Introduction

Excessive emission of CO2 by the use of fossil fuels gives rise to various environmental problems, including global warming, glacier melting, climate change, and ocean acidification [1,2,3]. However, CO2 can be regarded as an abundant, clean, nontoxic renewable carbon resource. Therefore, converting CO2 into valuable chemicals not only mitigates environmental problems but also reduces dependence on fossil energy [4, 5]. Direct hydrogenation is considered a feasible and promising way for CO2 utilization. Diverse products, such as methanol, syngas, olefins, liquid fuels, and higher alcohols, can be obtained using different catalysts. Fe-based catalysts are a great candidate if the target product is hydrocarbons. Generally, this process involves two steps, the reverse water–gas shift reaction (RWGS) over Fe3O4 and the Fischer–Tropsch synthesis reaction over iron carbides (FeCx) [6, 7].

To improve the catalytic properties of Fe-based catalysts, promoters are usually introduced to tune the physical and electronic structure of Fe species. Transition metals, such as Mn, Cu, Co, and Zn, have been extensively studied [8,9,10,11]. Among them, Zn exhibits versatile promotion effects, such as improving Fe dispersion, catalyzing the RWGS reaction, and tailoring product distribution [12,13,14]. Ma et al. [15] prepared Zn- and Na-co-modified Fe-based catalysts by coprecipitation and observed that Zn acting as a structure modifier considerably reduced the particle size of Fe species. Zhang et al. [16] introduced a Zn promoter to Fe-based FTS catalysts by microwave–hydrothermal method, which elevated both light olefin selectivity and catalyst stability. This was ascribed to the improved dispersion of Fe phases and decreased carbon deposition. Han’s group [17] revealed that ZnO stabilized the active phase (FeCx) in a bimetallic catalyst, thereby suppressing catalyst deactivation. Additionally, Wang et al. [13] observed that in the process of CO2 hydrogenation to olefins, the main role of ZnO was to promote the RWGS reaction. Similarly, Cai et al. [18] revealed that ZnO promoted the RWGS reaction by increasing CO2 adsorption. Furthermore, Zhang et al. [19] discovered that enhanced interaction between Fe and Zn altered the chemical environment of Fe species. The formation of the ZnFe2O4 solid solution structure inhibited the hydrogenation of olefin intermediates, thereby enhancing the selectivity of lower olefins. Gao et al. [20] revealed that the FeAlOx/Fe5C2 structure suppressed C–C coupling and promoted secondary olefin hydrogenation on the Fe5C2 surface. Following this, the extra addition of Zn would produce a synergistic effect between Zn and Al, which alleviated the undesirably strong interaction between Fe5C2 and spinel phases, finally improving the CO2 conversion. Han’s group [21] also observed that adding ZnO to Fe-based catalysts contributed to the generation of iron carbide, which enhanced the activity of CO2 hydrogenation to hydrocarbons. Additionally, the decoration of Na into Fe catalysts mainly altered the olefin formation.

The promotion effect of bimetallic catalysts has been discussed from different perspectives. The interaction and intimacy of promoters and active metals are critical in deciding catalytic properties. This study investigates the proximity effect of Fe–Zn bimetallic catalysts on activity and product distribution. Fe2O3 and ZnO nanoparticles with uniform size are prepared by the thermal decomposition method, and the distance between Fe and Zn is regulated by changing the loading sequences or mixing methods. Combining transmission electron microscopy (TEM), X-ray powder diffraction (XRD), X-ray photoelectron spectroscopy (XPS), H2 temperature-programmed reduction (H2-TPR), Mössbauer spectroscopy, etc., the promotion effect on the evolution of the Fe phase and catalytic properties is detailed.

Experimental Section

Chemicals

All chemicals were commercially purchased and directly used, including ferric chloride (FeCl3, Aladdin, 98.0%), zinc chloride (ZnCl2, Aladdin, ≥ 98.0%), zinc nitrate hexahydrate (Zn(NO3)2·6H2O, Aladdin, 99%), sodium oleate (C18H33NaO2, YuanLi Chemical Reagent Co., LTD, AR), oleic acid (Aladdin, > 90.0%), 1-octadecene (Aladdin, > 90.0%), and ethanol and hexane (YuanLi Chemical Reagent Co., LTD, AR). AlOOH·nH2O was purchased from Nankai University Catalyst Co. Deionized water was used for catalyst preparation.

Catalyst Preparation

Synthesis of Fe Nanoparticles and Fe/Al2O3 Catalysts

First, Fe nanoparticles were prepared by the thermal decomposition method [22,23,24,25]. Specifically, 6.5 g of ferric chloride and 36.5 g of sodium oleate were dissolved into a mixture containing 60 mL of distilled water, 80 mL of ethanol, and 140 mL of hexane. The homogeneous solution was heated at 60 °C for 4 h. The upper layer containing the iron oleate complex was then washed thrice with 100 mL of distilled water in a separatory funnel. Excess hexane was removed by reduced-pressure distillation to purify iron oleate. The obtained iron oleate was mixed with 3.0 g of oleic acid and 100 g of 1-octadecene and then heated at 320 °C for 30 min at a rate of 3.3 °C/min under nitrogen atmosphere. After cooling down, Fe nanoparticles were precipitated by adding 250 mL of ethanol to the solution and then separated by centrifugation. The products were washed several times with a mixture of hexane and ethanol and finally dispersed in 200 mL of hexane. The nanoparticles synthesized were uniform γ-Fe2O3 nanoparticles with a diameter of about 8.1 nm, as confirmed by XRD and TEM results (see Figs. S1 and S2). ICP results confirmed that no other elements (such as residual Na) were detected. α-Al2O3 supports were prepared by calcination of AlOOH·nH2O at 1200 °C for 4 h under air atmosphere. To load 10 wt. % Fe onto the α-Al2O3 supports, 1.0 g of α-Al2O3 was added into 100 mL of the above prepared hexane solution containing Fe nanoparticles, followed by stirring at room temperature for 24 h. Hexane was removed by rotary evaporation. The dried catalysts were then calcined at 350 °C for 4 h under air atmosphere, denoted as Fe/Al2O3.

Preparation of Catalysts with Different Fe–Zn Proximities

To obtain a series of Fe–Zn bimetallic catalysts with changeable Fe–Zn distance, Fe and Zn nanoparticles were separately prepared to load. ZnO nanoparticles were also fabricated by thermal decomposition as previously described, in which 6.5 g of ferric chloride was changed to 5.5 g of zinc chloride. The size of the ZnO nanoparticles obtained was approximately 17.6 nm, as confirmed by XRD and TEM results (Figs. S1 and S2). Additionally, no residual Na was detected in ZnO, as confirmed by ICP analysis. The prepared Fe and Zn nanoparticles were then loaded onto the α-Al2O3 supports by different methods. Specifically, the nanoparticles (nFe:nZn = 1) were together dispersed in hexane and mixed with α-Al2O3, followed by stirring at room temperature for 24 h. The suspension was dried by rotary evaporation, and it was then calcined at 350 °C for 4 h under static air, named Fe1Zn1-imp. Another, the prepared Fe and Zn nanoparticles were separately loaded on α-Al2O3 and dried and calcined under the same conditions as previously described. The obtained Fe/Al2O3 and Zn/Al2O3 (nFe:nZn = 1) were then physically mixed, named Fe1Zn1-mix. To further shorten the Fe–Zn distance, zinc nitrate solution was added into the prepared Fe nanoparticles in hexane solution and impregnated onto 1.0 g of α-Al2O3, named Fe1Zn1-ion. The theoretical Fe loading remained constant at about 10 wt. % for all catalysts. Figure 1 shows the schematic diagram of the fresh catalysts.

Fig. 1
figure 1

Schematic diagram of a Fe1Zn1-ion, b Fe1Zn1-imp, c Fe1Zn1-mix, and d Fe/Al2O3 catalysts

Full size image

Catalyst Characterizations

XRD patterns were measured by a SmartLab powder diffractometer with Cu Kα radiation (λ = 0.154 nm) at 60 kV and 220 mA. The samples were tested at a scanning speed of 8°/min in the 2θ range of 10–90°.

The specific surface area and pore structure properties of the catalysts were determined using a Micromeritics TriStar 3000 analyzer at − 196 °C. The specific surface area and pore size distributions of the catalysts were calculated by the Brunauer–Emmet–Teller and Barrett–Joyner–Halenda methods, respectively. Before testing, the catalysts were evacuated at 200 °C for 3 h to remove impurities and moisture adsorbed on the surface.

An inductively coupled plasma optical emission spectrometer (ICP-OES, Thermo iCAP 7400) was used to analyze the elemental content of Fe and Zn in the samples. Before measurements, the samples were dissolved in a solution of HCl, HNO3, and H3PO4 (volume ratio of 3: 1: 1) and treated at 200 °C for 4 h in a Teflon-lined stainless-steel autoclave.

The morphology and particle size of the samples were characterized by TEM spectroscopy (JEM-F200) at 200 kV. The particle size distribution was obtained by randomly measuring more than 300 nanoparticles in different regions. In the dark field, energy-dispersive X-ray (EDX)-mapping analysis was performed to obtain the elemental distribution.

H2-TPR was conducted on a Micromeritics AutoChem II 2920 with a thermal conductivity detector (TCD). 50 mg of the fresh catalyst was pretreated in Ar flow at 200 °C for 1 h. The gas was then switched to a mixture of H2 and Ar (10% H2/90% Ar). The sample was heated from 50 to 800 °C at a rate of 10 °C/min, during which the signal was recorded using the TCD.

XPS measurements were collected using a Thermo Fisher ESCALAB 250Xi XPS instrument with an Al Kα excitation source. The samples were tested in the vacuum with 5 × 10−8 Pa. The binding energy was calibrated by selecting as the reference the C 1 s photoelectron peak at 284.8 eV.

The graphitization degree of carbon species in spent catalysts was tested using a Raman spectrometer system (LabRAM HR Evolution, HORIBA Scientific) equipped with 532-nm laser excitation. The scanning range was 900–2000 cm−1, and the resolution was set to 1.838 cm−1.

57Fe Mössbauer spectroscopy was performed at -267 °C using a WSS-10 spectrometer controlled by Wissoft 2003 and a proportional counter (MFD-500AV, Japan) using 57Co(Rh) as a γ-ray radioactive source. The temperature was controlled using JANIS Cryostat SHI-850–5. The velocity was calibrated using a standard α-Fe foil at − 267 °C. The components of the Fe-based phase were identified by fitted hyperfine parameters, including isomer shift (IS), quadruple splitting (QS), and magnetic hyperfine field (H). The compositions of the Fe-based phase were obtained based on the areas of γ photon adsorption peaks, assuming that the Fe nuclei had the same adsorption probability.

CO2 temperature-programmed oxidation (CO2-TPO) was conducted on a Micromeritics AutoChem II 2920 instrument connected with a quadrupole mass spectrometry (MS) instrument. The catalysts were activated online with CO gas at 350 °C for 5 h. The activated samples were then purged with pure Ar for 1 h to remove excess CO. After the activated samples were cooled down to room temperature, the gas was switched to a 10% CO2/Ar mixture. The samples were heated from 50 to 800 °C at a rate of 10 °C/min, during which the signals were recorded using the MS instrument. H2 temperature-programmed desorption (H2-TPD) was conducted in the same facility. The catalysts were first reduced with 10% H2/Ar at 350 °C for 5 h and then purged by Ar flow. After the catalysts were cooled down to room temperature, a 10% H2/Ar gas mixture was introduced into the samples for 60 min. The MS signals were recorded upon heating the samples from 25 to 800 °C at a heating rate of 10 °C/min under Ar atmosphere.

Catalyst Tests

The catalytic performance in a fixed-bed reactor was observed. In a typical test, 200 mg of a catalyst was added to the tubular reactor (inner diameter: 8.0 mm). The catalyst was reduced in situ with H2 (30 mL/min) at 350 °C under 0.1 MPa for 5 h. The temperature was then reduced to 340 °C, and the pressure was adjusted to 2.0 MPa by introducing Ar into the reactor. The space velocity of the catalyst was 9000 mL/(g·h), and the volume ratio of the gas mixture was 67.5% H2/22.5% CO2/10% Ar.

The gaseous products and residual reactants were analyzed using an online gas chromatograph (Agilent 7890B). Specifically, permanent gas and CO2 were separated on an UltiMetal column and quantified using a TCD with Ar as an internal standard. C1–C6 hydrocarbons were separated on an HP-AL/S column and detected using a flame ionization detector.

CO2 conversion was calculated using the following equation: CO2 conv. = (FCO2,in − FCO2,out) / FCO2,in. CO selectivity was calculated by the following equation: CO selectivity = (Moles of CO2 converted to CO) / (Moles of input CO2 − Moles of output CO2). Selectivities to hydrocarbons were calculated on a carbon basis except for CO. Additionally, the activity of the catalysts was also expressed in terms of Fe time yield (FTY), meaning the moles of CO2 converted to hydrocarbons per gram of Fe per second:

$${\text{FTY = }}\frac{{{\text{Moles of CO}}_{{2}} {\text{ converted to hydrocarbons}}}}{{{\text{Mass of Fe atoms }} \times {\text{ Time}}}}$$

Results and Discussion

Properties of the as-Prepared Catalysts

The proximity between Fe2O3 and ZnO nanoparticles was examined by TEM equipped with EDX. As shown in Fig. 2a, Fe2O3 nanoparticles were uniformly deposited on the surface of α-Al2O3. In parallel, Fe2O3 and ZnO nanoparticles were observed to disperse over the α-Al2O3 surface with different distances for Fe–Zn bimetallic catalysts (see Fig. 2b–d). For Fe1Zn1-ion, the particle size was mainly centered at about 10 nm, which remained the same as the diameter of Fe2O3 nanoparticle precursors. However, for the Fe1Zn1-imp and Fe1Zn1-mix catalysts, Fe2O3 and ZnO nanoparticles were separated by a certain distance, whereas the ZnO nanocrystal was more adjacent to the Fe2O3 nanocrystal for Fe1Zn1-imp (see Fig. 2c–d). The proximity between Fe and Zn was further demonstrated by the EDX elemental mapping images of both fresh and spent catalysts (see Fig. 3 and Fig. S3). Both fresh and spent Fe1Zn1-ion samples exhibited a highly uniform and overlapping distribution of elements, indicating a close proximity between Fe and Zn. However, the elemental distribution of Fe and Zn in the Fe1Zn1-imp samples did not completely overlap (see Fig. 3b and Fig. S3b). The Fe and Zn elements became to enrich in certain regions. This suggested that a certain distance existed between Fe2O3 and ZnO particles in the catalyst. For the sample prepared by separate impregnation and physical mixing (Fe1Zn1-mix), Fe and Zn elements were obviously located at different regions (see Fig. 3c and Fig. S3c). Line scanning images further demonstrated that Fe and Zn elements clearly appeared in separate regions. This indicated that the distance between Fe2O3 and ZnO particles was farthest in the Fe1Zn1-mix catalyst. Notably, the proximity between Fe2O3 and ZnO for all the catalysts remained stable even after the reaction. Thus, the Fe–Zn bimetallic catalysts could be ranked on the basis of the relative distance between Fe2O3 and ZnO particles: Fe1Zn1-ion < Fe1Zn1-imp < Fe1Zn1-mix.

Fig.2
figure 2

TEM images of fresh catalysts: a Fe/Al2O3, b Fe1Zn1-ion, c Fe1Zn1-imp, and d Fe1Zn1-mix

Full size image
Fig.3
figure 3

EDX elemental mapping images of spent catalysts: a Fe1Zn1-ion, b Fe1Zn1-imp, and c Fe1Zn1-mix. d High-angle annular dark-field scanning transmission electron microscopy image and EDS line scanning of the spent Fe1Zn1-mix catalyst

Full size image

Fe-catalyzed CO2 hydrogenation included the RWGS and FTS reactions [25,26,27]. The latter is a structure-sensitive reaction in which the particle size of the Fe species is critical in catalytic performance [25]. As the series of Fe and Fe–Zn catalysts retained the same size of Fe nanoparticles derived from the thermal decomposition method, the impact of size effect on the performance could be eliminated. Additionally, the data presented in Table S1 showed that the specific surface areas of the catalysts were almost the same and that the bimetallic catalysts were composed of Fe (about 9 wt.%) and Zn (about 10 wt.%) in a molar ratio of about 1 (see Table S1). Therefore, the bimetallic Fe–Zn model catalysts have been successfully prepared for solely investigating the Fe–Zn proximity effect on CO2 hydrogenation.

The XRD patterns of fresh Fe/Al2O3, Fe1Zn1-ion, Fe1Zn1-imp, and Fe1Zn1-mix catalysts are shown in Fig. 4. The diffraction peaks at 30.3°, 35.7°, and 63.0° were assigned to γ-Fe2O3 (JCPDS 25-1402). The nature of the Fe species was further confirmed by the Fe 2p spectra (see Fig. S4). The satellite peak that appeared at ∼719.0 eV could be assigned to the Fe2O3 species [28]. Additionally, compared with Fe/Al2O3, the diffraction peaks of Fe1Zn1-imp and Fe1Zn1-mix at about 63° were shifted toward lower angles. This was ascribed to the stacking of the ZnO phase, which held a lower characteristic diffraction peak. However, for Fe1Zn1-ion, the shift became more apparent, indicating Zn was partially inserted into the lattice of γ-Fe2O3. Conversely, close contact between Zn and Fe was achieved for Fe1Zn1-ion.

Fig. 4
figure 4

XRD patterns of freshFe1Zn1-ion, Fe1Zn1-imp, and Fe1Zn1-mix catalysts

Full size image

H2-TPR was applied to analyze the interaction between Fe and Zn. As seen in Fig. 5a, the reduction temperature of Fe1Zn1-ion was higher than that of Fe1Zn1-imp and Fe1Zn1-mix. Additionally, Fe/Al2O3 was most prone to be reduced. This illustrates that Zn addition inhibits iron oxide reduction; the trend became increasingly obvious with closer Fe–Zn proximity. XRD patterns of the reduced catalysts are shown in Fig. 5b. The predominant ion phase of all the reduced catalysts was metallic Fe, which was evidenced by the intense diffraction peak located at 44.7°. Among the catalysts, the Fe/Al2O3 one exhibited the strongest diffraction peak of Fe, indicative of a high reduction degree under H2 treatment. Along with ZnO addition and further shortening Fe–Zn distance, the intensity of Fe diffraction peaks was gradually weakened. Noteworthily, the diffraction peak of FeO appeared on Fe1Zn1-ion. This indicated that the inhibition of Fe reduction would become prominent because of the closest proximity between Fe and Zn, in accordance with the H2-TPR results. Moreover, for reduced Fe–Zn bimetallic catalysts, no angle shift was observed compared with the characteristic peaks of metallic Fe-based catalysts, indicating the reduced catalysts did not contain the Fe–Zn alloy phase.

Fig. 5
figure 5

a H2-TPR profiles of Fe/Al2O3, Fe1Zn1-ion, Fe1Zn1-imp, and Fe1Zn1-mix catalysts; b XRD patterns of the reduced catalysts

Full size image

Catalytic Performance

To understand the effect of Fe–Zn proximity on CO2 hydrogenation performance, catalysts with different Fe–Zn distances were evaluated. As presented in Table 1, CO2 conversions on the Fe–Zn bimetallic catalysts were all higher than that on the Fe catalyst without a promoter due to the promotion effect of ZnO on the RWGS reaction [13]. By physically mixing ZnO/Al2O3 with Fe/Al2O3, the activity increased from 32.7% to 42.7%. Shortening the Fe–Zn distance (i.e., Fe1Zn1-imp) further improved the activity with almost identical CO selectivity. This demonstrated that a relatively closer distance between Fe and Zn contributed to enhancing the RWGS and FTS reactions. However, for the Fe1Zn1-ion sample, CO2 conversion turned to decrease, along with a remarkably increased CO selectivity. This indicated that the FTS reaction was obviously inhibited over Fe1Zn1-ion, then hindering the forward reaction of RWGS. The changes in activity and CO selectivity caused by controllable Fe–Zn distances revealed that suitable proximity can simultaneously facilitate the RWGS and FTS reactions.

Table 1 Catalytic performance of Fe1Zn1-ion, Fe1Zn1-imp, Fe1Zn1-mix, and Fe/Al2O3 catalysts
Full size table

The distribution of hydrocarbon products showed that the Fe–Zn bimetallic catalysts exhibited lower CH4 selectivity and higher C5+ selectivity than Fe/Al2O3. This indicated that the addition of ZnO promoted the C–C coupling process and shifted the product distribution toward hydrocarbons with a higher number of carbon atoms. Reportedly, the presence of ZnO increases the intensity of –CH3 or –CH2 groups, which are important intermediates for C–C coupling [29]. The Fe1Zn1-ion catalyst exhibited the highest C5+ selectivity, suggesting that a closer distance between Fe and Zn further promoted C–C coupling. H2-TPD profiles in Fig. S5 show that the series of Fe–Zn bimetallic catalysts had a lower temperature of H2 desorption in comparison with Fe/Al2O3, illustrating the weakened H2 adsorption caused by Zn addition. Consequently, the formation of CH4 and C2–4 hydrocarbons was suppressed. Within 60 h of the catalyst tests, Fe/Al2O3 exhibited a relatively apparent deactivation behavior (Fig. 6). However, the CO2 conversion on the stream became more stable over time upon introducing Zn as a promoter. Particularly for the Fe1Zn1-ion catalyst, the activity even became almost constant after 10 h. Therefore, with the decrease in distance between Fe and Zn, the stability was further improved.

Fig. 6
figure 6

CO2 conversion on time on stream for the Fe1Zn1-ion, Fe1Zn1-imp, Fe1Zn1-mix, and Fe/Al2O3 catalysts

Full size image

Structure–Performance Relationship

During the CO2 hydrogenation reaction, the reduced Fe catalyst undergoes carburization and oxidation. Finally, the Fe phases of the spent catalysts exist in the form of Fe3O4 and Fe5C2 (see Fig. 7), which are, respectively, regarded as the active phases for the RWGS and FTS reactions [6, 30, 31]. The Zn phases still existed in the form of ZnO. XPS was performed to determine the surface phase compositions of the spent catalysts (see Fig. 8a). Deconvolution of the Fe 2p spectra revealed two peaks appearing at approximately 707.5 and 710.5 eV, which could be, respectively, assigned to Fe5C2 and Fe3O4 [32,33,34,35]. Among all the spent catalysts, the Fe/Al2O3 catalyst exhibited the highest peak area ratio of Fe5C2/Fe3O4 (0.78), followed by the Fe1Zn1-mix (0.56) and Fe1Zn1-imp (0.45) catalysts. However, the Fe1Zn1-ion catalyst exhibited the lowest ratio (0.29). This means that the addition of Zn led to a decrease in the ratio of Fe5C2 to Fe3O4 and that as the Fe–Zn distance became smaller, more Fe3O4 formed during the reaction. Generally, Fe3O4 was the active phase for catalyzing the RWGS reaction, while Fe5C2 was responsible for CO hydrogenation. Figure 8b shows the relationship between the ratio of Fe5C2 and Fe3O4 and the reaction properties. FTY initially increased and then decreased as the ratio increased from 0.29 to 0.78. However, CO selectivity initially decreased and then remained almost unchanged with an increasing ratio. Considering the reversible characteristic of the RWGS reaction, expediting CO conversion would push the reaction equilibrium shift toward CO production, according to Le Chatelier’s Principle. Therefore, a balance should be established between RWGS and FTS performances. In this case, the Fe1Zn1-ion catalyst held a relatively higher Fe3O4 content but a lower Fe5C2 content; hence, the CO formed cannot be converted to hydrocarbons in time. The reaction equilibrium was the main limiting factor. For the Fe1Zn1-imp catalyst, a proper proportion of Fe3O4 and Fe5C2 enabled the timely conversion of CO intermediates generated by the RWGS reaction. Consequently, a superior activity was obtained on the Fe1Zn1-imp catalyst.

Fig. 7
figure 7

XRD patterns of the spent catalysts

Full size image
Fig. 8
figure 8

a XPS patterns of the spent catalysts and b FTY and CO selectivity as a function of the ratio of Fe5C2 and Fe3O4

Full size image

To further quantify the composition of the different Fe phases, the spent catalysts were characterized by Mössbauer spectroscopy (see Fig. 9a). The corresponding data are presented in Table 2. The patterns were fitted with 5 sextets. The 3 sextets with an H value of approximately 25 T, 22 T, and 13 T were attributed to χ-Fe5C2 and the 2 sextets with H of approximately 51 and 45 T corresponded to Fe3O4 [36]. The Fe/Al2O3 catalyst had the highest mass ratio of Fe5C2 to Fe3O4 in bulk phases, indicating that it was more prone to be carburized during the reaction. This is in line with the H2-TPR result (see Fig. 5a), in which Fe/Al2O3 was most easily reduced. After ZnO introduction, the content of Fe5C2 in bulk phases tended to decrease, demonstrating that the addition of ZnO inhibited the carburization of Fe. Additionally, as the Fe–Zn proximity increased, carburization of the Fe species was gradually suppressed, leading to a lower content of Fe5C2. To observe the oxidization degree of the fully carburized catalysts with different Fe–Zn distances by an oxidant (CO2), CO2-TPO of the CO-treated catalysts was conducted (see Fig. 9b). As shown in Fig. S6, all the CO-treated catalysts existed in the χ-Fe5C2 phase. Compared with the Fe/Al2O3 catalyst, the Fe–Zn bimetallic catalysts exhibited a lower oxidation temperature. This trend continued to become stronger with further shortening of the distance between the Fe and Zn. Therefore, adding a Zn promoter and modulating the Fe–Zn proximity affect the reduction of Fe oxides and oxidation of carbides, ultimately regulating the composition of Fe5C2 and Fe3O4.

Fig. 9
figure 9

a 57Fe Mössbauer spectra of the spent catalysts measured at − 267 °C and b CO2-TPO profiles of the activated catalysts

Full size image
Table 2 57Fe Mössbauer parameters of the spent catalysts
Full size table

Figure 10 shows the Raman spectra of the spent catalysts. All the spent catalysts exhibited two main bands at approximately 1340 and 1580 cm−1, which represented the D band (structural disorders) and G band (graphitized carbon), respectively [37,38,39]. The ratio of the D band to the G band (ID/IG) is typically used as an indicator to describe the graphitization degree. A higher ID/IG ratio indicates more defects and a higher disordered structure of carbon deposit on the surface. Compared with the Fe–Zn bimetallic catalysts, the Fe/Al2O3 catalyst held a lower ID/IG value, indicating that more graphite carbon deposits were formed during the reaction. Accordingly, a thick carbon layer could be seen in the TEM image of the spent Fe/Al2O3 catalyst (see Fig. S7). The graphitic carbon layer would block active phases and lead to catalyst deactivation, consistent with its poor stability (see Fig. 6). After the Zn introduction, the ID/IG value tended to increase, demonstrating that the presence of Zn inhibited the formation of graphite carbon. Noteworthily, the spent Fe1Zn1-ion catalyst had the highest ID/IG; additionally, the carbon layer was merely observed in the TEM images of the spent Fe1Zn1-ion catalyst. Hence, high proximity between Fe and Zn could further inhibit carbon deposition during the reaction and improve catalyst stability.

Fig. 10
figure 10

Raman spectra of the spent catalysts

Full size image

Conclusions

A series of Fe–Zn bimetallic catalysts with different Fe–Zn proximities were obtained by tuning the impregnation sequence or mixing method. Upon adding Zn and further shortening the Fe–Zn distance, the reduction of the Fe species exhibited a gradually suppressed tendency. The composition analysis of the Fe phases revealed that the interaction of Zn and Fe inhibited the carburization of the Fe species and promoted the oxidization of carbides during the reaction. A volcano-type relationship between the catalytic activity and the ratio of Fe5C2 and Fe3O4 was observed. Additionally, the Fe1Zn1-imp catalyst with a medium ratio of Fe5C2 and Fe3O4 content simultaneously contributed to the RWGS reaction and CO hydrogenation, thus facilitating the entire CO2 hydrogenation reaction. Meanwhile, adding Zn and further reducing the Fe–Zn distance promoted the C–C coupling process, resulting in higher C5+ selectivity. Compared with the Fe2O3/Al2O3 catalyst, the Fe-Zn bimetallic catalysts exhibited higher catalytic stability. Particularly, close proximity between Zn and Fe species was observed to inhibit carbon deposition. The design of this model catalyst provided a novel way of studying the proximity effects for bimetallic catalysts. Additionally, it enriched the understanding of the relationship between the composition and activity of active phases (Fe5C2 and Fe3O4) in Fe-catalyzed CO2 hydrogenation reactions.