Synthesis of liquid fuel via direct hydrogenation of CO₂

Significance

CO₂ is a greenhouse gas. Synthesis of liquid fuel using CO₂ and H₂ is promising for the sustainability of mankind. The reported technologies usually proceed via CO intermediate, which needs high temperature, and tend to cause low selectivity. Direct hydrogenation of CO₂ to liquid fuel, not via CO, is a challenging issue. In this work, we designed a Co₆/MnO_x nanocatalyst that could successfully avoid the CO route. The reaction could proceed at 200 °C, which is much lower than those reported so far. The selectivity of the liquid fuel in total products reached 53.2 C-mol%, which is among the highest reported to date.

Abstract

Synthesis of liquid fuels (C₅₊ hydrocarbons) via CO₂ hydrogenation is very promising. Hydrogenation of CO₂ to liquid hydrocarbons usually proceeds through tandem catalysis of reverse water gas shift (RWGS) reaction to produce CO, and subsequent CO hydrogenation to hydrocarbons via Fischer–Tropsch synthesis (FTS). CO₂ is a thermodynamically stable and chemically inert molecule, and RWGS reaction is endothermic and needs a higher temperature, whereas FTS reaction is exothermic and is thermodynamically favored at a lower temperature. Therefore, the reported technologies have some obvious drawbacks, such as high temperature, low selectivity, and use of complex catalysts. Herein we discovered that a simple Co₆/MnO_x nanocatalyst could efficiently catalyze CO₂ hydrogenation. The reaction proceeded at 200 °C, which is much lower than those reported so far. The selectivity of liquid hydrocarbon (C₅ to C₂₆, mostly n-paraffin) in total product could reach 53.2 C-mol%, which is among the highest reported to date. Interestingly, CO was hardly detectable during the reaction. The in situ Fourier transform infrared characterization and ¹³CO labeling test confirmed that the reaction was not via CO, accounting for the eminent catalytic results. This report represents significant progress in CO₂ chemistry and CO₂ transformation.

Transformation of CO₂ into value-added chemicals has attracted much attention in past decades (1–5). Liquid hydrocarbons (more than five carbon atoms, i.e., C₅₊) are widely used as liquid fuels, which are currently produced from fossil resources, especially petroleum. The growing demand for liquid fuels and gradual depletion of petroleum lead to the urgent need for their production from renewable resources. Hydrogen gas is a clean reductant, which can be produced from water by artificial photosynthesis or by electricity from renewable energy such as wind and solar energy, and this area is developing rapidly (6, 7). Moreover, CO₂ hydrogenation into liquid hydrocarbons is an interesting topic in chemistry (8–15). However, at the present stage, it is still a great challenge. Firstly, CO₂ is a fully oxidized, thermodynamically stable and chemically inert molecule. Secondly, the difficulty of CO₂ adsorption on a catalyst surface may result in a low C/H ratio at catalytic sites, favoring methane formation and inhibiting the chain growth probability (8, 9). To date, most reports of CO₂ hydrogenation have been focusing on C₁ or short-chain products, such as CO, HCOOH, CH₃OH, CH₄, and C₂ to C₄ olefins, while studies on producing liquid (C₅₊) hydrocarbons are rarely reported (8–11).

Hydrogenation of CO₂ to liquid hydrocarbons usually proceeds through tandem catalysis of reverse water gas shift (RWGS) reaction and subsequent CO hydrogenation to hydrocarbons via Fischer–Tropsch synthesis (FTS) (8, 9). The RWGS reaction is endothermic and needs a higher temperature, whereas the FTS reaction is exothermic and favors a lower temperature. Hence this reaction path has inherent drawbacks. Firstly, complicated catalysts are usually required to tailor the cascade reactions, and a high temperature (usually above 300 °C) is often needed to drive these reactions. Secondly, the equilibrium of the RWGS reaction limits the pressure of CO generated in situ, which restricts the selectivity of liquid hydrocarbon in subsequent FTS reaction. Hence, an additional reforming catalyst is usually required to obtain good catalytic results. Thirdly, considerable CO is inevitably retained in the final product, which usually occupies about 20 to 45% of the total product. Although, in several cases, methanol acted as an intermediate to produce liquid hydrocarbon from CO₂ hydrogenation at high temperature, considerable CO byproduct also existed in the final product. Iron-based catalysts, which could catalyze both RWGS and FTS reactions, were predominantly used as a catalyst for CO₂ hydrogenation to longer-chain hydrocarbons, especially olefins (16, 17). Very recently, significant progress has been made in hydrogenating CO₂ into gasoline fuel. It was found that, in a flow reactor, Na−Fe₃O₄/HZSM-5 multifunctional catalyst (14) and In₂O₃/HZSM-5 bifunctional catalyst (15) showed excellent performance in converting CO₂ and H₂ into liquid fuel at about 320 °C. In the reaction, CO was an intermediate and/or considerable byproduct. Obviously, a novel catalyst for liquid fuel synthesis via CO₂ hydrogenation that can avoid CO is highly desirable.

In a previous paper, we developed a route to synthesize acetic acid via methanol hydrocarboxylation over a homogeneous catalyst, where CO was almost undetectable and CO₂ directly participated in forming carboxyl of the acetic acid (5). Here we report the high-selective synthesis of liquid hydrocarbons via CO₂ hydrogenation at a lower temperature using a simple Co₆/MnO_x nanocatalyst. Interestingly, very little CO was detected in the product. Further study indicates that hydrogenation of CO₂ to form C₅₊ paraffins did not proceed through the CO pathway.

Results and Discussion

Catalyst Structure.

The Co₆/MnO_x nanocatalyst was prepared by a coprecipitation method. The transmission electron microscope (TEM) image shows that the size of the catalyst particles was mostly in the range of 25 nm to 40 nm (Fig. 1A). The N₂ adsorption test revealed that the Brunauer–Emmett–Teller surface area of the piled catalyst was only 49.4 m²/g (SI Appendix, Fig. S1), indicating that it was not a porous material. The metallic Co (Co⁰) catalyst and the Co−Mn bimetallic catalysts fabricated by the above method were all spherical-like, and the sizes of the bimetallic catalysts were slightly larger (SI Appendix, Fig. S2). In the TEM images of Co₆/MnO_x, the lattice spacings of 1.92 and 2.02 Å (Fig. 1B) corresponded to the (101) plane and the (002) plane of Co⁰, and the lattice spacing of 2.57 Å (Fig. 1C) could be assigned to the (111) plane of MnO. The X-ray diffraction (XRD) pattern also reveals two kinds of crystal phases, i.e., Co⁰ bulk phase and MnO, in the catalyst (SI Appendix, Fig. S3). It is known that cobalt oxides can be easily reduced to metallic Co by H₂ (18). The temperature-programmed reduction (TPR) analysis proved that the Mn atoms could evidently retard the reduction of Co oxides because strong interaction existed between Co−Mn atoms (SI Appendix, Fig. S4). The X-ray photoelectron spectroscopy (XPS) characterization suggests that Co²⁺ and Mn³⁺ were on the surface of the catalyst (SI Appendix, Fig. S5 A–C). The elementary mapping shows that Co, Mn, and O atoms were well dispersed in the catalyst (SI Appendix, Fig. S6). The inductively coupled plasma optical emission spectroscopy (ICP-OES) and XPS analysis demonstrated that compositions of the catalysts and catalyst surface were consistent with the design values (SI Appendix, Table S1). In brief, the Co₆/MnO_x consisted of Co⁰ and MnO crystallites, and strong interaction existed between the atoms.

Fig. 1.

TEM images of Co₆/MnO_x and fast Fourier transform patterns (Insets): the fresh catalyst (A−C); the catalyst after CO₂ hydrogenation (D−F), and the catalyst after CO hydrogenation (G−I).

Catalytic Performance.

The target reaction was effectively catalyzed by Co₆/MnO_x in squalane solvent at milder conditions (Table 1). Liquid hydrocarbons (C₅ to C₂₆) were the major product, and their selectivity in the total product was as high as 53.2 C-mol% (Table 1, entry 1). The alkane products were all linear alkanes (n-paraffins). The byproducts were mostly C₁ to C₄ hydrocarbons. The C atoms in the hydrocarbons were all from CO₂, because no product was observed without the reactant CO₂ (entry 2). The reaction temperature was 200 °C, which is much lower than those reported in the literature using a flow reactor. The detailed product distribution (SI Appendix, Fig. S7 A and B) indicated that a high proportion of n-paraffins, especially C₅₊ n-paraffins, was obtained in the reaction. The chain length distribution of the hydrocarbon products followed the Anderson−Schulz−Flory (ASF) statistics (SI Appendix, Fig. S7C). The chain growth probability (α) was 0.78, which is consistent with a large portion of longer-chain products. Interestingly, the amount of CO was very small (0.4 C-mol%). Further studies also showed that CO was hardly detectable during the whole reaction, which will be discussed in the following paragraphs. We would like to mention that the above results were collected using a batch reactor, where CO₂ and H₂ reacted on suspending catalyst particles under stirring. The product generated remained in the reactor during the reaction. While in a flow reactor, reactant gases of constant pressure passed through the packed catalyst bed, and the products generated were simultaneously taken out of the reactor by flowing gas.

Table 1.

CO₂ hydrogenation using different catalysts

Entry	Catalyst	Solvent	Selectivity (C-mol%)				Activity (mmolCO2·gcat−1·h−1)
			CO	C1–4	C5+	ROH
1*	Co6/MnOx	squalane	0.4	46.4	53.2	0	15.1
2†	Co6/MnOx	squalane	—	—	—	—	0
3	Co0	squalane	0.6	80.7	18.7	0	1.7
4	CoO	squalane	0	0	0	0	0
5	Co3O4	squalane	0	0	0	0	0
6	Mn	squalane	0	0	0	0	0
7	MnO	squalane	0	0	0	0	0
8	MnO2	squalane	0	0	0	0	0
9	Co14/MnOx	squalane	0.5	58.5	41.0	0	6.6
10	Co10/MnOx	squalane	0.5	47.2	52.3	0	11.2
11	Co2/MnOx	squalane	0.4	76.2	23.4	0	13.3
12	Co6/ZnOx	squalane	0.1	80.7	19.2	0	4.8
13	Co6/AlOx	squalane	0.1	94.2	5.7	0	4.1
14	Co6/CeOx	squalane	0.4	89.8	9.8	0	6.3
15	Co6/MnOx	—	0.7	60.9	38.4	0	11.3
16	Co6/MnOx	cyclohexane	0.4	48.3	51.3	0	14.7
17	Co6/MnOx	benzene	0.5	95.3	4.2	0	7.9
18	Co6/MnOx	DMI	0.6	87.6	6.8	5.0	2.6
19	Co6/MnOx	water	0.3	92.9	0	6.8	3.2

Reaction conditions: 20 mg of catalyst, 1.0 mL of solvent (squalane), 200 °C, 15 h, initial pressure 8 MPa (CO₂/H₂ = 1).

^*The conversion of CO₂ was 15.3%, and the selectivity of paraffin in the total hydrocarbons was 78.1 C-mol%; moreover, the selectivity of paraffin in the C₅₊ products was 67.2 C-mol%.

^†No CO₂ was added. The error of activity was ±1.0 mmol_CO2·g_cat⁻¹·h⁻¹ and the error of selectivity was ±1.5 C-mol%, which were obtained by three repeated experiments.

To optimize the catalyst, we tested the Co⁰ and its oxides (CoO, Co₃O₄) in CO₂ hydrogenation. The results revealed that Co⁰ could catalyze the synthesis of C₅₊ hydrocarbons at very low activity and C₅₊ selectivity, and the largest compound contained only 16 C atoms (Table 1, entry 3), whereas the Co oxides could not catalyze the reaction at all (entries 4, 5). We also used the metallic Mn and its oxides (MnO, MnO₂), but the reaction did not occur either (entries 6 to 8). Thus, we deduce that Co was the major catalyst, and Mn was the promoter. After testing the Co-Mn bimetallic catalysts of different atom ratios, we found that Co₆/MnO_x showed the best performance (entries 1, 9 to 11). We also tested other Co based bimetallic catalyst prepared by a similar method, including Co₆/ZnO_x, Co₆/AlO_x, and Co₆/CeO_x, but the results were poor (entries 12 to 14). Thus, Co₆/MnO_x was appropriate for the reaction.

The solvent was also important for the reaction. Without solvent, the reaction could take place at lower activity and selectivity (Table 1, entry 15). Both squalane and cyclohexane as solvent could effectively improve the catalytic performance, and selectivity in squalane was slightly higher (entries 1, 16). We also tested benzene and found that the solvent aromaticity inhibited the reaction (entry 17). When we used polar solvents, such as 1,3-Dimethyl-2-imidazolidinone (DMI) and water, both reaction activity and selectivity were much lower (entries 18, 19). Interestingly, minor methanol and ethanol were observed. In short, squalane solvent and Co₆/MnO_x catalyst were a suitable combination for the reaction.

Fig. 2A shows the effect of reaction temperature upon the reaction activity and selectivity. The C₅₊ products began to form at 140 °C. When the temperature was gradually elevated to 200 °C, the reaction rate and C₅₊ selectivity increased remarkably. The contribution of catalytic activity from 200 °C to 220 °C was not as significant as that from 180 °C to 200 °C. In addition, the increase of C₅₊ selectivity also became slower from 200 °C to 220 °C, suggesting that 200 °C was a suitable temperature. At this temperature, we studied the influence of the total pressure at an equal proportion of CO₂ and H₂ (Fig. 2B). Both the activity and C₅₊ selectivity were enhanced markedly with the elevating total pressure, while they changed slowly when the pressure exceeded 8 MPa. We further fixed the total pressure at 8 MPa and studied the influence of CO₂/H₂ ratio on the reaction (Fig. 2C). The activity rose rapidly with increasing CO₂ pressure, which became less evident when the ratio of CO₂/H₂ reached 4/4. On the other hand, the C₅₊ selectivity decreased with elevating CO₂ pressure, and the rate of decrease became faster after the CO₂/H₂ ratio was larger than 4/4. Thus, 4 MPa of CO₂/4 MPa of H₂ was the best for the reaction. Fig. 2D illustrates the time course of the reaction. The CO₂ was consumed continuously during the reaction, and the C₅₊ selectivity rose gradually with reaction time. The GC analysis also verified that the chain length of the products increased with increasing reaction time. It is noteworthy that, at all above conditions, CO was hardly detectable.

Fig. 2.

Effect of reaction conditions [(A) reaction temperature, (B) total pressure, (C) CO₂/H₂ ratio, and (D) reaction time] on the CO₂ hydrogenation using 20 mg of Co₆/MnO_x and 1 mL of squalane. Other conditions: (A) 4 MPa of CO₂ and 4 MPa of H₂, 15 h; (B) P_CO2/P_H2 = 1, 200 °C, 15 h; (C) initial pressure 8 MPa, 200 °C, 15 h; and (D) 4 MPa of CO₂ and 4 MPa of H₂, 200 °C. Symbols: triangle, C₅₊ selectivity; circle, C_1–4 selectivity; square, CO selectivity; star, activity; diamond, CO₂ consumption.

The reusability of the catalyst is shown in SI Appendix, Fig. S8A. A trend of slight decrease in selectivity and activity was observed. The TEM, N₂ adsorption, XRD, and XPS characterization also showed that the catalyst changed slightly after the recycling test (Fig. 1 and SI Appendix, Figs. S1, S3, and S5). In the hot filtration test, no product was generated after the catalyst was filtered out, indicating that the catalyst worked in a heterogeneous mode (SI Appendix, Fig. S8B). It should be mentioned that industrial use of a heterogeneous catalyst usually operates in a flow reactor. Further work should be done to optimize operation conditions using a flow reactor for practical application.

Reaction Mechanism.

Co⁰ is a well-known FTS catalyst using syngas, but it is generally regarded as a methanation catalyst in CO₂ hydrogenation (9, 16). This can be ascribed to its weak CO₂ adsorption and strong H₂ adsorption ability, which results in high H/C ratio on the catalyst surface. The high H/C ratio inhibited the chain growth process and favored methane generation (19). Inspired by this, we tested the temperature-programmed desorption of CO₂ (CO₂-TPD) of Co⁰ catalyst and the Co−Mn catalysts. The results demonstrated that the amount of CO₂ adsorbed on the Co−Mn catalysts was much higher than that on Co⁰ (Fig. 3A). In addition, CO₂ adsorption on Co₆/MnO_x was the highest in all of the catalysts, especially near the reaction temperature. The CO₂-TPD data were consistent with the catalytic results. The CO₂ desorbed at a higher temperature (e.g., 188 °C) directly correlated with the C₅₊ selectivity. The H₂ adsorption ability of the catalyst was also crucial. The H₂-TPD analysis revealed that the amount of H₂ adsorbed on Co₆/MnO_x was remarkably lower than that on the Co⁰ catalyst, especially at higher temperature (150 °C to 250 °C; Fig. 3B). The above results verified that the C/H ratio on Co₆/MnO_x must be much higher than that on Co⁰, accounting for the exceptional behavior of Co₆/MnO_x. We would like to mention that competitive adsorption of the reactant gases (e.g., CO₂, H₂) on the catalyst surface existed at reaction conditions. We further conducted the CO₂-TPD and H₂-TPD characterization of other Co-based bimetallic catalysts (Co₆/ZnO_x, Co₆/AlO_x, Co₆/CeO_x), as shown in Fig. 3 C and D. The results demonstrated that addition of Zn, Al, and Ce promoters excellently accelerated the CO₂ adsorption, but could not suppress or even promote the H₂ adsorption on the catalyst, while addition of Mn promoter to Co catalyst not only remarkably enhanced the CO₂ adsorption but also significantly suppressed the H₂ adsorption. This is the reason Mn is the appropriate promoter for producing long-chain paraffins. The TPD data could also explain why the C₅₊ selectivity over Co₆/ZnO_x was higher than those over Co₆/AlO_x and Co₆/CeO_x (Table 1, entries 12 to 14).

Fig. 3.

CO₂-TPD (A and C) and H₂-TPD (B and D) data of the catalysts. The TPD signals have been normalized to the mass of the tested samples.

We characterized Co₆/MnO_x catalyst pretreated with H₂ at 200 °C by the semi-in situ XPS method. The result revealed that, at reaction condition, Co⁰/Co²⁺ coexisted on the catalyst surface and Mn³⁺ was in situ reduced to Mn²⁺ (SI Appendix, Fig. S5 D–F). The outstanding performance of Co₆/MnO_x may be due to the synergy of the surface Co⁰, Co²⁺, and Mn²⁺ atoms during the reaction. The semi-in situ XPS study also manifested that the CO₂ adsorption was improved when the temperature increased from 50 °C to 150 °C (SI Appendix, Fig. S9). At 200 °C, the adsorbed CO₂ was reduced to hydrocarbons by the H atoms preadsorbed on the catalyst surface. To detect the intermediates of the reaction, we further conducted an in situ Fourier transform infrared (FTIR) study of CO₂ adsorption on the pretreated catalyst at 200 °C. The result of the in situ FTIR study is shown in Fig. 4. It can be seen that CO₂ was adsorbed on the catalyst surface and gradually reduced to CH₂/CH₃, via CO₂^δ−, HCOO⁻, −CH₂OH, and/or CH₃O⁻ intermediates (Fig. 4A). The weak peak at 1,434 cm⁻¹ is assigned to asymmetrically adsorbed CO₂ (CO₃²⁻). At 2 min, it was quickly reduced to CO₂^δ− by H atoms on the catalyst surface, as can be seen from the peaks at 1,289 and 1,606 cm⁻¹. At 2.5 min, it was observed that formate (HCOO⁻) species began to form by reducing CO₂^δ−. The peaks at 1,373 and 1,594 cm⁻¹ correspond to symmetric and asymmetric OCO stretches of HCOO⁻. The HCOO⁻ was further reduced to primary alcohol group (-CH₂OH, 1,028 cm⁻¹) and/or methoxy group (CH₃O⁻, 1,068 cm⁻¹). Finally, the −CH₂OH/CH₃O⁻ were reduced to CH₂ group (1,474 cm⁻¹) and/or CH₃ group (1,388 cm⁻¹). The broad peak around 1,474 cm⁻¹ revealed the coexistence of the CH₂ and CH₃ groups. No carboxylic acid, alcohol, or ether was detected in the reaction product, indicating that the HCOO⁻, −CH₂OH, and/or CH₃O⁻ intermediates were finally converted into CH₂ and/or CH₃. With the consumption of HCOO⁻, −CH₂OH, and/or CH₃O⁻ species, the peaks of CH₂ and CH₃ became more and more remarkable, which further proved such conversion. The peak at 843 cm⁻¹ can be attributed to out-of-plane bending vibration of C−H in alkenes. The assignment of the above peaks agrees well with that in the literature (20). No gaseous CO or adsorbed CO (CO_ad) was observed (Fig. 4B). The FTIR study suggested that the reaction did not proceed through the CO pathway.

Fig. 4.

In situ FTIR spectra [(A) key intermediates and (B) wide band] recorded during the CO₂ adsorption on Co₆/MnO_x. The catalyst was pretreated with H₂ at 200 °C for 2 h before the adsorption test. The black straight line at the bottom was the baseline recorded before CO₂ adsorption. The spectra were recorded after the CO₂ was introduced to the catalyst, and the time sequence, from bottom to top, was 2, 2.5, 4.5, 7.5, 15, 25, 35, 45, and 55 min.

Only a very little CO was detected during the reaction over Co₆/MnO_x (Fig. 2), which agreed with the very low RWGS ability of the Co catalyst (9, 16). Moreover, addition of Mn to Co-based catalyst may accelerate the dissociation and disproportionation of CO (21). To investigate the contribution of the trace CO in the C₅₊ hydrocarbons synthesis, we conducted the reaction using CO (0.5 MPa, 4 MPa) instead of CO₂. It was observed that the reaction rate was much lower, and little C₅₊ product was observed (SI Appendix, Table S2, entries 1, 2). The products were mostly C₁–C₄ hydrocarbons with a small amount of methanol and ethanol, which were different from those in CO₂ hydrogenation. The CO-TPD and H₂-TPD data indicate that addition of Mn promoter inhibited the adsorption of CO and H₂ on the Co catalyst, especially at the reaction temperature (200 °C) (Fig. 3B and SI Appendix, Fig. S10). Furthermore, most of the Co atoms on the catalyst surface were blocked by the carbon deposit after the CO hydrogenation (SI Appendix, Fig. S11). The TEM and XPS data also suggested that CO altered the Co/Mn valence and resulted in surface reconstruction (Fig. 1 and SI Appendix, Fig. S5). These facts help to explain the poor result of CO hydrogenation over Co₆/MnO_x. We further conducted a ¹³CO labeling test by adding a small amount of ¹³CO in the reactant gases (i.e., 0.2 MPa of ¹³CO, 3.8 MPa of CO₂, 4 MPa of H₂). The gas chromatography mass spectrometry (GC-MS) spectra of this test are shown in SI Appendix, Fig. S12. As expected, both ¹³CO and CO₂ could generate the target paraffins, whereas the ¹³C of ¹³CO did not enter the products generated from CO₂, and the CO₂ did not take part in the products formed by ¹³CO either. This phenomenon revealed that the CO₂ and ¹³CO formed long-chain paraffins separately and independently in the reaction. There is a strong possibility that hydrogenation of CO₂ and CO proceeded on segregated and different sites of the catalyst. Otherwise, the ¹³CH₂ monomer from ¹³CO and the CH₂ monomer from CO₂ would enter the same paraffin molecule. The surface segregation of the cobalt catalyst during the FTS reaction has been reported elsewhere (22). The results of the ¹³CO labeling test further confirmed that hydrogenation of CO₂ to produce paraffins over Co₆/MnO_x did not proceed via CO. We also used CH₃OH or HCOOH, instead of CO₂, to react with H₂, but the results were poor (SI Appendix, Table S2, entries 3 and 4). This may be because the dissociative adsorption of formic acid and methanol on the catalyst surface was difficult at the reaction condition, which was necessary for further transformation.

Based on all of the results above, we proposed the possible reaction mechanism. Firstly, CO₂ and H₂ adsorbed on the Co₆/MnO_x surface, where the Mn promoter enhanced the CO₂ adsorption and weakened the H₂ adsorption. Secondly, the adsorbed CO₂ was reduced to CH₂ and CH₃ by H atoms on the catalyst, via CO₂^δ−, HCOO⁻, −CH₂OH, and/or CH₃O⁻ intermediates. The CH₂ is known as the principle monomer in FTS reactions, while the CH₃ species are common starters for chains to grow (22). Finally, the liquid hydrocarbons were formed by chain growth steps with CH₂ and CH₃. The largest molecule in the product had 26 C atoms, and the product distribution followed the ASF statistics (α = 0.78), indicating that the chain growth steps are very similar to those of the well-known Co⁰ catalyzed FTS reaction (22). The presence of Co⁰ atoms on the Co₆/MnO_x surface at reaction condition further supported this mechanism. The major mechanistic contribution of this work is direct formation of CH₂/CH₃ in CO₂ hydrogenation, not via CO. Obviously, the parts of the pathway related to FTS are consistent with the well-established Fischer−Tropsch chemistry (e.g., the monomer, the chain growth process, and the ASF distribution of products).

Concluding Remarks

In summary, we discovered a simple Co₆/MnO_x nanocatalyst that could efficiently catalyze CO₂ hydrogenation to normal C₅₊ hydrocarbons at a lower temperature. Co was the major catalyst, while Mn promoter enhanced CO₂ adsorption and weakened H₂ adsorption on the catalyst. During the reaction, CO₂ adsorbed on the catalyst and was reduced by H atoms to CH₂ monomer and CH₃ species, via CO₂^δ−, HCOO⁻, −CH₂OH, and/or CH₃O⁻ intermediates. The liquid hydrocarbons were generated by further chain growth steps, which are similar to those of the well-known Co⁰ catalyzed FTS reaction. The reaction could proceed by avoiding CO, which opens an avenue for liquid fuel synthesis and CO₂ transformation.

Materials and Methods

Catalyst Preparation.

The Co−Mn bimetallic catalysts were prepared by a coprecipitation method. In a typical procedure, Co(NO₃)₂·6H₂O (0.873 g, 3.0 mmol) and Mn(NO₃)₂·4H₂O (0.125 g, 0.5 mmol) were dissolved in 40 mL of water and were stirred for 1 h. Then the solution was added into 100 mL of Na₂CO₃ solution (0.5 mol/L) within 1 h of stirring. After stirring for another 3 h, the solid was centrifuged and washed with 500 mL of water. It was dried at 110 °C for 10 h and calcined at 400 °C for 3 h, and was reduced under hydrogen flow at 400 °C for 1 h. Finally, the catalyst was passivated with a low content of oxygen (e.g., 1%) in N₂ at room temperature for 30 min, and Co₆/MnO_x was obtained. The catalysts with Co/Mn molar ratios of 2, 10, and 14 were also prepared using a similar method. Co₆/ZnO_x, Co₆/AlO_x, and Co₆/CeO_x were also fabricated by the above method, using Co(NO₃)₂·6H₂O to react with Zn(NO₃)₂·6H₂O, Al(NO₃)₃·9H₂O, and Ce(NO₃)₃·6H₂O, respectively. Furthermore, the metallic Co catalyst (Co⁰) was synthesized by a similar method, during which Co(NO₃)₂·6H₂O (0.873 g, 3.0 mmol) was used as precursor. The source and specification of the chemicals used in this work are given in SI Appendix. The catalysts fabricated above had good air stability.

Catalyst Characterization.

The as-prepared catalysts were characterized by N₂ adsorption, XRD, XPS, TEM, ICP-OES, H₂-TPR, and TPD. More experimental details are provided in SI Appendix.

In Situ Characterization.

The semi-in situ XPS measurement was aided by a special vacuum seal sample unit to transfer the sample prepared in the glove box. To study the CO₂ adsorption, Co₆/MnO_x was pretreated with 4 MPa of H₂ at 200 °C for 2 h. The reactor was cooled, and the H₂ was released and replaced with 1 MPa of CO₂ three times. Then the reactor was charged with 2 MPa of CO₂ and kept at the specified temperature for 1 h. Finally, the reactor was cooled to room temperature, and the gases were released. The sample was prepared in a glove box, and transferred to the sample chamber of an XPS spectrometer using the sample unit. To reduce the loss of adsorbates during the XPS analysis, the sample chamber was kept at ultra-low temperature using liquid nitrogen.

The in situ FTIR spectra were recorded with a Nicolet FTIR Spectrometer 6700 equipped with a liquid-nitrogen-cooled narrow-band mercury–cadmium–telluride detector and diffuse reflectance infrared Fourier transform spectroscopy optics. The details of the apparatus and operation procedure were described elsewhere (23). To get clear spectra, the black Co₆/MnO_x and KBr powders were mixed and ground together. Then the mixture was pressed into the sample cup. At 200 °C, the sample was firstly purged with N₂ for 2 h, then it was treated with H₂ for 2 h. After further purging the sample with N₂ for 1 h, CO₂ was introduced to the sample, and the CO₂ adsorption began. All of the above experiments were conducted at ambient pressure. The flow rate of the above gases was 200 mL/min. A background spectrum was recorded before CO₂ adsorption. Spectra were collected as difference spectra with the background spectrum. The surface products were shown as positive bands, while loss of surface species was shown as negative bands. In this way, the formation and disappearance of the surface species during the process can be clearly recorded.

CO₂ Hydrogenation.

The CO₂ hydrogenation was conducted in a 16-mL stainless steel batch reactor with a Teflon lining. In a typical experiment, the catalyst (20 mg) and solvent (1.0 mL) were added in the reactor under air. The reactor was sealed, and the air was removed by flushing with 1 MPa of CO₂ three times. Then, 4 MPa of CO₂ and 4 MPa of H₂ were charged at room temperature. The reaction was performed at 200 °C for the desired time. Then the reactor was cooled in an ice-water bath, and the gas was released slowly and collected for analysis in an HP 4890B GC with thermal conductivity detector. The liquid products were analyzed in an HP 7890B GC with flame ionization detector using toluene as an internal standard. The liquid products were identified using GC-MS (Agilent-7890B-5977A) as well as by comparing the retention times with the standards in the GC traces. The yields of the products were calculated from the GC data. The conversion of CO₂ was calculated from the amount of reacted CO₂ divided by the total CO₂ added to the reactor. The total amount of CO₂ charged into the reactor could be obtained by weighing, and the reacted CO₂ was known from the amount of total carbon in the product.

The Recycling Test.

The solid catalyst was separated and washed with cyclohexane (5 × 3 mL) and dried under vacuum at 60 °C for 10 h. Then the catalyst was used directly for the next run.

The Hot Filtration Test.

The hot filtration test was conducted at typical reaction conditions. After 6 h, the catalyst was filtered, and the solution was analyzed by GC. The reaction in the resulting solution proceeded for another 9 h, and then the reaction solution was analyzed again by GC.