Nano-Intermetallic InNi3C0.5 Compound Discovered as a Superior Catalyst for CO2 Reutilization.

CO2 circular economy is urgently calling for the effective large-scale CO2 reutilization technologies. The reverse water-gas shift (RWGS) reaction is the most techno-economically viable candidate for dealing with massive-volume CO2 via downstream mature Fischer-Tropsch and methanol syntheses, but the desired groundbreaking catalyst represents a grand challenge. Here, we report the discovery of a nano-intermetallic InNi3C0.5 catalyst, for example, being particularly active, selective, and stable for the RWGS reaction. The InNi3C0.5(111) surface is dominantly exposed and gifted with dual active sites (3Ni-In and 3Ni-C), which in synergy efficiently dissociate CO2 into CO* (on 3Ni-C) and O* (on 3Ni-In). O* can facilely react with 3Ni-C-offered H* to form H2O. Interestingly, CO* is mainly desorbed at and above 400°C, whereas alternatively hydrogenated to CH3OH highly selectively below 300°C. Moreover, this nano-intermetallic can also fully hydrogenate CO-derived dimethyl oxalate to ethylene glycol (commodity chemical) with high selectivity (above 96%) and favorable stability.

Introduction

Concerns about the vital global warming and ocean acidification problems caused by CO2 excessive emission (Karl and Trenberth, 2003, Orr et al., 2005) have triggered extensive researches on its large-scale reutilization via effective, economical, and sustainable technologies for a CO2 circular economy (Aresta et al., 2014, Porosoff et al., 2016). However, industrialized CO2 reutilization is just limited to the synthesis of urea and polycarbonate (occupying only 0.5% [Shima et al., 2012, Su et al., 2017] of CO2 emissions), whereas enzymatic and electro-/photo-chemical strategies are hampered by their low CO2-conversion efficiency (Wang et al., 2008, Kondratenko et al., 2013). To achieve the large-scale CO2 reutilization, CO2 hydrogenation with renewable-energy-generated H2 to CO by the reverse water-gas shift (RWGS) reaction is the most techno-economically viable candidate (Porosoff et al., 2016, Kondratenko et al., 2013, Xu and Moulijn, 1996, Porosoff and Chen, 2013, Zhang et al., 2017), thanks not only to its high efficiency, enabling to deal with vast amounts of CO2, but also to the great versatility of syngas (CO + H2, product gas of RWGS reaction) to produce commodity chemicals and fuels (occupying 40% CO2 emissions [Zhang et al., 2017] via mature Fischer-Tropsch and methanol (CH3OH) syntheses [Porosoff et al., 2016, Kondrat et al., 2016]).

The RWGS reaction is an equilibrium-limited endothermic reaction (required enthalpy of 41.17 kJ mol−1). According to Le Châtelier's principle, high-temperature (about 400–800°C) thermodynamically favors high CO2 conversion and high CO selectivity, but the undesired methanation also proceeds under the preferred RWGS conditions (Chen et al., 2001, Wu et al., 2015, Gonçalves et al., 2017, Yang et al., 2017). Therefore, a techno-economically available catalyst with outstanding CO2-to-syngas performance is the prerequisite for the large-scale RWGS implementation. To date, homogeneous complexes and heterogeneous solids catalysts have been extensively explored. The homogeneous catalysts show satisfactory activity and selectivity (Federsel et al., 2010), but their difficult recovery from the reaction mixture makes them unattractive. The heterogeneous catalysts are more competitive in terms of ready catalyst-product separation and continuous processes. They mainly include the nanoparticles of precious metals (e.g., Au, Ag, Pt) (Porosoff et al., 2016, Yang et al., 2017) and non-precious metals (e.g., Cu, Ni) (Zhang et al., 2017, Chen et al., 2001, Wu et al., 2015, Gonçalves et al., 2017) dispersed on supports (e.g., SiO2, Al2O3, CeO2, MoCx) (Porosoff et al., 2016, Zhang et al., 2017, Chen et al., 2001, Wu et al., 2015, Gonçalves et al., 2017, Yang et al., 2017). Despite the excellent RWGS activity, the precious-metal catalysts suffer from their limited natural abundance. Cu and Ni catalysts are intensively studied but are not promising owing to either serious sintering (Cu) (Zhang et al., 2017, Chen et al., 2001) or high methanation activity (Ni) (Wu et al., 2015, Gonçalves et al., 2017). Given the chemical inertness of CO2 molecule (Xu and Moulijn, 1996), the heart of RWGS is to exquisitely design and tailor a groundbreaking catalytic material with both high efficiency and low cost, but this represents a grand challenge within the CO2-conversion field.

Against all odds, the tantalizing progresses in nano-intermetallic catalysis (Stamenkovic et al., 2007, Studt et al., 2014) open an opportunity for designing and tailoring qualified RWGS catalysts because nano-intermetallic has fascinating prospects in catalysis field, with their tunable components and ratios, variable constructions, and reconfigurable electronic structures, distinctly different from their single metals (Stamenkovic et al., 2007, Armbrüster et al., 2012, Ji et al., 2010). Particularly, their precise atomic ordering structure can provide rational predictions of the effects of geometry and electronic structure on their catalytic properties for required reactions (Wang et al., 2013, Nicholson et al., 2014, Qin et al., 2018). One of the recent pertinent examples is the discovery of a Ni5Ga3 nano-intermetallic, which strikingly shows that the Ni, originally active for CO2 methanation, turns itself suddenly into a qualified CO2-to-CH3OH catalyst after Ga alloying (Studt et al., 2014), because this intermetallic offers the unique Ga-rich sites for CH3OH formation. Encouraged by these big achievements toward nano-intermetallic catalysis, we believe that the nano-intermetallic can pave a road to the rational engineering of more intelligent catalysts gifted with flexibly arranged atomic structures and tailor-made catalytic properties for the RWGS reaction as well as other reactions for CO2 reutilization.

Here, we present a nano-intermetallic InNi3C0.5 catalyst that is particularly active, selective, and stable for the RWGS reaction under extremely wide reaction conditions. Such nano-intermetallic is fabricated via carburizing the In-Ni nano-intermetallic in the real RWGS stream and is gifted with dual active sites (i.e., 3Ni-In and 3Ni-C) on the InNi3C0.5(111) surface. The dual sites act in synergy to facilely dissociate CO2* (adsorbed on 3Ni-In sites) into CO* (on 3Ni-C sites) and O* (on 3Ni-In sites), and the O* can favorably react with 3Ni-C offered H* to form H2O. Most notably, the CO* is mainly desorbed into gas phase at and above 400°C but can be highly selectively hydrogenated to form CH3OH below 300°C with a promising CO2-to-CH3OH capacity. Furthermore, this nano-intermetallic can fully hydrogenate dimethyl oxalate (obtainable from oxidative coupling of CO (Fenton and Steinwand, 1974), product of the RWGS) to ethylene glycol (a commodity chemical) with high selectivity (above 96%) and favorable stability.

Results

Discovery of InNi3C0.5 and Its Application for RWGS Reaction

To exquisitely tailor a groundbreaking RWGS catalyst, the elaborate choice of appropriate elements oriented by this reaction should be initially conducted but poses a great challenge because the relevant elements for this reaction traverse most of the periodic table. The first metal that mostly attracts attention is Ni, because Ni-based catalysts are typically used for the RWGS reaction despite CH4 formation (Wu et al., 2015, Gonçalves et al., 2017). Moreover, In is another attractive element, because In-based catalysts are burgeoning in CO2 conversion (Ye et al., 2012, Park et al., 2017, Larrazábal et al., 2016), and, for example, the intermetallic AgIn catalyst is highly efficient for electrochemical reduction of CO2 to CO (Park et al., 2017, Larrazábal et al., 2016). We thus surmise that In-Ni intermetallic could reconstruct geometric-electronic structures of Ni, which might be feasible to switch Ni catalysis in CO2 reduction from CH4 formation to CO formation.

A series of pure intermetallics of InNi, InNi2, and InNi3 were successfully synthesized (Figure 1A) and then were evaluated for the RWGS reaction. Comparison with the conventional Cu-based catalysts (Zhang et al., 2017, Chen et al., 2001) reveals that the intermetallic In-Ni catalysts deliver exciting intrinsic RWGS performances, especially for InNi3 with a high CO formation rate of 1.96 mmol gcat−1 min−1 and a considerably low CH4 selectivity (Figure S1). It is very intriguing to find that after reaction the InNi, InNi2, and InNi3 phases are in situ changed in association with a new phase formation of InNi3C0.5 (Figure 1B, identified in following section). Consistently, the InNi3C0.5 formation is thermodynamically favorable with large ordering energy (such as 2.72 eV for InNi3 carburization with CO, Figure S2), which also portends that the InNi3C0.5 is stable under the RWGS conditions. Notably, only InNi3 could be fully transformed into pure InNi3C0.5 owing to the identical stoichiometric In:Ni ratios of 1:3 and offers the highest RWGS performance, indicating that InNi3C0.5 should be responsible for the RWGS reaction.

Figure 1

XRD Patterns of the Various Catalysts

(A) In2O3-NiO mixture (green) and In-Ni intermetallics with different In:Ni molar ratio (InNi, black; InNi2, blue; InNi3, red).

(B) Used intermetallics of InNi (black), InNi2 (blue), and InNi3 (red) after the RWGS reaction (500°C, GHSV of 30,000 mL gcat−1 h−1, H2/CO2/N2 molar ratio of 66/22/12, 0.1 MPa).

The above-mentioned results and analyses make us confident that the InNi3C0.5 intermetallic is a superior RWGS catalyst. To make it a practical catalyst, the thin-felt Al2O3/Al-fiber substrate consisting of 10 vol% 60-μm Al2O3/Al-fiber and 90 vol% voidage (Wang et al., 2016) was used to support 9 wt% InNi3C0.5. This strategy permits the engineering of InNi3C0.5 nano-intermetallic at “nano-meso-macro” triple-scale levels of both porosity and structure in one step (Figures 2A–2C, S3A, and S3B), thereby making the catalyst development and reaction engineering (for enhanced heat/mass transfer) go hand in hand (Wang et al., 2016, Li et al., 2015). The InNi3C0.5/Al2O3/Al-fiber catalyst was tested for the RWGS reaction in a tubular fixed-bed reactor. As expected, this catalyst always achieves high CO2 conversions very close to the thermodynamic equilibrium values with above 97% CO selectivity under the wide reaction conditions (Figures 2D–2F). For example, a 53% CO2 conversion is obtainable, quite close to the equilibrium value of 54%, at 540°C and a gas hourly space velocity (GHSV) of 54,000 mL gcat−1 h−1. This catalyst delivers a very high intrinsic activity with a turnover frequency (TOF) of 11.0 CO per active site per second at 540°C (see detailed TOF calculation in Supplemental Information), almost one to two orders of magnitude higher than that seen with most platinum/oxide and non-noble-metal catalysts (Table S1). Furthermore, a kinetic study was carried out over the InNi3C0.5/Al2O3/Al-fiber catalyst, and the apparent activation energy was calculated with the result as shown in Figure S3D. InNi3C0.5/Al2O3/Al-fiber provided a much lower Ea (60 kJ/mol) than Cu/ZnO-based catalysts (112 kJ/mol, Schumann et al., 2015), further indicating that this catalyst has a high intrinsic activity. Also encouraging is the exclusive CO selectivity (above 98%) with pressure increasing from 1.0 to 4.0 MPa at 540°C (Figure 2F), despite the fact that CH4 formation is much favorable at high pressure over the conventional Ni-based catalysts (Wu et al., 2015, Gonçalves et al., 2017, Li et al., 2015).

Figure 2

Structural and Morphological Features of the InNi3C0.5/Al2O3/Al-fiber Catalyst and Its RWGS Performance

(A) Optical photograph (top) and scanning electron microscopy (SEM) image (bottom) of the fresh catalyst.

(B) High-magnitude SEM image of the fresh catalyst.

(C) TEM images of the fresh catalyst (inset: lattice fringes with distance of 0.218 nm corresponding to the InNi3C0.5(111) surface).

(D–F) CO2 conversion, product selectivity, and CO-formation rate as a function of (D) reaction temperature (at a GHSV of 21,600 mL gcat−1 h−1 and 4.0 MPa), (E) GHSV (at 540°C and 4.0 MPa), and (F) reaction pressure (at a GHSV of 54,000 mL gcat−1 h−1 and 540°C) for a feed gas of H2/CO2/N2 with molar ratio of 66/22/12.

(G and H) Time on stream under different reaction conditions and comparison (H) with the reported literature data over the commercial Cu/ZnO/Al2O3 (Zhang et al., 2017) and Cu/β-Mo2C (Zhang et al., 2017) catalysts.

Stability is a significant consideration for catalysts in practical applications. Our InNi3C0.5/Al2O3/Al-fiber catalyst is very stable with 52%–53% CO2 conversion and 97%–99% CO selectivity throughout the entire 150 h testing at a GHSV of 54,000 mL gcat−1 h−1 and 540°C (Figure 2G). Even at a high GHSV of 300,000 mL gcat−1 h−1 and 600°C, the InNi3C0.5/Al2O3/Al-fiber catalyst also shows a high stability with no deactivation sign throughout 65 h testing (Figure 2H). In comparison, the Cu/β-Mo2C catalyst maintains 85% of its initial activity after 40 h reaction and the Cu/ZnO/Al2O3 catalyst loses more than 60% of its initial activity within 15 h reaction under the identical reaction conditions (Zhang et al., 2017). It is not surprising that the InNi3C0.5 crystalline phase, surface morphology, and structure of the used catalysts are preserved unchanged (Figures S3E–S3H), consistent with the excellent activity/selectivity maintenance in Figures 2G and 2H. To the best of our knowledge, the InNi3C0.5 intermetallic has never been used before for any application in catalysis, and herein we discover its superior RWGS performance—including CO2 conversion, CO selectivity, and especially high-temperature stability—over the reported state-of-the-art catalysts (Table S1).

Structure Identification

To definitely identify the crystal structure and composition of the as-formed carbide-intermetallic from In-Ni intermetallics, such pure carbide-intermetallic was synthesized via fully carburizing InNi3, and its X-ray diffraction (XRD) pattern completely coincides with the one of InNi3C0.5 that has an anti-perovskite-type structure (Joint Committee on Powder Diffraction File No. 28-0468; Figure 3A and Table S2). Moreover, the In:Ni:C molar ratio of the as-synthesized InNi3C0.5 was determined to be 1:2.99:0.49 (see elemental analyses in Supplemental Information), quite close to its stoichiometric ratio. Figure 3B shows its structural model containing eight InNi3 units. For each unit, eight In atoms occupy the eight corners and six Ni atoms occupy the six face centers; four C atoms randomly disperse in these eight body centers, but with the most stable configuration in a regular tetrahedron (Figure S4). The Wulff equilibrium shape of the InNi3C0.5 nanocrystal was further optimized, and its optimum shape exposes fourteen surfaces consisting of eight hexagons and six squares (Figure 3C). The InNi3C0.5(111) is the most stable surface of the hexagonal shapes with the lowest surface free energy (Table S3). Interestingly, high-resolution transmission electron microscopy (TEM) also displays an approximate hexagonal morphology of the real synthetic InNi3C0.5 nanoparticles (Figures 3D, 3E, and S5), and the lattice spacing of 0.218 nm is assignable to the InNi3C0.5(111) surface.

Figure 3

Structure and Morphology of the InNi3C0.5 Nano-Intermetallic

(A) XRD pattern of the as-synthesized InNi3C0.5 nano-intermetallic.

(B) Ball-and-stick perspective of polyhedral InNi3C0.5 (a 2 × 2 × 2 super cell) with cubic (Fmm) anti-perovskite-type structure (green octahedron, C-Ni6).

(C) Optimum Wulff equilibrium shape of InNi3C0.5 crystal.

(D) TEM image of the as-synthesized InNi3C0.5 nano-intermetallic, showing its uniform hexagonal shape (marked by yellow dashed lines).

(E) High-resolution TEM image of a typical InNi3C0.5 nanoparticle with hexagonal shape and a lattice spacing of 0.218 nm corresponding to the InNi3C0.5(111) surface.

Theoretical Calculations Study

In the last decade, significant advances have been achieved in the atomistic-theoretical calculations, enabling us to computationally construct molecular and crystalline structures and to reveal the reaction pathways on the catalyst surface at atomic-molecular level (Nicholson et al., 2014, Qin et al., 2018, Studt et al., 2014, Mao et al., 2017). Therefore, the RWGS reaction mechanism on InNi3C0.5 is first investigated by the density functional theory (DFT) calculations. We selected the most stable InNi3C0.5(111) as the ideal surface and established the dual active sites (h1: Hollow(3Ni-In); h2: Hollow(3Ni-C); Figure 4A) from nine kinds of possible active sites (see detailed results in Table S4). As shown in Figure 4B, the CO2 molecule is chemically adsorbed via a bending configuration to form CO2* on h1 site, and the H2 molecule spontaneously dissociates into H* that can be adsorbed on both h1 and h2 sites. Electron density distribution for the dual active sites is richer than the others, which makes them more nucleophilic and more favorable for CO2 activation (Figure S6). Therefore, the CO2* facilely dissociates into CO* adsorbed on h2 site and O* adsorbed on h1 site with moderate exothermicity (namely, reaction energy Er, −0.38 eV) and a low activation barrier (Ea, 0.32 eV), but with higher Ea of CO2* hydrogenation to formate (HCOO*, 0.42 eV) and to carboxyl (COOH*, 0.75 eV, Figure S7 and Table S5). Clearly, the CO2* dissociation to CO* and O* (i.e., redox pathway) is preferred over the formate and carboxyl pathways on the InNi3C0.5(111) surface. Furthermore, the formed O* on h1 site preferably reacts with H* on the neighboring h2 site to produce an OH* group (Ea, 0.73 eV), and subsequently, two OH* groups on the dual sites are easily transformed into H2O* (Ea, 0.25 eV) that is finally desorbed into the gas phase (Ea, 0.35 eV). The dual active sites provide much lower Ea than the sole h1 sites for the above-mentioned steps (see detailed results in Table S5), probably the consequences of appropriate adsorption of reaction intermediates in terms of their adsorption strength (Table S4) and the distance between them (the dual active sites have shorter adjacent h1-h2 distance of 3.106 Å than the sole h1 sites with an adjacent h1-h1 distance of 5.345 Å, Figure 4A). In contrast, CO2* dissociation on Cu(111) becomes endothermic (Er, +1.06 eV, thermodynamically unfavorable) and is kinetically unfavorable (Ea of 1.55 eV versus 0.32 eV on InNi3C0.5(111), Figures 4B, S8, and S9).

Figure 4

Dual Active Sites and Reaction Pathways for CO2 Hydrogenation on the InNi3C0.5(111) Surface

(A) Side (left) and top (middle) views of the InNi3C0.5(111) surface, and detailed structure (right) of the dual active sites of “3Ni-In” (i.e., Hollow I by three Ni atoms and one In atom) and “3Ni-C” (i.e., Hollow II by three Ni atoms and one C atom).

(B and C) Energy profiles on the InNi3C0.5(111) surface for (B) the most favorable pathways to RWGS reaction and for (C) the competitive pathways for CO2 hydrogenation to CO (green), CH3OH (blue), and CH4 (black), where the black, red, and white balls represent the C, O, and H atoms in the reactive species, respectively.

The formed CO* either undergoes further hydrogenation to CH4 and/or CH3OH or desorbs into the gas phase. Figure 4C shows that CO* desorption overcomes a slightly higher Ea of 1.36 eV at 0 K than the formation of CH4 (CH3*-to-CH4*, 1.27 eV) and CH3OH (CO*-to-HCO*, 1.05 eV), clearly exhibiting a possibility of CH3OH formation (see detailed results and discussion in Figures S10 and S11). It should be noted, however, that CO* desorption is thermodynamically more favorable at elevated temperatures (Figures S12 and S13) owing to the significant entropy contributions (Graciani et al., 2014), and therefore CO* is preferentially desorbed into gas phase rather than hydrogenated into CH3OH at our real RWGS temperature of 420°C –600°C (see experimental results in Figures 2D–2F).

Infrared Spectroscopy Study

To verify the RWGS reaction pathway on InNi3C0.5 from experimental perspective, the in situ Fourier transform infrared (FTIR) spectroscopy analysis was carried out on pure InNi3C0.5 in a continuous H2/CO2/N2 (molar ratio of 66/22/12) flow at ambient pressure. As shown in Figure 5A, the linear adsorbed CO* species are formed from CO2 dissociation even at 50°C, evidenced by infrared (IR) bands (Martin et al., 2016) at 2132, 2107, 2094, 2077, and 2055 cm−1. Along with the increase in the temperature, the IR band intensity of linear adsorbed CO* becomes slightly stronger from 50°C to 175°C, remains almost unchanged from 200°C to 250°C, and then diminishes until disappearance at 325°C. In addition, two new bands at 1942 and 1824 cm−1 assignable to the bridge-absorbed CO* species (Dou et al., 2017) are observed at 100°C while becoming stronger and stronger along with the temperature. Plentiful gaseous CO starts to be detected only at 300°C, and its formation is favored with the temperature. Neither CH4 (at 3013 cm−1) (Dou et al., 2017) nor formate and carboxyl species (at 1281 and 1360–1600 cm−1) (Dou et al., 2017) are detectable in the whole temperature range studied, coinciding with the DFT-suggested preferable formation of CO over CH4, formate, and carboxyl. It should be also noticed that no adsorbed CO2* species are detectable; a possible explanation is that the CO2 adsorption-dissociation is too fast to be monitored by IR, also coinciding with the DFT-indicated very low Ea of only 0.32 eV for CO2* dissociation. These IR spectra undoubtedly validate the DFT results: CO2 can be efficiently converted to CO via redox pathway rather than formate and carboxyl ones.

Figure 5

In Situ FTIR and On-line Mass Spectrometric Analyses of CO2 Hydrogenation on the InNi3C0.5 Catalyst

(A and B) In situ FTIR spectra of CO2 hydrogenation against reaction temperature on the InNi3C0.5 catalyst in wavenumber range of (A) 2,210–1,730 cm−1 and (B) 1,230–980 cm−1.

(C) Mass spectrometric signals of the carbonaceous species for CO2 hydrogenation: CO2 signal (m/z = 44), CH3OH signal (m/z = 31), CO signal (m/z = 28), and CH4 signal (m/z = 16).

Moreover, DFT calculations on InNi3C0.5(111) surface predict the possibility of CH3OH formation (Figures 4C and S10 and Table S5). CO* is first hydrogenated into HCO* (Ea, 1.05 eV), which is easily hydrogenated into CH2O* (Ea, 0.32 eV); CH2O* can be continuously hydrogenated into CH2OH* (Ea, 0.65 eV) or CH3O* (Ea, 0.60 eV); however, CH2OH* is more favorably hydrogenated into CH3OH* (Ea, 0.88 eV) over CH3O* to CH3OH* (Ea, 1.71 eV). Therefore, we infer that CH3O* should be detectable by IR owing to its high accumulation and that CH3OH can be formed through the CO*-to-HCO*-to-CH2O*-to-CH2OH*-to-CH3OH* pathway (see detailed results and discussion in Figure S10). Indeed, CH3O* with IR band at 1033 cm−1 are detectable at 200°C –325°C (Figure 5B), whereas CH3OH is detected by the on-line mass spectrometry (MS) at 220°C–310°C accompanied by gaseous CO formation above 300°C (Figure 5C). Notably, the absence of CH2O* and CH2OH* in in situ IR spectra is probably a consequence of the low residence time of these species on the surface under atmospheric conditions (Graciani et al., 2014). These IR and MS spectra consistently display that CO* is hydrogenated into CH3OH highly selectively below 300°C, whereas it is dominantly desorbed into gas phase above 300°C.

Extended Application for CO2-to-CH3OH and Carbonyl-to-Hydroxyl Transformations

The above-mentioned DFT and FTIR results also make us confident that the InNi3C0.5 nano-intermetallic is a potential catalyst for the CO2 hydrogenation to CH3OH, which becomes more and more competing in recent years. With reaction temperature reduced from 400°C–600°C (for the RWGS reaction) to 300°C and below, the InNi3C0.5/Al2O3/Al-fiber indeed turns itself suddenly into a CO2-to-CH3OH catalyst, being capable of converting 1%–8% CO2 into CH3OH with 60%–98% selectivity (corresponding to the CH3OH space time yield of 70–330 gMeOH kgcat−1 h−1) at 200°C –300°C (Table S6). The preferable CH3OH formation rather than CO formation below 300°C is attributed to the fact that low temperatures thermodynamically favor further hydrogenation of CO* to CH3OH* (Figures S12 and S13). These results exhibit an interesting temperature-dependent selectivity switching for CO2 hydrogenation.

Moreover, in the light that CO2 molecule has the carbonyl property and InNi3C0.5 intermetallic can efficiently activate CO2 molecule, we wonder whether this catalyst is favorable for other carbonyl-compounds transformation, such as the hydrogenation of aldehydes/ketones/esters to corresponding alcohols. To avoid the adverse influence of acid groups on the surface of Al2O3, we directly supported the InNi3C0.5 nano-intermetallic onto a thin-sheet Ni-foam substrate with 110 pores per inch (Figure S14, see detailed preparation in Supplemental Information). Indeed, the InNi3C0.5/Ni-foam catalyst presents the satisfying activity and high product selectivity (Tables 1 and S7), providing the general and efficient ability to activate the C=O bond for carbonyl-to-hydroxyl transformation. Notably, ethylene glycol (EG) is an important commodity chemical, used for polyester manufacture, anti-freeze compounds, and solvents (Yue et al., 2012), and the gas-phase hydrogenation of dimethyl oxalate (DMO) to EG (its commercialization is on the way) is an attractive alternative EG synthesis using syngas (Fenton and Steinwand, 1974) derived from non-oil resources (such as coal, natural gas, and biomass) even from CO2 through the RWGS reaction. This foam-structured catalyst is capable of completely converting DMO at a high EG selectivity of 96% with a promising stability (Table 1). Moreover, the InNi3C0.5/Ni-foam also shows favorable RWGS and CO2-to-CH3OH performances that are comparable with those seen with the InNi3C0.5/Al2O3/Al-fiber (Tables S8 and S9).

Table 1

Hydrogenation of Carbonyl Compounds Catalyzed by the InNi3C0.5/Ni-foam Catalyst

Substrate	Target Product	T (°C)	P (MPa)	WHSVa (h−1)	Conv. (%)	Sel. (%)
DMO	EG	210	2.5	0.44	100/100b	96.0/96.1b
Acetone	Isopropanol	150	0.1	6.0	65.3	>99
Furfural	Furfuryl alcohol	180	0.5	0.7	98.3	91.0
Cyclohexanone	Cyclohexanol	170	0.1	6.0	62.6	>99
Butanone	Butanol	120	0.1	5.0	57.3	98.1
Salicylaldehyde	Salicylol	350	0.1	8.0	67.4	97.2
n-Nonaldehyde	n-Nonyl alcohol	300	0.1	7.0	56.0	95.1

The molar ratio of H2 to DMO is 135, and the ratios of H2 to other substrates are 10. The by-products selectivities were summarized in Table S7.

Weight hourly space velocity (WHSV) of substrates.

DMO conversion and EG selectivity after 500 h reaction.

Discussion

In summary, we have discovered an outstanding nano-intermetallic InNi3C0.5 catalyst system via RWGS-reaction-oriented pre-design combined with atomistic-theoretical calculations and experimental verifications. Practical fiber/foam-structured InNi3C0.5 nano-intermetallic catalysts engineered from nano- to macro-scale in one step have been developed, achieving unprecedented performance in the RWGS reaction and showing potential to catalyze CO2 hydrogenation to CH3OH. Most notably, such nano-intermetallic catalysts are also highly active, highly selective, and highly particularly stable for the DMO-to-EG process (EG synthesis using syngas derived from non-oil resources even from CO2 through the RWGS reaction). We anticipate our essay to be a new point closer toward the ultimate goal of catalysis, namely, designing and tailoring the catalysts atom by atom with precise structure, and our findings might lead to commercial exploitation of such kind of nano-intermetallic catalysts for applications in highly efficient reduction of CO2 to CO as well as carbonyl-to-hydroxyl transformation.

Limitations of the Study

The large-scale H2 production should be from the renewable solar, hydraulic, and wind energy.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.