molecules, atoms, and ions).
To improve the desired product selectivity of the reactions in plasma, the combination of catalysts with plasma technology (so-called plasma catalysis) is a promising strategy, since catalysts usually have a special feature of regulating product distribution.
NTP can be generated through various types of discharges-i.e., microwave discharges, glow discharges, gliding arc discharges, dielectric barrier discharges (DBD), etc.-but DBD are the best option to be used in plasma catalysis. Indeed, DBDs are usually operated at atmospheric pressure and ambient temperature, and the integration of DBD plasma with catalysts has the advantages of simple operation and low cost. Thus plasma-driven direct hydrogenation of CO 2 is mostly based on DBD plasmas. The possible plasma/catalyst synergism is illustrated in Figure 19 [13].
A lot of research has been performed for pure CO 2 splitting, in various types of plasma reactors, including DBD, microwave discharge and gliding arc discharge, without catalysts [114][115][116][117]. A typical experimental set-up of a DBD plasma for CO 2 decomposition is shown in Figure 20. Paulussen et al. studied the conversion of CO 2 to CO and oxygen in DBD [114], and they found that the gas flow rate is the most crucial parameter affecting the CO 2 conversion. At 0.05 L min −1 flow rate, 14.75 W cm −3 power density and 60 kHz discharge frequency, 30% CO 2 conversion was achieved. The performance might be further enhanced by optimizing the discharge parameters (i.e., power, frequency, dielectric material) or by implementing parallel reactors.
usually have a special feature of regulating product distribution.
NTP can be generated through various types of discharges-i.e., microwave discharges, glow discharges, gliding arc discharges, dielectric barrier discharges (DBD), etc.-but DBD are the best option to be used in plasma catalysis. Indeed, DBDs are usually operated at atmospheric pressure and ambient temperature, and the integration of DBD plasma with catalysts has the advantages of simple operation and low cost. Thus plasma-driven direct hydrogenation of CO2 is mostly based on DBD plasmas. The possible plasma/catalyst synergism is illustrated in Figure 19 [13]. Figure 19. Overview of the possible effects of the catalyst on the plasma and vice versa, possibly leading to synergism in plasma-catalysis, reproduced with permission from [13]. Copyright Royal Society of Chemistry, 2017.
A lot of research has been performed for pure CO2 splitting, in various types of plasma reactors, including DBD, microwave discharge and gliding arc discharge, without catalysts [114][115][116][117]. A typical experimental set-up of a DBD plasma for CO2 decomposition is shown in Figure 20. Paulussen et al. studied the conversion of CO2 to CO and oxygen in DBD [114], and they found that the gas flow rate is the most crucial parameter affecting the CO2 conversion. At 0.05 L min −1 flow rate, 14.75 W cm −3 power density and 60 kHz discharge frequency, 30% CO2 conversion was achieved. The performance might be further enhanced by optimizing the discharge parameters (i.e., power, frequency, dielectric material) or by implementing parallel reactors. Dry reforming of methane (DRM), i.e., the combined conversion of CH4 and CO2 by plasma and/or plasma catalysis, has attracted extensive attention in recent years. For instance, Li et al. studied CO2 reforming of CH4 by taking advantage of atmospheric pressure glow discharge plasma [118]. Liu et al. reported high-efficient conversion of CO2 and CH4 in AC-pulsed tornado gliding arc plasma [119]. Kolb et al. investigated DRM in a DBD reactor [120]. In the above-mentioned studies [118][119][120], syngas (CO and H2) was produced as the main product. Recently, Wang et al. reported a novel onestep reforming of CO2 and CH4 into liquid oxygenate products, dominated by acetic acid, at room temperature by the coupling of DBD plasma and catalysts [121]. They examined the effect of CH4/CO2 molar ratio and of various catalysts (γ-Al2O3, Cu-γ-Al2O3, Au-γ-Al2O3, Pt-γ-Al2O3). Interestingly, compared with plasma-only mode, the coupling of plasma and catalysts tuned the selectivity of liquid chemicals, and oxygenates selectivity was achieved up to approximately 60% when Cu catalyst was used. Details about the effects of operating mode and catalysts on the CO2 conversion reaction results are shown in Figure 21. Although much efforts need to be made to reveal the unknown mechanisms, the results are attractive and show an excellent application prospect. Dry reforming of methane (DRM), i.e., the combined conversion of CH 4 and CO 2 by plasma and/or plasma catalysis, has attracted extensive attention in recent years. For instance, Li et al. studied CO 2 reforming of CH 4 by taking advantage of atmospheric pressure glow discharge plasma [118]. Liu et al. reported high-efficient conversion of CO 2 and CH 4 in AC-pulsed tornado gliding arc plasma [119]. Kolb et al. investigated DRM in a DBD reactor [120]. In the above-mentioned studies [118][119][120], syngas (CO and H 2 ) was produced as the main product. Recently, Wang et al. reported a novel one-step reforming of CO 2 and CH 4 into liquid oxygenate products, dominated by acetic acid, at room temperature by the coupling of DBD plasma and catalysts [121]. They examined the effect of CH 4 /CO 2 molar ratio and of various catalysts (γ-Al 2 O 3 , Cu-γ-Al 2 O 3 , Au-γ-Al 2 O 3 , Pt-γ-Al 2 O 3 ). Interestingly, compared with plasma-only mode, the coupling of plasma and catalysts tuned the selectivity of liquid chemicals, and oxygenates selectivity was achieved up to approximately 60% when Cu catalyst was used. Details about the effects of operating mode and catalysts on the CO 2 conversion reaction results are shown in Figure 21. Although much efforts need to be made to reveal the unknown mechanisms, the results are attractive and show an excellent application prospect. Besides the above-mentioned metal catalysts, zeolites have also been used in plasma catalytic DRM. Zhang et al. studied the catalytic performance of zeolite catalysts (i.e., NaA, NaY, and HY) for the direct conversion of CH4 and CO2 at relatively low temperature range and ambient pressure via DBD plasma [122], and the products were dominated by syngas and C4 hydrocarbons. Form the investigated catalysts (NaA, NaY, and HY), HY zeolite catalyst exhibited the best performance (26.7% CO2 conversion and 52.1% C4 hydrocarbon selectivity), which was mainly attributed to the appropriate pore size and electrostatic properties of HY zeolite.
Although direct hydrogenation of CO2 to CH3OH is an exothermic reaction, studies of heterogeneous catalysis for CO2 hydrogenation to CH3OH were usually operated at high temperature and high pressure, mainly caused by the high stability of CO2 and the low equilibrium constant at atmospheric pressure. Plasma catalysis, however, is a promising approach to enable CO2 conversion to CH3OH at ambient conditions, and has gradually attracted more and more interest. For instance, Eliasson et al. reported the direct hydrogenation of CO2 to CH3OH by coupling of DBD plasma and a discharge-activated catalyst (CuO-ZnO-Al2O3) [123]. By comparison of the experiments, with catalyst only, discharge only, and discharge + catalyst, they found that DBD plasma effectively lowered the optimal reaction temperature corresponding to the best catalytic performance. Indeed, the optimal reaction temperature was 493 K for catalyst only, while in terms of discharge only and discharge + catalyst, the optimal reaction temperature for CO2 conversion was 373 K. The maximum selectivity for the methanol formation (10%) was achieved at a temperature of 373 K, which implies that the plasma improved the catalytic activity of CuO-ZnO-Al2O3 catalyst at low temperature. Besides the above-mentioned metal catalysts, zeolites have also been used in plasma catalytic DRM. Zhang et al. studied the catalytic performance of zeolite catalysts (i.e., NaA, NaY, and HY) for the direct conversion of CH 4 and CO 2 at relatively low temperature range and ambient pressure via DBD plasma [122], and the products were dominated by syngas and C 4 hydrocarbons. Form the investigated catalysts (NaA, NaY, and HY), HY zeolite catalyst exhibited the best performance (26.7% CO 2 conversion and 52.1% C 4 hydrocarbon selectivity), which was mainly attributed to the appropriate pore size and electrostatic properties of HY zeolite.
Although direct hydrogenation of CO 2 to CH 3 OH is an exothermic reaction, studies of heterogeneous catalysis for CO 2 hydrogenation to CH 3 OH were usually operated at high temperature and high pressure, mainly caused by the high stability of CO 2 and the low equilibrium constant at atmospheric pressure. Plasma catalysis, however, is a promising approach to enable CO 2 conversion to CH 3 OH at ambient conditions, and has gradually attracted more and more interest. For instance, Eliasson et al. reported the direct hydrogenation of CO 2 to CH 3 OH by coupling of DBD plasma and a discharge-activated catalyst (CuO-ZnO-Al 2 O 3 ) [123]. By comparison of the experiments, with catalyst only, discharge only, and discharge + catalyst, they found that DBD plasma effectively lowered the optimal reaction temperature corresponding to the best catalytic performance. Indeed, the optimal reaction temperature was 493 K for catalyst only, while in terms of discharge only and discharge + catalyst, the optimal reaction temperature for CO 2 conversion was 373 K. The maximum selectivity for the methanol formation (10%) was achieved at a temperature of 373 K, which implies that the plasma improved the catalytic activity of CuO-ZnO-Al 2 O 3 catalyst at low temperature. Additionally, the authors found that low input power and high pressure are beneficial for the improvement of the methanol selectivity.
Zeng et al. studied CO 2 hydrogenation by combining various catalysts (i.e., Cu/γ-Al 2 O 3 , Mn/γ-Al 2 O 3 , and Cu-Mn/γ-Al 2 O 3 ) with DBD plasma in a coaxial packed-bed [124]. The experimental results showed that the addition of catalysts in the reactor improved the conversion of CO 2 . At the same time, with the increase of the H 2 /CO 2 molar ratio, the CO 2 conversion was improved, and the CO yield was also enhanced. It is also worth mentioning that, compared with plasma only experiments, the energy efficiency was enhanced by adding catalysts, although the synergetic mechanism between catalysts and plasma is still unknown.
Recently, Wang et al. examined the influence of plasma reactor structure and catalysts on CO 2 conversion and CH 3 OH selectivity for plasma catalytic CO 2 hydrogenation [125], and a schematic diagram of the experimental setup and images of the H 2/ CO 2 discharge are shown in Figure 22. They investigated three kinds of reactors, i.e., a cylindrical reactor (aluminum foil sheet as ground electrode), a double dielectric barrier discharge reactor (water as ground electrode), and a single dielectric barrier discharge reactor (water as a ground electrode). The single DBD reactor equipped with a special water-electrode showed the optimal reaction performance (21.2% CO 2 conversion and 53.7% CH 3 OH selectivity). In addition, they tested the catalytic performance of Pt/γ-Al 2 O 3 catalysts and Cu/γ-Al 2 O 3 catalysts in the optimized reactor for direct conversion of CO 2 to CH 3 OH. As shown in Figure 23, both Pt/γ-Al 2 O 3 catalysts and Cu/γ-Al 2 O 3 catalysts improved not only the CO 2 conversion, but also the CH 3 OH selectivity, and Cu/γ-Al 2 O 3 catalysts showed a better catalytic performance (21.2% CO 2 conversion and 53.7% methanol selectivity). That is, a strong synergistic effect between the plasma and Cu/γ-Al 2 O 3 catalysts promoted the hydrogenation of CO 2 to methanol, although the reaction temperature in the optimized reactor remained near room temperature. Additionally, the authors found that low input power and high pressure are beneficial for the improvement of the methanol selectivity. Zeng et al. studied CO2 hydrogenation by combining various catalysts (i.e., Cu/γ-Al2O3, Mn/γ-Al2O3, and Cu-Mn/γ-Al2O3) with DBD plasma in a coaxial packed-bed [124]. The experimental results showed that the addition of catalysts in the reactor improved the conversion of CO2. At the same time, with the increase of the H2/CO2 molar ratio, the CO2 conversion was improved, and the CO yield was also enhanced. It is also worth mentioning that, compared with plasma only experiments, the energy efficiency was enhanced by adding catalysts, although the synergetic mechanism between catalysts and plasma is still unknown.
Recently, Wang et al. examined the influence of plasma reactor structure and catalysts on CO2 conversion and CH3OH selectivity for plasma catalytic CO2 hydrogenation [125], and a schematic diagram of the experimental setup and images of the H2/CO2 discharge are shown in Figure 22. They investigated three kinds of reactors, i.e., a cylindrical reactor (aluminum foil sheet as ground electrode), a double dielectric barrier discharge reactor (water as ground electrode), and a single dielectric barrier discharge reactor (water as a ground electrode). The single DBD reactor equipped with a special water-electrode showed the optimal reaction performance (21.2% CO2 conversion and 53.7% CH3OH selectivity). In addition, they tested the catalytic performance of Pt/γ-Al2O3 catalysts and Cu/γ-Al2O3 catalysts in the optimized reactor for direct conversion of CO2 to CH3OH. As shown in Figure 23, both Pt/γ-Al2O3 catalysts and Cu/γ-Al2O3 catalysts improved not only the CO2 conversion, but also the CH3OH selectivity, and Cu/γ-Al2O3 catalysts showed a better catalytic performance (21.2% CO2 conversion and 53.7% methanol selectivity). That is, a strong synergistic effect between the plasma and Cu/γ-Al2O3 catalysts promoted the hydrogenation of CO2 to methanol, although the reaction temperature in the optimized reactor remained near room temperature.
It is obvious that the combination of the plasma and catalysts can enhance the catalytic reaction at room temperature and atmospheric pressure. The maximum methanol selectivity of 53.7% was achieved with a CO2 conversion of 21.2% via the plasma-catalysis process (see Figure 23). All the above studies show a strong synergistic effect between plasma and catalysts. Images of H2/CO2 discharge generated in DBD reactor without catalyst, reprinted with permission from [125]. Copyright American Chemical Society, 2017. Figure 23. Effect of H2/CO2 molar ratio and catalysts on the reaction performance of the plasma hydrogenation process, reprinted with permission from [125]. Copyright American Chemical Society, 2017.
Using one-dimensional fluid modeling [126], De Bie et al. proposed a chemical reaction network in the plasma (i.e., without catalysts) for conversion of CO2 to value-added chemicals, i.e., CO, CH4, CH2O, CH3OH, and hydrocarbons. The simulation results indicated the dominant reaction pathways for the conversion of CO2 and H2, as illustrated in Scheme 5. According to the model, the combination between H atoms and CHO radicals is the most important reaction to form CO, while this reaction is counterbalanced by the reorganization of H with CO into CHO radicals. Therefore, the most effective net CO formation reaction is dissociation of CO2 influenced by electrons. The production of CH4 was generally driven by two reactions, i.e., three-body recombination reaction between CH3 and H radicals, and charge transfer reaction between CH5 + and H2O. However, the latter reaction is partly balanced by the loss of CH4, resulting from a charge transfer reaction with H3 + . The production of CH2O is closely related to the initial CO2 fraction in the gas mixture. At a low initial fraction of CO2, the reaction between CO2 and CH2 radicals seem to be the most important channel for the formation of CH2O, while at higher initial CO2 fractions, CH2O is also produced out of two CHO radicals to some extent. In addition, as predicted by the model, the most important channel for the formation of CH3OH is the three-body reaction between CH3 and OH radicals, while the three-body reaction between CH2OH and H radicals is also an effective production channel for CH3OH. Furthermore, they also found that a higher density of CH3 and CH2 radicals would be essential to tune the distribution of end products. Therefore, it can be predicted that the degree of hydrogenation in the reaction has a significant influence for the targeted products. It is obvious that the combination of the plasma and catalysts can enhance the catalytic reaction at room temperature and atmospheric pressure. The maximum methanol selectivity of 53.7% was achieved with a CO 2 conversion of 21.2% via the plasma-catalysis process (see Figure 23). All the above studies show a strong synergistic effect between plasma and catalysts.
Using one-dimensional fluid modeling [126], De Bie et al. proposed a chemical reaction network in the plasma (i.e., without catalysts) for conversion of CO 2 to value-added chemicals, i.e., CO, CH 4 , CH 2 O, CH 3 OH, and hydrocarbons. The simulation results indicated the dominant reaction pathways for the conversion of CO 2 and H 2 , as illustrated in Scheme 5. According to the model, the combination between H atoms and CHO radicals is the most important reaction to form CO, while this reaction is counterbalanced by the reorganization of H with CO into CHO radicals. Therefore, the most effective net CO formation reaction is dissociation of CO 2 influenced by electrons. The production of CH 4 was generally driven by two reactions, i.e., three-body recombination reaction between CH 3 and H radicals, and charge transfer reaction between CH 5 + and H 2 O. However, the latter reaction is partly balanced by the loss of CH 4 , resulting from a charge transfer reaction with H 3 + . The production of CH 2 O is closely related to the initial CO 2 fraction in the gas mixture. At a low initial fraction of CO 2 , the reaction between CO 2 and CH 2 radicals seem to be the most important channel for the formation of CH 2 O, while at higher initial CO 2 fractions, CH 2 O is also produced out of two CHO radicals to some extent. In addition, as predicted by the model, the most important channel for the formation of CH 3 OH is the three-body reaction between CH 3 and OH radicals, while the three-body reaction between CH 2 OH and H radicals is also an effective production channel for CH 3 OH. Furthermore, they also found that a higher density of CH 3 and CH 2 radicals would be essential to tune the distribution of end products. Therefore, it can be predicted that the degree of hydrogenation in the reaction has a significant influence for the targeted products. Figure 24 summarizes the selectivity of CO, CH 4 , or CH 3 OH, as a function of the CO 2 conversion, obtained from the recent reports discussed above. Several main trends are clear. Firstly, the main products are CO and CH 4 for direct hydrogenation of CO 2 in plasma catalysis, while the selectivity of CH 3 OH is relatively low. Moreover, it is obvious that researchers have paid more attention to the reduction of CO 2 to CO and CH 4 up to now. However, the production of other hydrocarbons, such as olefins and gasoline hydrocarbons, would also be a promising direction, to achieve the maximum utilization of CO 2 by plasma catalysis. Therefore, some insights from heterogeneous catalysis, especially the combination of metal catalysts, metal oxide catalysts, and zeolite catalysts could be helpful to develop a methodology for plasma catalytic hydrogenation of CO 2 . On the other hand, there is no guarantee that good catalysts in thermal processes would also perform well in plasma catalysis, because of the clearly different operating conditions (e.g., lower temperature, abundance of reactive species, excited species, charges, and electric field present in plasma).  Figure 24 summarizes the selectivity of CO, CH4, or CH3OH, as a function of the CO2 conversion, obtained from the recent reports discussed above. Several main trends are clear. Firstly, the main products are CO and CH4 for direct hydrogenation of CO2 in plasma catalysis, while the selectivity of CH3OH is relatively low. Moreover, it is obvious that researchers have paid more attention to the reduction of CO2 to CO and CH4 up to now. However, the production of other hydrocarbons, such as olefins and gasoline hydrocarbons, would also be a promising direction, to achieve the maximum utilization of CO2 by plasma catalysis. Therefore, some insights from heterogeneous catalysis, especially the combination of metal catalysts, metal oxide catalysts, and zeolite catalysts could be helpful to develop a methodology for plasma catalytic hydrogenation of CO2. On the other hand, there is no guarantee that good catalysts in thermal processes would also perform well in plasma catalysis, because of the clearly different operating conditions (e.g., lower temperature, abundance of reactive species, excited species, charges, and electric field present in plasma). Scheme 5. Dominant reaction pathways for the plasma-based conversion (without catalysts) of CO 2 and H 2 into various products, in a 50/50 CO 2 /H 2 gas mixture. The thickness of the arrow lines is proportional to the rates of the net reactions. The stable molecules are indicated with black rectangles, reprinted with permission from [126]. Copyright American Chemical Society, 2016. Scheme 5. Dominant reaction pathways for the plasma-based conversion (without catalysts) of CO2 and H2 into various products, in a 50/50 CO2/H2 gas mixture. The thickness of the arrow lines is proportional to the rates of the net reactions. The stable molecules are indicated with black rectangles, reprinted with permission from [126]. Copyright American Chemical Society, 2016. Figure 24 summarizes the selectivity of CO, CH4, or CH3OH, as a function of the CO2 conversion, obtained from the recent reports discussed above. Several main trends are clear. Firstly, the main products are CO and CH4 for direct hydrogenation of CO2 in plasma catalysis, while the selectivity of CH3OH is relatively low. Moreover, it is obvious that researchers have paid more attention to the reduction of CO2 to CO and CH4 up to now. However, the production of other hydrocarbons, such as olefins and gasoline hydrocarbons, would also be a promising direction, to achieve the maximum utilization of CO2 by plasma catalysis. Therefore, some insights from heterogeneous catalysis, especially the combination of metal catalysts, metal oxide catalysts, and zeolite catalysts could be helpful to develop a methodology for plasma catalytic hydrogenation of CO2. On the other hand, there is no guarantee that good catalysts in thermal processes would also perform well in plasma catalysis, because of the clearly different operating conditions (e.g., lower temperature, abundance of reactive species, excited species, charges, and electric field present in plasma). Overview of selectivity into CO, CH 4 , and CH 3 OH, as a function of CO 2 conversion, based on all reports available in literature about plasma catalysis (all in DBD reactors). The references where these data are adopted from are discussed in the text.
Up to now, in view of the complexity of this interdisciplinary field, the development of plasma catalysis still needs major research efforts. On one hand, plasma catalysis has potential for industrial application, because it drives the CO 2 conversion reaction at ambient temperature and atmospheric pressure, breaking thermodynamic equilibrium to make full use of feedstock. On the other hand, the excessive energy consumption-caused by high input power and heat loss-is an important disadvantage, but it can be mitigated with the further development of renewable energy (i.e., wind, solar, and tidal energy). Additionally, to improve the reaction activity for CO 2 conversion, it is crucial to explore the reaction mechanisms by in situ characterization and computer modeling, to improve the synergistic effect between plasma and catalysts. Finally, we need to search for suitable catalytic materials to strength the reaction performance and decrease the production costs.

Outlook and Conclusions
Currently, fuels and base chemicals are nearly all produced from non-renewable fossil energy (oil, natural gas, and coal), and CO 2 is generally the end product (e.g., upon burning fossil fuels) or a waste product in chemical industry. This indicates that we should use CO 2 as the main carbon source when fossil energy would get depleted in the future. Thus, in the long term, hydrogenation of CO 2 (as well as CO 2 conversion with other H-sources) into value-added chemicals and fuels is very significant, since it can close the carbon cycle, as shown in Figure 25. However, some crucial issues should be addressed in advance for application of CO 2 hydrogenation in industry.
on all reports available in literature about plasma catalysis (all in DBD reactors). The references where these data are adopted from are discussed in the text.
Up to now, in view of the complexity of this interdisciplinary field, the development of plasma catalysis still needs major research efforts. On one hand, plasma catalysis has potential for industrial application, because it drives the CO2 conversion reaction at ambient temperature and atmospheric pressure, breaking thermodynamic equilibrium to make full use of feedstock. On the other hand, the excessive energy consumption-caused by high input power and heat loss-is an important disadvantage, but it can be mitigated with the further development of renewable energy (i.e., wind, solar, and tidal energy). Additionally, to improve the reaction activity for CO2 conversion, it is crucial to explore the reaction mechanisms by in situ characterization and computer modeling, to improve the synergistic effect between plasma and catalysts. Finally, we need to search for suitable catalytic materials to strength the reaction performance and decrease the production costs.

Outlook and Conclusions
Currently, fuels and base chemicals are nearly all produced from non-renewable fossil energy (oil, natural gas, and coal), and CO2 is generally the end product (e.g., upon burning fossil fuels) or a waste product in chemical industry. This indicates that we should use CO2 as the main carbon source when fossil energy would get depleted in the future. Thus, in the long term, hydrogenation of CO2 (as well as CO2 conversion with other H-sources) into value-added chemicals and fuels is very significant, since it can close the carbon cycle, as shown in Figure 25. However, some crucial issues should be addressed in advance for application of CO2 hydrogenation in industry. From a relatively economical point of view, the direct hydrogenation of CO2 is not yet a viable approach. On one hand, CO2 of a certain purity generally depends on the development of capture and separation techniques. On the other hand, compared with the price of the feedstock (H2, 10,000 $/ton), the price of the main products (i.e., liquefied natural gas, 770 $/ton, and CH3OH, 340 $/ton) obtained by CO2 hydrogenation is too low to make this an economically viable process. However, once the production of H2 can be realized with fully-fledged solar power technology, the production cost for CO2 hydrogenation will be greatly reduced. Meanwhile, the CO2 conversion driven by solar energy (artificial photosynthesis) is also a promising routing.
Technologically, the direct hydrogenation of CO2 to CO is an endothermic reaction (ΔH = 41.2 kJ mol −1 ), and a higher reaction temperature is beneficial for the production of CO according to Le From a relatively economical point of view, the direct hydrogenation of CO 2 is not yet a viable approach. On one hand, CO 2 of a certain purity generally depends on the development of capture and separation techniques. On the other hand, compared with the price of the feedstock (H 2 , 10,000 $/ton), the price of the main products (i.e., liquefied natural gas, 770 $/ton, and CH 3 OH, 340 $/ton) obtained by CO 2 hydrogenation is too low to make this an economically viable process. However, once the production of H 2 can be realized with fully-fledged solar power technology, the production cost for CO 2 hydrogenation will be greatly reduced. Meanwhile, the CO 2 conversion driven by solar energy (artificial photosynthesis) is also a promising routing.
Technologically, the direct hydrogenation of CO 2 to CO is an endothermic reaction (∆H = 41.2 kJ mol −1 ), and a higher reaction temperature is beneficial for the production of CO according to Le Chatelier's principle. Based on the available experimental results (Table 1), the CO selectivity can reach nearly 100% under conditions of 873 K and 1 MPa. On the other hand, the production of other products (i.e., CH 4 , CH 3 OH, DME, and light olefins) from CO 2 is exothermic, and in theory a low temperature favors the equilibrium conversion. However, the inert CO 2 molecule generally needs relatively high reaction temperature to be activated. The competition between these two factors makes the reaction performance not very satisfactory. Up to now, from the perspective of technology readiness level, methanol synthesis is a far more advanced production process, with a high selectivity up to 80-90% with Cu-based catalyst (Table 3). However, there are some limitations in heterogeneous catalysis for direct hydrogenation of CO 2 , such as catalyst sintering at high temperature, the influence of water produced in the reaction process and low CO 2 conversion caused by thermo-dynamical restriction in terms of the formation of CH 3 OH and hydrocarbons. Hence, we should explore novel catalytic materials and reactors to break the thermodynamic equilibrium, as well as design bi-functional and/or multi-functional catalysts (metal, metal oxides, and zeolites) to couple the chemical reactions (RWGS, methanation, methanol synthesis, and/or MTO).
Plasma catalysis, a new field of catalysis, has attracted sufficient attention in recent years, due to its simple operating conditions (ambient temperature and atmospheric pressure) and unique advantages in activating inert molecules. However, the energy efficiency is still too high for commercial exploitation. On the other hand, renewable energy (e.g., wind, solar, and tidal energy) will further develop in the near future, and it will be a perfect match with plasma catalysis (because plasma is generated by electricity and can simply be switched on/off-allowing storage of fluctuating energy), hence making the problem of limited energy efficiency less dramatic. Nevertheless, significant efforts must be devoted to elucidate the reaction mechanisms in plasma catalysis, to improve not only the energy efficiency, but certainly also the selectivity towards value-added products. Indeed, plasma catalysis is complicated from chemistry and physics point of view, but the potential of plasma technology for industrial applications deserves major research efforts. A combination of computer simulations with experiments will be needed for an in-depth understanding of the reaction mechanisms, responsible for the synergy between plasma and catalysts. Although the development process in plasma catalysis may be slow due to its complex character, it would bring great benefits to human society, once it is developed mature enough, and once appropriate catalysts for plasma catalysis can be designed. Therefore, future research should definitely emphasize on a better understanding and rational screening of highly active catalysts.
Compared with the numerous studies on CO 2 hydrogenation by heterogeneous catalysis, much more research should be carried out in the field of plasma catalysis, to improve the CO 2 conversion and target products selectivity (and even to exploit new products, such as olefins, gasoline hydrocarbons, and aromatics), since the results achieved by plasma catalysis (Figure 24) are far from those of heterogeneous catalysis (Figure 18). To achieve this goal, the advantage of plasma should be exploited in full, and at the same time, insights from heterogeneous catalysis (e.g., catalyst combination, reaction combination, active sites design, etc.) can help to further improve the potential of this promising field of catalysis.