Preparation and Performances of ZIF-67-Derived FeCo Bimetallic Catalysts for CO 2 Hydrogenation to Light Olefins

A novel sodium-promoted Fe-Co/NC catalyst prepared by incipient-wet-impregnation method using ZIF-67 as a support was employed to convert CO2 to light olefins through hydrogenation reaction. Properties of the synthesized catalysts calcinated at various temperatures (from 400 to 700 °C) were investigated by XRD, SEM, TEM and Mӧssbauer spectroscopy. Characterization results showed that the support could be fully converted into carbon support above 500 °C, which could anchor metal particles, thus resulting in a uniform dispersion of active components. Furthermore, the Fe-Co alloy was formed during N2 calcination, and was converted into active components, such as Fe3O4, Fe5C2, and Co2C during the reaction. The reaction result indicated that FeCo/NC-600 catalyst exhibited the highest selectivity of light olefins (C2= − C4=, 27%) and CO2 conversion could reach around 37% when this catalyst pyrolyzed at 600 °C in N2. The highest selectivity for light olefins may be related to the combination of suitable particle size and sufficient active sites of iron carbide.


Introduction
Due to the global warming concerns and the increasing of ocean acidity, highly efficient capture, and utilization of CO 2 has attracted much attention [1][2][3]. CO 2 can be used in food additives, oil and gas extraction, and synthesis of chemical material [4], while several high value-added products, such as methane, olefins, liquid fuels, and alcohols can be obtained by CO 2 hydrogenation reaction. Among them, light olefins (C 2 = − C 4 =) are kinds of indispensable chemicals, and a series of studies have been carried out to research the hydrogenation of CO 2 to light olefins [5][6][7]. However, adjusting the ratio of different active sites to obtain the best adsorption competition between CO 2 and CO, and obtaining high CO 2 conversion while inhibiting high selectivity of methane and light saturated hydrocarbons are still huge challenges [8].
To our knowledge, Fe-based catalysts have been extensively applied in the CO 2 hydrogenation reaction due to similar characteristics to the traditional Fischer-Tropsch (FT) synthesis process [9][10][11][12]. In general, the mechanism of the CO 2 hydrogenation reaction over Fe-based catalysts is based on a revised Fischer-Tropsch path, in which CO is firstly generated by the reverse water gas reaction (RWGS), and then hydrocarbon compounds were generated by the FT reaction [13,14]. Fe 3 O 4 and Table 1 displayed the elemental composition and physical properties of the synthesis samples. The catalyst element composition was shown in Table 1 under different pretreatment temperatures, of which the relative proportion of cobalt and iron remain at 2, and the metal content increased ascribing to the H/N loss in MOFs during the pyrolysis process. Figure S1 shows the TG and DTG curves of ZIF-67 impregnated with iron. The catalyst had four-step thermal decomposition behavior: The first weight loss around 120 • C was considered to be related to the loss of physically adsorbed water. The weight loss with increasing temperature to around 150-200 • C may be due to the release of residual molecules (dimethylimidazole or methanol) and gas from the activation site [48]. The slight decomposition at about 300 • C may be attributed to the loss of iron acetylacetonate. On the other hand, due to the decomposition of the skeleton, ZIF-67 had a significant mass loss at 400-500 • C [49].  V meso (cm − The nitrogen adsorption-desorption isotherms of FeCo/NC-T are shown in Figure 1 and Table 1 to detect the structural properties. According to IUPAC classification, ZIF-67 shows a sharp N 2 uptake at a lower relative pressure (<0.05), suggesting that ZIF-67 has a good microporous structure [50]. However, FeCo/NC-T displays a type IV isotherm with the H3 hysteresis loop. The hysteresis loops caused by the capillary condensation closed at P/P 0 ≈ 0.4, demonstrating the generation of mesopores [51]. The pore structure of ZIF-67 was gradually destroyed due to the raise of pyrolysis temperature, and part of the specific surface area sacrificed. Due to incomplete pyrolysis at 400 • C, part of the skeletal structure was retained, and the specific surface area of micropores decreased from 1265 to 454 m 2 g −1 .
As the temperature further increased, the micropores of support gradually disappeared, and part of the micropores were transformed into mesopores [52].
The morphology and structure of prepared ZIF-67 and its derivatives by SEM are shown in Figure 2. ZIF-67 shows a typical rhombic dodecahedron shape ( Figure 2a). As the pyrolysis temperature increased, the original structure of ZIF-67 was damaged, the particle surface became very rough and the boundary of the particle interface became blurred. Some self-grown carbon nanotubes were detected when temperature reached above 500 • C as shown in Figure 2b,d-f, which may be related to the catalytic growth induced by iron [53][54][55].  XRD patterns of FeCo/NC-T are displayed in Figure 3. FeCo/NC-400 shows characteristic diffraction peaks ascribing to the reserved ZIF-67 framework structure at such temperature. Peaks at 30.08°, 35.43°, 43.06°, 56.94°, and 62.53° were attributed to the CoFe2O4 crystal phase (JCPDS no. 79-1744). When pyrolysis temperature reached above 500 °C, the support skeleton completely collapsed, and only a peak of 44.86° was observed, which was assigned to the FeCo alloy phase (JCPDS no. . With the raise of pyrolysis temperature, the intensity of the Fe-Co alloy diffraction peaks increased, and the FeCo/NC-700 demonstrated the best crystallinity.   XRD patterns of FeCo/NC-T are displayed in Figure 3. FeCo/NC-400 shows characteristic diffraction peaks ascribing to the reserved ZIF-67 framework structure at such temperature. Peaks at 30.08°, 35.43°, 43.06°, 56.94°, and 62.53° were attributed to the CoFe2O4 crystal phase (JCPDS no. 79-1744). When pyrolysis temperature reached above 500 °C, the support skeleton completely collapsed, and only a peak of 44.86° was observed, which was assigned to the FeCo alloy phase (JCPDS no. . With the raise of pyrolysis temperature, the intensity of the Fe-Co alloy diffraction peaks increased, and the FeCo/NC-700 demonstrated the best crystallinity. XRD patterns of FeCo/NC-T are displayed in Figure 3. FeCo/NC-400 shows characteristic diffraction peaks ascribing to the reserved ZIF-67 framework structure at such temperature. Peaks at 30.08 • , 35.43 • , 43.06 • , 56.94 • , and 62.53 • were attributed to the CoFe 2 O 4 crystal phase (JCPDS no. 79-1744). When pyrolysis temperature reached above 500 • C, the support skeleton completely collapsed, and only a peak of 44.86 • was observed, which was assigned to the FeCo alloy phase (JCPDS no. . With the raise of pyrolysis temperature, the intensity of the Fe-Co alloy diffraction peaks increased, and the FeCo/NC-700 demonstrated the best crystallinity.  The TEM result is shown in Figure 4 and Figure S2. On one hand, with the rise of pyrolysis temperature, the average particle size of the metal increased. The average particle diameters of FeCo/NC-400 to FeCo/NC-700 were 9.9 nm, 12.9 nm, 14.9 nm, and 40.8 nm, respectively ( Figure S2ad). Metal agglomeration became obvious when the pyrolysis temperature reached 700 °C. Moreover, pyrolysis at high temperature exposed more cobalt nanoparticles, which promoted formation of iron-cobalt alloys. On the other hand, it can be found from Figure 4 that as the temperature reached above 500 °C, a small amount of carbon nanotubes with a thickness of ca. 8-12 nm formed on the support (Figure 4b-d). Except for the metal nanoparticles distributed on the outer surface of the support, some metals are encapsulated in the mouth of the carbon nanotubes [56].  The TEM result is shown in Figure 4 and Figure S2. On one hand, with the rise of pyrolysis temperature, the average particle size of the metal increased. The average particle diameters of FeCo/NC-400 to FeCo/NC-700 were 9.9 nm, 12.9 nm, 14.9 nm, and 40.8 nm, respectively ( Figure S2a-d).
Metal agglomeration became obvious when the pyrolysis temperature reached 700 • C. Moreover, pyrolysis at high temperature exposed more cobalt nanoparticles, which promoted formation of iron-cobalt alloys. On the other hand, it can be found from Figure 4 that as the temperature reached above 500 • C, a small amount of carbon nanotubes with a thickness of ca. 8-12 nm formed on the support (Figure 4b-d). Except for the metal nanoparticles distributed on the outer surface of the support, some metals are encapsulated in the mouth of the carbon nanotubes [56].  The TEM result is shown in Figure 4 and Figure S2. On one hand, with the rise of pyrolysis temperature, the average particle size of the metal increased. The average particle diameters of FeCo/NC-400 to FeCo/NC-700 were 9.9 nm, 12.9 nm, 14.9 nm, and 40.8 nm, respectively ( Figure S2ad). Metal agglomeration became obvious when the pyrolysis temperature reached 700 °C. Moreover, pyrolysis at high temperature exposed more cobalt nanoparticles, which promoted formation of iron-cobalt alloys. On the other hand, it can be found from Figure 4 that as the temperature reached above 500 °C, a small amount of carbon nanotubes with a thickness of ca. 8-12 nm formed on the support (Figure 4b-d). Except for the metal nanoparticles distributed on the outer surface of the support, some metals are encapsulated in the mouth of the carbon nanotubes [56].   The structure of the catalyst was further studied by HRTEM, which is shown in Figure 5. When pyrolyzed at 400 • C (Figure 5a), the measured lattice fringes were 0.201 nm, 0.242 nm, and 0.484 nm, which corresponded to the crystal plane of FeCo (110), CoFe 2 O 4 (006), and CoFe 2 O 4 (003), respectively. When the temperature reached above 500 • C, only the FeCo (110) crystal plane could be found (Figure 5b), which might be attributed to the gas evolution with oxygen atoms during the decomposition of the ligand or the unstable decomposition of the oxide. In addition, the metal nanoparticles produced by MOFs carbonization were wrapped with graphitized carbon, which could effectively protect metal from sintering. However, when the pyrolysis temperature reaches 700 • C, the metal will obviously agglomerate without being protected by carbon [57].
The structure of the catalyst was further studied by HRTEM, which is shown in Figure 5. When pyrolyzed at 400 °C (Figure 5a), the measured lattice fringes were 0.201 nm, 0.242 nm, and 0.484 nm, which corresponded to the crystal plane of FeCo (110), CoFe2O4 (006), and CoFe2O4 (003), respectively. When the temperature reached above 500 °C, only the FeCo (110) crystal plane could be found (Figure 5b), which might be attributed to the gas evolution with oxygen atoms during the decomposition of the ligand or the unstable decomposition of the oxide. In addition, the metal nanoparticles produced by MOFs carbonization were wrapped with graphitized carbon, which could effectively protect metal from sintering. However, when the pyrolysis temperature reaches 700 °C, the metal will obviously agglomerate without being protected by carbon [57].
To observe the distribution of element species in detail, high angle annular dark field scanning transmission electron microscopy (HAADF-STEM) and EDS element mapping was used to analyze FeCo/NC-600. The metal nanoparticles (NPS) were relatively uniformly dispersed on the support (Figure 6b-d). The results show that in the particles of stem marked with a green line (Figure 6a), distribution position of Fe element was almost identical with that of Co element, but the Fe-rich position deviated from the Co-rich position in the NPs (Figure 5c). This result further proved the existence of FeCo alloy, which was corresponded to the XRD results.  To study the relationship between the degree of graphitization and the calcination temperature, FeCo/NC-T was analyzed by Raman spectra as displayed in Figure 7a. The D-band at 1345 cm −1 and the G-band at 1580 cm −1 were observed, attributed to disordered carbon and graphitic carbon [58]. To observe the distribution of element species in detail, high angle annular dark field scanning transmission electron microscopy (HAADF-STEM) and EDS element mapping was used to analyze FeCo/NC-600. The metal nanoparticles (NP S ) were relatively uniformly dispersed on the support (Figure 6b-d). The results show that in the particles of stem marked with a green line (Figure 6a), distribution position of Fe element was almost identical with that of Co element, but the Fe-rich position deviated from the Co-rich position in the NPs (Figure 5c). This result further proved the existence of FeCo alloy, which was corresponded to the XRD results.
Catalysts 2020 6 of 16 The structure of the catalyst was further studied by HRTEM, which is shown in Figure 5. When pyrolyzed at 400 °C (Figure 5a), the measured lattice fringes were 0.201 nm, 0.242 nm, and 0.484 nm, which corresponded to the crystal plane of FeCo (110), CoFe2O4 (006), and CoFe2O4 (003), respectively. When the temperature reached above 500 °C, only the FeCo (110) crystal plane could be found (Figure 5b), which might be attributed to the gas evolution with oxygen atoms during the decomposition of the ligand or the unstable decomposition of the oxide. In addition, the metal nanoparticles produced by MOFs carbonization were wrapped with graphitized carbon, which could effectively protect metal from sintering. However, when the pyrolysis temperature reaches 700 °C, the metal will obviously agglomerate without being protected by carbon [57].
To observe the distribution of element species in detail, high angle annular dark field scanning transmission electron microscopy (HAADF-STEM) and EDS element mapping was used to analyze FeCo/NC-600. The metal nanoparticles (NPS) were relatively uniformly dispersed on the support (Figure 6b-d). The results show that in the particles of stem marked with a green line (Figure 6a), distribution position of Fe element was almost identical with that of Co element, but the Fe-rich position deviated from the Co-rich position in the NPs (Figure 5c). This result further proved the existence of FeCo alloy, which was corresponded to the XRD results.  To study the relationship between the degree of graphitization and the calcination temperature, FeCo/NC-T was analyzed by Raman spectra as displayed in Figure 7a. The D-band at 1345 cm −1 and the G-band at 1580 cm −1 were observed, attributed to disordered carbon and graphitic carbon [58]. To study the relationship between the degree of graphitization and the calcination temperature, FeCo/NC-T was analyzed by Raman spectra as displayed in Figure 7a. The D-band at 1345 cm −1 and Catalysts 2020, 10, 455 7 of 17 the G-band at 1580 cm −1 were observed, attributed to disordered carbon and graphitic carbon [58]. The intensity proportion of the D-band to the G-band (ID/IG) shows the degree of graphitization of the carbon material. With the increasing of pyrolysis temperature, ID/IG decreased from 1.04 to 0.89, indicating that the graphitization degree became larger. In addition, the peak around 1425 cm −1 at 400 • C could be attributed to the N = N stretching vibration during the formation of the C-N = N bond [37]. the carbon material. With the increasing of pyrolysis temperature, ID/IG decreased from 1.04 to 0.89, indicating that the graphitization degree became larger. In addition, the peak around 1425 cm −1 at 400 °C could be attributed to the N = N stretching vibration during the formation of the C-N = N bond [37].
To explore the basicity of the catalyst surface and the amount of CO2 adsorbed, we used the CO2-TPD method on the FeCo/NC-T catalyst (Figure 7b). In the test temperature range, three kinds of desorption peaks of 120 °C, 270 °C, and 350 °C were mainly observed. Among them, peaks around 120 °C comes from the weakly adsorbed alkaline sites, and the peaks at 270 °C and 350 °C correspond to the alkaline sites of medium intensity chemisorption [33,59]. It can be seen that with the raise of pyrolysis temperature, the adsorption of medium intensity became less and the weak physical adsorption increased. Table S1 shows the amount of adsorbed CO2. With the rise of pyrolysis temperature, the quantity of CO2 adsorbed on the surface decreased. More adsorption of CO2 would contribute to higher CO2 conversion. XPS spectra were taken to study the chemical composition of nitrogen on the catalyst surface ( Figure 8). The high-resolution N 1s spectra could be divided into three subpeaks at 398.5, 400.5, and 401.1, which belonged to the three N species of pyridinic N, pyrrolic N, and graphitic N (Figure 8ad). The quantity of three N species was evaluated by fitting the area to the N 1s curve. The results show that with the rise of pyrolysis temperature, the total N content and pyridine nitrogen content decrease, and the graphite nitrogen content increased (Figure 8e) [35,60].
The lower pyrolysis temperature led to a very low degree of graphitization of the support NC layer, and no graphitic nitrogen was found in FeCo/NC-400 ( Figure 8a). As the temperature reached above 500 °C, the graphitization degree of the support increased and a small amount of graphitic nitrogen began to form (Figure 8b-d) [61]. The presence of nitrogen can provide anchor sites, which promotes high dispersion of the metal [40]. Pyridine-N would regulate the surface electron density and match the active metal, and then affect the CO2 activity [62]. To explore the basicity of the catalyst surface and the amount of CO 2 adsorbed, we used the CO 2 -TPD method on the FeCo/NC-T catalyst (Figure 7b). In the test temperature range, three kinds of desorption peaks of 120 • C, 270 • C, and 350 • C were mainly observed. Among them, peaks around 120 • C comes from the weakly adsorbed alkaline sites, and the peaks at 270 • C and 350 • C correspond to the alkaline sites of medium intensity chemisorption [33,59]. It can be seen that with the raise of pyrolysis temperature, the adsorption of medium intensity became less and the weak physical adsorption increased. Table S1 shows the amount of adsorbed CO 2 . With the rise of pyrolysis temperature, the quantity of CO 2 adsorbed on the surface decreased. More adsorption of CO 2 would contribute to higher CO 2 conversion.
XPS spectra were taken to study the chemical composition of nitrogen on the catalyst surface ( Figure 8). The high-resolution N 1s spectra could be divided into three subpeaks at 398.5, 400.5, and 401.1, which belonged to the three N species of pyridinic N, pyrrolic N, and graphitic N (Figure 8a-d). The quantity of three N species was evaluated by fitting the area to the N 1s curve. The results show that with the rise of pyrolysis temperature, the total N content and pyridine nitrogen content decrease, and the graphite nitrogen content increased (Figure 8e) [35,60]. The intensity proportion of the D-band to the G-band (ID/IG) shows the degree of graphitization of the carbon material. With the increasing of pyrolysis temperature, ID/IG decreased from 1.04 to 0.89, indicating that the graphitization degree became larger. In addition, the peak around 1425 cm −1 at 400 °C could be attributed to the N = N stretching vibration during the formation of the C-N = N bond [37].
To explore the basicity of the catalyst surface and the amount of CO2 adsorbed, we used the CO2-TPD method on the FeCo/NC-T catalyst (Figure 7b). In the test temperature range, three kinds of desorption peaks of 120 °C, 270 °C, and 350 °C were mainly observed. Among them, peaks around 120 °C comes from the weakly adsorbed alkaline sites, and the peaks at 270 °C and 350 °C correspond to the alkaline sites of medium intensity chemisorption [33,59]. It can be seen that with the raise of pyrolysis temperature, the adsorption of medium intensity became less and the weak physical adsorption increased. Table S1 shows the amount of adsorbed CO2. With the rise of pyrolysis temperature, the quantity of CO2 adsorbed on the surface decreased. More adsorption of CO2 would contribute to higher CO2 conversion. XPS spectra were taken to study the chemical composition of nitrogen on the catalyst surface ( Figure 8). The high-resolution N 1s spectra could be divided into three subpeaks at 398.5, 400.5, and 401.1, which belonged to the three N species of pyridinic N, pyrrolic N, and graphitic N (Figure 8ad). The quantity of three N species was evaluated by fitting the area to the N 1s curve. The results show that with the rise of pyrolysis temperature, the total N content and pyridine nitrogen content decrease, and the graphite nitrogen content increased (Figure 8e) [35,60].
The lower pyrolysis temperature led to a very low degree of graphitization of the support NC layer, and no graphitic nitrogen was found in FeCo/NC-400 ( Figure 8a). As the temperature reached above 500 °C, the graphitization degree of the support increased and a small amount of graphitic nitrogen began to form (Figure 8b-d) [61]. The presence of nitrogen can provide anchor sites, which promotes high dispersion of the metal [40]. Pyridine-N would regulate the surface electron density and match the active metal, and then affect the CO2 activity [62].  Table 2 summarizes the relationship between CO2 conversion and product selectivity under a prepared catalyst during 32 h on stream. Additionally, FeCo/Al2O3 was evaluated as a comparison. The CO2 conversion under FeCo/NC-400 and FeCo/NC-500 were 48.37% and 47.45%, which were higher than that of FeCo/NC-600 (37.03%) and FeCo/NC-700 (35.89%). FeCo/NC-400 shows the highest activity, ascribing to enhanced adsorbed CO2 detected from CO2-TPD. The turnover frequency (TOF) value of the FeCo/NC-T catalyst decreased with the increase of the pyrolysis temperature, which is the same change trend as the CO2 conversion. It shows that after eliminating the influence of metal content, pyrolysis under low temperature conditions will facilitate the formation of iron oxides, thereby providing higher RWGS activity.

Activity Test
In addition, Pyridine-N can benefit the electron transfer of the catalyst and provide more anchor points for the metal active phase. XPS analysis of FeCo/NC-400 demonstrated the highest content of pyridine-N. Moreover, the smaller particle size of FeCo/NC-400 was helpful to more exposed active sites. The connection of Fe and O facilitated the Fe3O4 formation, which acts as the RWGS active phase. The result is shown in Mӧssbauer spectra as follows.  The lower pyrolysis temperature led to a very low degree of graphitization of the support NC layer, and no graphitic nitrogen was found in FeCo/NC-400 ( Figure 8a). As the temperature reached above 500 • C, the graphitization degree of the support increased and a small amount of graphitic nitrogen began to form (Figure 8b-d) [61]. The presence of nitrogen can provide anchor sites, which promotes high dispersion of the metal [40]. Pyridine-N would regulate the surface electron density and match the active metal, and then affect the CO 2 activity [62]. Table 2 summarizes the relationship between CO 2 conversion and product selectivity under a prepared catalyst during 32 h on stream. Additionally, FeCo/Al 2 O 3 was evaluated as a comparison. The CO 2 conversion under FeCo/NC-400 and FeCo/NC-500 were 48.37% and 47.45%, which were higher than that of FeCo/NC-600 (37.03%) and FeCo/NC-700 (35.89%). FeCo/NC-400 shows the highest activity, ascribing to enhanced adsorbed CO 2 detected from CO 2 -TPD. The turnover frequency (TOF) value of the FeCo/NC-T catalyst decreased with the increase of the pyrolysis temperature, which is the same change trend as the CO 2 conversion. It shows that after eliminating the influence of metal content, pyrolysis under low temperature conditions will facilitate the formation of iron oxides, thereby providing higher RWGS activity.

Activity Test
In addition, Pyridine-N can benefit the electron transfer of the catalyst and provide more anchor points for the metal active phase. XPS analysis of FeCo/NC-400 demonstrated the highest content of pyridine-N. Moreover, the smaller particle size of FeCo/NC-400 was helpful to more exposed active sites. The connection of Fe and O facilitated the Fe 3 O 4 formation, which acts as the RWGS active phase. The result is shown in Mössbauer spectra as follows. Mössbauer spectra are taken to identify the iron phases of the catalysts. The 57 Fe Mössbauer spectra of the FeCo/NC-600 catalyst are presented in Figure 9. The hyperfine parameters including central shift (CS), quadrupole splitting (QS), and hyperfine magnetic field (H) are shown in Table  S2 and Figure S3. The spectrum (Figure 9a) consisted of one magnetic sextet with one central paramagnetic doublet. Stanfield and Delgass found [63] the hyperfine magnetic field within the range of 330-360 kOe (330 kOe for α-Fe). Hereby, the dominating sextet with parameters CS~0.013 mm/s and H~339.5 kOe was attributed to the α-(Fe,Co) alloy phase. The presence of the doublet could be due to the superparamagnetic (SP) effect of very small α-(Fe,Co) alloy particles (<10 nm), revealing the wide size distribution of metal particles. The after-reaction spectrum (Figure 9b) shows complex features with one doublet and six magnetic sextets, which can be assigned to iron oxides (FeOx) [64], α-(Fe,Co) alloy, and iron carbide (FeCx) [65] as listed in Table S2. The SP doublets in all after-reaction spectra had similar hyperfine parameters (CS~0.27-0.30 mm/s, QS~0.70-0.81 mm/s), which were ascribed to the ferric ions (Fe 3+ ) from the FeOx phase. Gnanamani reported [66] that the doublet appearing in the room-temperature spectrum was resolved into a magnetite phase at 20 K. Herein, even though without low-temperature Mössbauer spectra, the SP doublet of FeOx phase could be inferred, because the sextet of magnetite phase disappeared in the Mössbauer spectrum of the FeCo/NC-700 catalyst. The mole fraction of FeCo, FeOx, and FeCx phases in various catalysts after reaction is shown in Figure 9c. With the rise of pyrolysis temperature, the content of Fe 5 C 2 increased from 17% (FeCo/NC-400) to 55% (FeCo/NC-700). Correspondingly, the content of Fe 3 O 4 decreased from 44% to 27%.
Methane selectivity under FeCo/NC-600 (44.50%) and FeCo/NC-700(47.40%) were significantly lower than that under FeCo/NC-400 (67.31%) and FeCo/NC-500 (57.88%). Relatively, the selectivity of C 2 = − C 4 = shows an upward trend, from the lowest 11.14% (FeCo/NC-400) to the highest 27.05% (FeCo/NC-600). The result shows that with the rise of pyrolysis temperature, the CO 2 conversion and CH 4 selectivity decreased, but the selectivity of light olefins increased. Higher pyrolysis temperature promoted the production of iron carbides in the reaction, which was detected from Mössbauer spectroscopy. These iron carbides are beneficial to the production of olefins and inhibit methane production. At the same time, the decrease in the content of Fe 3 O 4 may result in a decrease in the activity of the RWGS reaction, which affects the conversion of CO 2 . However, when the pyrolysis temperature reached 700 • C, the metal carbide active sites exposed decreased due to obvious agglomeration, therefore FeCo/NC-600 exhibited the highest selectivity for low olefins (27.05%). In addition, the C 2 -C 4 olefins to paraffins ratio (O/P) of the product rise with increasing pyrolysis temperature, and FeCo/NC-700 shows the highest O/P (1.69). The decrease in the selectivity of C 2 -C 4 paraffins at 700 • C might be attributed to the decrease in active sites that hinder the readsorption of light olefins and subsequent chain growth processes, which led to an increase in the value of O/P. The basic characterization of the FeCo/Al 2 O 3 catalyst was added to the supporting materials ( Figure S6 and Table S3), including XRD, BET, inductively coupled plasma (ICP), SEM, etc. It can be seen from BET that FeCo/Al 2 O 3 had a sharp N 2 absorption at a lower pressure and a H1 hysteresis loop, indicating the presence of micropores and mesopores. The proportion of iron and cobalt obtained by ICP was consistent with FeCo/NC, which ensures similar reaction conditions. While, the result of SEM shows that the FeCo/Al 2 O 3 still retained a relatively good morphology when calcined at 600 • C. It can be seen from the XRD that iron and cobalt existed in the form of M 3 O 4 (M = Fe,Co) on Al 2 O 3 . The characteristic diffraction peaks of these two oxides were close to indistinguishable (Fe3O4 JCPDS no. 26-1136 and Co 3 O 4 JCPDS no. 80-1541), but they did not form an alloy phase like FeCo/NC, so the synergy of bimetals was not obvious. At the same time, the strong interaction between the Al 2 O 3 support and the metal was not conducive to the formation and stability of metal carbides. Therefore, although the FeCo/Al 2 O 3 catalyst had a high RWGS activity, it lacked the active site for generating and stabilizing olefins, resulting in undesirable methane and C 2 -C 4 paraffins with high selectivity. CO is a key factor for the reaction. Compared with a single iron, the addition of cobalt can effectively promote the conversion of CO. In the FTS process, the formation of alloy between cobalt and iron is beneficial to promote the oxidation and carburization of iron in the reaction. On the other hand, cobalt serves as the active site of FTS and formed the Co2C phase during the reaction, which would promote light olefin production. The addition of sodium could effectively stabilize the presence of the Co2C phase, as mentioned above. The metal Co will also generate some methane in the reaction, although the selectivity of methane has significantly decreased compared to the oxide support, from 76.19% to 44.50% (FeCo/NC-600). However, methane selectivity is still higher than expected due to the higher cobalt content in the support. To further regulate the active phase interface for better light olefin selectivity, it is important to properly control the metal content and the proportion of bimetal.
The XRD and HR-TEM results of the spent catalyst are shown in Figure 10. It can be obviously seen that the change of metal phase compared with that before the reaction (Figure 10a), and many new phases are shown in the XRD spectrum compared with spectrum before reaction (Figure 3). Fe3O4 and Fe5C2 phases could be observed, which are important active sites in the RWGS and FTS processes. In addition, the generation of Co2C phase was also observed, which is believed to be beneficial to the production of olefins in the FTS reaction [42]. The HR-TEM images are shown in CO is a key factor for the reaction. Compared with a single iron, the addition of cobalt can effectively promote the conversion of CO. In the FTS process, the formation of alloy between cobalt and iron is beneficial to promote the oxidation and carburization of iron in the reaction. On the other hand, cobalt serves as the active site of FTS and formed the Co 2 C phase during the reaction, which would promote light olefin production. The addition of sodium could effectively stabilize the presence of the Co 2 C phase, as mentioned above. The metal Co will also generate some methane in the reaction, although the selectivity of methane has significantly decreased compared to the oxide support, from 76.19% to 44.50% (FeCo/NC-600). However, methane selectivity is still higher than expected due to the higher cobalt content in the support. To further regulate the active phase interface for better light olefin selectivity, it is important to properly control the metal content and the proportion of bimetal.
The XRD and HR-TEM results of the spent catalyst are shown in Figure 10. It can be obviously seen that the change of metal phase compared with that before the reaction (Figure 10a), and many new phases are shown in the XRD spectrum compared with spectrum before reaction (Figure 3). Fe 3 O 4 and Fe 5 C 2 phases could be observed, which are important active sites in the RWGS and FTS processes. In addition, the generation of Co 2 C phase was also observed, which is believed to be beneficial to the production of olefins in the FTS reaction [42]. The HR-TEM images are shown in Figure 10b.The stripes with a lattice spacing of 0.280 nm measured on the spent FeCo/NC-600 catalyst correspond to the crystal plane of Fe 5 C 2 (021). This is because the CO intermediate formed during the reaction promotes the carburizing process of Fe. The phase transformation of iron is corresponding to the previous results of Mössbauer spectroscopy.

Preparation of Fe/ZIF-67
Fe/ZIF-67 catalysts were synthesized by traditional method of incipient wetness impregnation (IWI). Commonly, 0.2 g iron acetylacetonate was dissolved in 1.8 mL methanol. Subsequently, it was added to the surface of 1 g ZIF-67 sample with stirring. The obtained product was dried at 25 °C for 8 h and at 110 °C overnight, and the impregnation behavior was repeated five times. NaNO3 aqueous solution (0.136 g of NaNO3 dissolved in 1.5 g of water) was added to the sample obtained above, and the obtained product was dried at 25 °C for 8 h and dried at 110 °C overnight. The obtained sample was named Fe/ZIF-67.

Preparation of FeCo/NC
FeCo/NC was prepared by the pyrolysis of pre-synthesized Fe/ZIF-67. The sample was heated up to specific temperature in nitrogen atmosphere (40 ml min -1 )with a heating speed of 3 °C/min and held for 150 min, then cooled down to 25 °C. The obtained product was named as FeCo/NC-T.

Synthesis of FeCo/Al2O3
In the same way as above, an aqueous solution of 1.94 g of Fe (NO3)3 and 3.74 g of Co (NO3)2 was respectively mingled in deionized water, then it was added to Al2O3 (1.5 g). Promoted with the same amount of sodium for comparison. Subsequently the product was heated up to 600 °C in N2 atmosphere (40 mL min −1 )with a heating speed of 3 °C/min and held for 2 h, then cooling down to 30 °C. The obtained sample was named as FeCo/Al2O3.

Catalyst Characterization
The element content of the FeCo/NC catalyst was analyzed by an inductively coupled plasma (ICP, Optima 2100DV, PE, MA, USA), and the FeCo/NC samples need to be digested with a microwave before the test. X-ray diffraction (XRD, Bruker, Karlsruhe, Germany) of the samples were tested with a D8-Focus diffractometer at the scan speed of 8 °C /min from 5 to 80 °C by using Cu-Kα (40 kV, 200 mA) irradiation. Scanning electron microscopy (SEM, HITACHI, Tokyo, Japan) was tested on a Regulus 8100 (30 kV accelerating voltage). The high-resolution transmission electron microscopy (HRTEM, HITACHI, Tokyo, Japan) and scanning transmission electron microscopy (STEM) images was taken by using a Thermo Fischer Talos F200X microscope with mapping and

Preparation of Fe/ZIF-67
Fe/ZIF-67 catalysts were synthesized by traditional method of incipient wetness impregnation (IWI). Commonly, 0.2 g iron acetylacetonate was dissolved in 1.8 mL methanol. Subsequently, it was added to the surface of 1 g ZIF-67 sample with stirring. The obtained product was dried at 25 • C for 8 h and at 110 • C overnight, and the impregnation behavior was repeated five times. NaNO 3 aqueous solution (0.136 g of NaNO 3 dissolved in 1.5 g of water) was added to the sample obtained above, and the obtained product was dried at 25 • C for 8 h and dried at 110 • C overnight. The obtained sample was named Fe/ZIF-67.

Preparation of FeCo/NC
FeCo/NC was prepared by the pyrolysis of pre-synthesized Fe/ZIF-67. The sample was heated up to specific temperature in nitrogen atmosphere (40 mL min −1 ) with a heating speed of 3 • C/min and held for 150 min, then cooled down to 25 • C. The obtained product was named as FeCo/NC-T.

Synthesis of FeCo/Al 2 O 3
In the same way as above, an aqueous solution of 1.94 g of Fe (NO 3 ) 3 and 3.74 g of Co (NO 3 ) 2 was respectively mingled in deionized water, then it was added to Al 2 O 3 (1.5 g). Promoted with the same amount of sodium for comparison. Subsequently the product was heated up to 600 • C in N 2 atmosphere (40 mL min −1 ) with a heating speed of 3 • C/min and held for 2 h, then cooling down to 30 • C. The obtained sample was named as FeCo/Al 2 O 3 .

Catalyst Characterization
The element content of the FeCo/NC catalyst was analyzed by an inductively coupled plasma (ICP, Optima 2100DV, PE, MA, USA), and the FeCo/NC samples need to be digested with a microwave before the test. X-ray diffraction (XRD, Bruker, Karlsruhe, Germany) of the samples were tested with a D8-Focus diffractometer at the scan speed of 8 • C /min from 5 to 80 • C by using Cu-Kα (40 kV, 200 mA) irradiation. Scanning electron microscopy (SEM, HITACHI, Tokyo, Japan) was tested on a Regulus 8100 (30 kV accelerating voltage). The high-resolution transmission electron microscopy (HRTEM, HITACHI, Tokyo, Japan) and scanning transmission electron microscopy (STEM) images was taken by using a Thermo Fischer Talos F200X microscope with mapping and energy dispersive spectroscopic (EDS, TFS, MA, USA) analysis. In order to eliminate the variables that may exist in the discussion, all the pyrolyzed samples were pretreated with H 2 before the test.
Thermogravimetric (TG, TA, DE, USA) analysis was taken by using a Q500A equipment. The result was obtained with a ramping speed of 8 • C/min from 20 to 780 • C under a N 2 atmosphere. ASAP-2020 physical adsorption instrument was used to test the catalyst pore structure and specific surface area.
The degree of graphitization of carbon support after pyrolysis was tested by Raman with an excitation source of 532 nm.
Use X-ray photoelectron spectroscopy (XPS, TFS, MA, USA) to identify the surface properties of FeCo/NC samples with a monochromatic Al Ka radiation on a Thermo Scientific K-Alpha+ spectrometer. The curves were corrected according to carbon 1s spectra at 284.8 eV.
The phases and contents of iron species were identified by Mössbauer spectra (MS-500, Wissel, Germany) at room temperature. α-Fe absorber and 57 Co(Rh) source (25 mCi) was used. The curves fittings were measured by Lorentzian multiplet analysis. CO 2 temperature programmed desorption (CO 2 -TPD) was taken by using an AMI-300 apparatus. The type and amount of alkalinity on the catalyst surface and the CO 2 adsorption capacity were tested by the change of CO 2 concentration with the peak temperature. Firstly, a catalyst of 60 mg was pretreated under 10% H 2 /Ar (25 mL·min −1 ) at 400 • C for 60 min. Subsequently, the temperature was cooled down to 40 • C under Ar atmosphere and stay for 20 min. The test was taken by injecting continuous pulses of CO 2 into the He flow. The result was collected from 50 to 350 • C at a ramping rate of 8 • C·min −1 in a He flow. H 2 temperature programmed reduction (H 2 -TPR) was taken by using an AutoChem II 2920 apparatus to get the best pretreatment temperature. The detailed description is listed in supporting.
Karl Fisher titration method was used to analyze the water content of liquid products on an 870 KF Titrino plus instrument

Catalytic Performances
Typically, a fixed-bed reactor with a 10 mm inner diameter was taken to test the CO 2 hydrogenation performance. 0.5 g sample (20-30 mesh) mixed with 3.5 g quarta sands (20-30 mesh) was loaded. Before the test, the sample was pretreated in H 2 (400 • C, 40 mL min −1 ) for 2 h. Subsequently, the reactor was cooled down to the 50 • C and the mixed gas with a H 2 /CO 2 /N 2 ratio of 69/23/8 (N 2 used as an internal standard) was led in the reaction still under 2.0 MPa and 350 • C, the space velocity = 6240 mL g −1 h −1 .
The reaction result was evaluated by operating two gas chromatographs. CO 2 , CO, N 2 , and CH 4 were studied by using a GC Micro 490 with a TCD detector. While hydrocarbons and oxygenates were researched by using a GC 2010 with an FID detector. CO 2 conversion was evaluated by carbon atom balance according to the Equation: CO 2 conversion = CO 2 in − CO 2 out CO 2 in × 100% CO 2 in and CO 2 out mean the molar fraction of CO 2 at the inlet and outlet of the reactor. CO and different hydrocarbon selectivity were given as follows: CO selectivity = CO out CO 2 in − CO 2 out × 100%

Conclusions
ZIF-67 was pyrolyzed at different temperatures to synthesize a carbon-encapsulated Fe-Co bimetallic catalyst, and the surface properties were promoted by sodium. We used different characterization methods to characterize the catalyst and evaluate the CO 2 hydrogenation activity, then further clarified the synergy of bimetals and the effect of pyrolysis temperatures on the samples. Compared with the traditional oxide support, the formation of the NC layer can effectively anchor and disperse the metal nanoparticles, promoting the reduction and carburization of the metal during the reaction, thereby forming a higher olefin selectivity. On the other hand, different pyrolysis temperatures had a profound effect on the degree of carbonization of the support, metal loading, particle size, and degree of carburization during the reaction. When the pyrolysis temperature was above 500 • C, the ZIF-67 was completely pyrolyzed, and the metal nanoparticles were well embedded in the generated carbon support. With the raise of the pyrolysis temperature, the carburization of the metal during the reaction was facilitated. The Fe 3 O 4 content of the iron phase during the reaction decreased from 44% (400 • C) to 27% (700 • C). Relatively, the Fe 5 C 2 content increased from 17% (400 • C) to 55% (700 • C). Meanwhile, part of the metal cobalt was converted into Co 2 C, which also promoted the conversion of CO to light olefin in the FTS reaction. When the pyrolysis temperature reached 700 • C, the active sites exposed decreased due to sintering. Therefore FeCo/NC-600 exhibited the highest selectivity for low olefins (27.05%). These results are important for the subsequent further adjustment of the catalyst structure and metal composition, as well as the understanding of strategies for the control of active phase transition products of CO 2 hydrogenation. Such catalysts are currently being studied in more detail.