Introduction

Rising emission of greenhouse gases and their impact on global climate increasingly attracts attention on the sustainable utilization of CO2 [1]. In this context, the hydrogenation of CO2 is widely considered to yield valuable products like methane [2, 3], higher hydrocarbons [4,5,6] and methanol [7, 8]. For the CO2 based synthesis of hydrocarbons Fe catalysts are known to be effective [6, 9, 10], whereas they undergo complex phase transformations and structural changes upon reaction [9, 11,12,13]. Catalytic activity is strongly influenced by reduction, oxidation and carburization of bulk Fe phases as well as by different types of carbon entities formed during hydrogenation [14]. The resulting dynamic behavior of the Fe catalysts hinders the unraveling of active sites and the understanding of activation and deactivation processes. These effects and processes occurring during CO2 methanation on Fe catalysts are highlighted in a recent review article [15]. Furthermore, important structural properties of the catalysts, such as particle and crystallite size often reveal broad distributions, which additionally impede the correlation of structure and activity, particularly for bare Fe catalysts [10]. Finally, iron oxides are also known to catalyze the water gas shift reaction [16].

For investigation of structure–activity relations, nanostructured core–shell catalysts offer unique properties making them suitable as model materials in catalysis research [17, 18]. For instance, changes of size and shape of the active core nanoparticles upon conversion can easily be studied by transmission electron microscopy (TEM), as the core–shell arrangement is well-defined and stable during reaction [19].

Fe based core–shell catalysts are already studied for a broad variety of important catalytic processes, e.g. esterification [20], ammonia decomposition [19], alcohol oxidation [21], olefin hydrogenation [22] and Fischer–Tropsch synthesis [23]. Contrary, no papers are reported so far for the investigation of Fe containing core–shell catalysts in the hydrogenation of CO2 to CH4 (\({\text{CO}}_{{2}} {\text{ } + \text{ 4H}}_{{2}} \rightleftharpoons {\text{CH}}_{{4}} {\text{ } + \text{ 2H}}_{{2}} {\text{O; }}\Delta {\text{H } = \text{ }} - {\text{ 165 kJ mol}}^{{ - 1}}\)). Additionally, the reverse water gas shift reaction is often reported as a competing reaction of methanation when using Fe catalysts (\({\text{CO}}_{{2}} {\text{ } + \text{ H}}_{{2}} \rightleftharpoons {\text{CO } + \text{ H}}_{{2}} {\text{O; }}\Delta {\text{H } = \text{ 41 kJ mol}}^{{ - 1}}\)) [11].

In this study, core–shell model catalysts with different sizes of nanoparticulate Fe cores embedded in a SiO2 shell were prepared and thoroughly characterized. Additionally, the catalysts were tested for the reaction of CO2 and H2 with particular focus on the selectivity of CH4 versus CO as well as the influence of Fe particle size on catalytic activity. For evaluation of structural changes upon the conversion and the effect of Fe core size on catalytic performance, fresh and spent samples were physical-chemically characterized by X-ray diffraction (XRD), 57Fe Mössbauer spectroscopy, transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS) as well as temperature-programmed hydrogenation (TPH). Bare γ-Fe2O3, which was recently reported as effective CO2 methanation catalyst, was taken as a reference sample [10].

Experimental

Preparation of the core–shell catalysts

Spherical silica encapsulated iron catalysts were prepared using an adapted reverse microemulsion method. For the oil phase 100 mL cyclohexane (VWR Chemicals, AnalaR NORMAPUR®) were mixed with 34.2 g of polyoxyethylene (10) cetyl ether (Brij® C10, Sigma Aldrich/Merck) in a round-bottom flask and heated to 50 °C. Subsequently, 5 mL of a 1 M aqueous FeCl3 solution (ACROS Organics, p.a.) were added to the mixture under continuous stirring. After 30 min of homogenization 3 mL of hydrazine hydrate (aqueous solution, 35 wt%, Sigma Aldrich/Merck) were added dropwise and slowly to the solution followed by continuously stirring for 1 h. The color of the solution changed from bright yellow to dark brown which indicates the formation of iron nanoparticles. To encapsulate the obtained iron oxide nanoparticles in silica 21.4 mL tetraethyl orthosilicate (TEOS, Merck, p.a.) was added to the suspension followed by 3 mL of a 5 M aqueous ammonia solution (Carl Roth, 30wt%, ROTIPURAN®). This mixture was stirred continuously for 2 h. Afterwards the microemulsion was destabilized by addition of 2-propanol (VWR Chemicals, technical) and centrifuged at 8000 rpm for 15 min. The obtained particles were washed several times in 2-propanol, dried over night at 80 °C and calcined at 420 °C for 4 h (ramp 2 K/min). With defined variation of the surfactant type and SiO2 precursor it is possible to control Fe particle sizes in a range from 3.5 to 8 nm.

Reaction conditions for catalytic tests

For catalytic tests 200 mg of as-prepared core–shell samples as well as the bare γ-Fe2O3 reference was used taking a grain size of 125 to 250 µm. Grains were made by pressing the powder samples at 0.5 MPa followed by crushing and sieving. Respective catalyst was fixed in a glass rube reactor (i.d. 8 mm) with quartz wool, while the temperature was monitored by two K-type thermocouples placed directly in front of and behind the fixed-bed. Prior to CO2 hydrogenation, the catalyst was activated in 20 vol% H2 in N2 with a total flow of 500 ml/min (STP) at 350 °C for 3 h. Subsequently, reaction temperature was adjusted in N2 flow and the hydrogenation was started at 1 bar and a stoichiometric molar H2/CO2 ratio of 4 with a CO2 proportion of 10 vol.%. N2 was taken as balance. The total flow was 500 ml/min (STP) resulting in a space velocity of ca. 52,000 h−1 (69.2 mmol(CO2)/mol(Fe)/s).

Catalyst characterization

Powder X-ray diffraction (XRD) was performed on a D8 Discover (Bruker-AXS) with Co Kα radiation (35 kV, 45 mA). The diffractograms were recorded in 2θ mode from 15° to 85° with a step width of 0.06° and a time of 4 s per step. DIFFRAC EVA software (Bruker-AXS) was used for assignment of the reflexes, while quantitative phase composition was evaluated by DIFFRAC TOPAS software (Bruker-AXS) employing standard fitting procedure according to the Rietveld refinement.

Investigation of Fe species as well as phase composition was done with 57Fe Mössbauer spectroscopy using a 100 mCi 57Co source embedded in a Rh matrix. The spectrometer (WissEl) was operated in constant acceleration mode at ambient conditions. Experimental Mössbauer spectra were fitted using least square procedure assuming Lorentzian peak shapes, while isomer shift, quadrupole splitting and hyperfine field are reported relative to α-Fe.

Bright field TEM micrographs were taken by a Zeiss TEM 109 equipped with a CCD camera. For sample preparation for these measurements, a droplet of ethanol (VWR 96% Ph. Eur.) solution containing the sample powder (ca. 0.5 mg/ml) was deposited on a carbonized Cu grid, followed by evaporation of the ethanol.

X-ray photoelectron spectroscopy (XPS) was performed for the core–shell catalysts using an Escalab Xi + (Thermo Scientific) with Avantage software for data acquisition and analysis. All analyses were carried out under high vacuum conditions (10–7 mbar) using monochromatic Al Kα radiation. Surface charge was compensated by an electron flood gun. High resolution spectra of Si2p, O1s, C1s and Fe2p were measured with a pass energy of 25 eV.

Results and discussion

Preparation of the spherical encapsulated Fe catalysts was done by microemulsion method as recently published [24] (detailed description in ESI) adjusting a Fe mass fraction of 15 wt% for all samples. The Fe particle size distribution of three different as-prepared core–shell catalysts was obtained from TEM micrographs (Fig. 1, Fig. S1) resulting in average core sizes of 4, 6 and 8 nm, which is expressed by the respective sample codes used within this paper, e.g. Fe(6)@SiO2 for average core size of 6 nm. The total particle size of all samples amounts to ca. 16 nm leading to a decreasing shell thickness with increase in core size.

Fig. 1
figure 1

TEM micrograph of Fe(8)@SiO2 (left) and Fe core particle size distribution of the core–shell catalysts (right) each based on analysis of more than 200 particles

Full size image

Respective XRD patterns reveal a broad reflex at ca. 27° (2θ) assigned to the amorphous SiO2 shell, while γ-Fe2O3 or Fe3O4 represent the major Fe phase (Fig. 2, left). Differentiation of both phases by XRD is rather difficult, since they imply very similar crystalline structures and lattice parameters. However, 57Fe Mössbauer spectra excluded Fe3O4, since only superparamagnetic high-spin Fe3+ species appeared. These entities are reflected by their isomer shift (δ = 0.29–0.35 mm s−1) as well as quadrupole splitting (ΔEQ = 0.51–0.68 mm s−1) (Fig. 2 right, Table S2) [25, 26]. Also, the absence of magnetic coupling, which would produce a sextet pattern, suggests Fe2O3 particles smaller than ca. 13 nm [27, 28]. Furthermore, the crystallite size of γ-Fe2O3 estimated from Scherrer equation based on the X-ray diffractograms amounts to 4, 5 and 6 nm for Fe(4)@SiO2, Fe(6)@SiO2 and Fe(8)@SiO2, respectively, showing a similar trend as compared to TEM.

Fig. 2
figure 2

XRD patterns (left) with reflexes assigned to γ-Fe2O3 (filled diamond) and SiO2 (open square) and 57Fe Mössbauer spectra (right) at room temperature of the as-prepared core–shell catalysts; (hkl) indices refer to γ-Fe2O3

Full size image

Exemplarily, Fig. 3 and Table S3 demonstrates the CO2 hydrogenation performance of Fe(8)@SiO2 referring to a gas hourly space velocity of 52,000 h−1, an operation pressure of 1 bar and a stoichiometric H2/CO2 ratio. The catalytic tests were performed after H2 activation and are related to steady-state conditions (detailed reaction conditions in ESI). In the reaction of CO2 with H2, CO and CH4 are the only carbon containing products detected in the gas phase. Catalytic performance is expressed in terms of reaction rate of CH4 and CO (molar amount of CH4 or CO formed per time and total Fe amount). For temperatures between 300 and 550 °C, CO represents the major product, while its formation rate continuously increases with rising temperature. This effect is ascribed to the endothermic reverse water gas shift reaction, which is thermodynamically favored at elevated temperatures. The relatively low CH4 selectivity (4% at 300 °C and decreasing with inclining temperatures) is attributed to the high space velocity (52,000 h−1) and atmospheric operation pressure. The rate of CH4 formation goes through a maximum at 400 °C associated with kinetic limitation at lower temperatures and increasingly accelerating backward reaction at higher temperatures. It has to be stated, that the present work aims to investigate the influence of particle size on the production of CO and CH4 without optimization of the CH4 yield. Thus, although relatively small total CO2 conversions below 6% are achieved, the reaction conditions are suitable for this fundamental approach. Moreover, atmospheric pressure conditions lead to suppression of higher hydrocarbon production thus admitting the methanation as model reaction for CO2 hydrogenation without chain propagation.

Fig. 3
figure 3

Steady-state formation rates of CH4 and CO taken after syngas exposure for 30 min for different temperatures in conversion of CO2 and H2 on Fe(8)@SiO2 catalyst; conditions: p = 1 bar, y(H2)/y(CO2) = 4, GHSV = 52,000 h−1

Full size image

The apparent activation energy of CO2 conversion on Fe(8)@SiO2 is estimated as 60 kJ/mol (Fig. S2), which underlines the absence of significant mass transfer limitations. Note that no significant CO2 consumption is found in hydrogenation on bare SiO2 clearly showing the catalytic effect of the Fe cores.

The effect of Fe core size on catalytic performance was investigated at 400 °C, since at this temperature Fe(8)@SiO2 reveals highest methanation efficiency (Fig. 3). As a result, Fig. 4 demonstrates inclining rate of CH4 and CO formation with increase in size of Fe particles encapsulated inside the SiO2 shell. For comparison, bare γ-Fe2O3 (crystallite size ca. 25 nm; XRD pattern in Fig. S4) shows about 10 times higher CH4 and 15 times higher CO production rate as referred to the best core–shell catalyst, i.e. Fe(8)@SiO2. While CO2 conversion is generally enhanced for larger Fe particles, the CH4 selectivity is nearly the same (ca. 3%) for the three investigated core–shell catalysts.

Fig. 4
figure 4

Steady-state formation rates of CH4 and CO in conversion of CO2 and H2 after 17 h at 400 °C on the core–shell catalysts with different Fe core sizes as well as bare γ-Fe2O3 as reference; conditions: p = 1 bar, y(H2)/y(CO2) = 4, GHSV = 52,000 h−1

Full size image

The TEM images of the samples taken after CO2 conversion show brighter regions in the inner part of the core–shell structures (Fig. S3). This effect suggests some shrinkage of the cores associated with reduction of the Fe oxide precursor to metallic Fe upon the H2 activation. This is in line with the higher density of α-Fe (ρ = 7.87 g/cm3) as compared to γ-Fe2O3 (ρ = 4.86 g/cm3) [29]. Moreover, it is assumed, that the volume reduction does not substantially change by carburization, which partially occurs during CO2 hydrogenation, e.g. ρ(Fe3C) = 7.69 g/cm3. The observed shrinkage also indicates the permeability of the SiO2 shell, which is crucial for the gas transport to the catalytically active Fe core. Additionally, the TEM images reveal an intact core–shell structure after hydrogenation without any thermal sintering of the Fe cores or appearance of possible coke deposits.

For detailed characterization of carbonaceous species formed during syngas reaction, TPH analyses were performed after 17 h time on stream referring to steady-state conditions (Fig. 5). The TPH profiles obtained for the three core–shell catalysts exhibit minor formation of CH4 in the temperature range from 280 to 400 °C, while a strong CH4 feature occurs between 420 and 700 °C. In the literature, specific hydrogenation regimes are roughly ascribed to different types of carbon species including adsorbed surface carbon (270–390 °C), polymeric surface carbon (420–460 °C), iron carbides (480–690 °C) and graphitic deposits (above 690 °C) [30]. Based on this assignment, the major CH4 formation is tentatively referred to carbidic carbon species originated from the carburization of Fe cores upon CO2 hydrogenation. Carburization is quantitatively checked by integrating the respective TPH signal (400–700 °C) revealing carbon amounts of 22, 35 and 40 µmol for the samples with an average Fe core size of 4, 6 and 8 nm, respectively. From these results it is deduced, that the carburization of the catalysts clearly depends on the Fe core size, i.e. the proportion of carbon increases with the size of the Fe particle. Additionally, the quantitative carbon analyses by TPH imply molar C/Fe ratios below 0.1 for the three catalysts confirming incomplete iron carbide formation. Assuming exclusive formation of Fe3C, which is known as preferred carbide phase in hydrogenation of CO2 under the established conditions [10], the carburization degree of Fe(8)@SiO2 is about 20% only.

Fig. 5
figure 5

TPH patterns and respective total carbon amounts of the different core–shell catalyst after CO2 hydrogenation for 17 h at 400 °C and 1 bar; Fe amount is 535 µmol for all samples

Full size image

57Fe Mössbauer spectra of the spent core–shell catalysts (Fig. S5, Table S4) comprise two duplets, respectively, whereas no iron carbide sextets appear. However, iron carbide particles smaller than ca. 10 nm are known to form typical superparamagnetic duplets with isomer shift and quadrupole splitting close to the values of the pronounced duplet (δ = 0.37–0.51 mm s−1, ΔEQ = 0.72–0.98 mm s−1). In the spent samples, the latter features reveal fractions above 65%. Albeit, accurate quantification of superparamagnetic carbide structures is limited, since the Mössbauer parameters of carbidic and oxidic Fe3+ duplets are in a similar range and superimposition due to relatively broad line widths cannot be excluded [31]. It may be speculated, that superparamagnetic iron carbide species contribute to this intense duplet, since the quadrupole splitting of the spent samples tentatively increases as compared to the as-prepared catalysts. Also, the calculated Fe3C amount based on the TPH results generally fits to the relative areas of this duplet (68–87%) corroborating the incomplete carbide formation during CO2 hydrogenation. Moreover, an additional duplet is found for the spent samples, which is assigned to Fe2+ species according to the isomer shift (δ = 0.83–1.26 mm s−1) and quadrupole splitting (ΔEQ = 1.96–2.27 mm s−1). Furthermore, the proportion of Fe2+ increases with rising Fe core size from 13% (Fe(4)@SiO2) to 27% (Fe(6)@SiO2) and finally 32% (Fe(8)@SiO2).

Furthermore, no significant differences in the surface compositions of the as-prepared and spent catalysts are observed by XPS considering the quantities of Si, O, C and Fe (Fig. S6). However, molar C and Fe surface fractions are below 1% consequently leading to small intensities in respective spectra. Moreover, the formation of graphitic carbon species during conversion of CO2 and H2 is excluded due to the absence of the corresponding peak at ca. 284.5 eV [32]. This corroborates the TEM results, which do not provide indication for carbon deposits. In contrast to that, TPH study of bare γ-Fe2O3 performed after the CO2–H2 reaction shows drastic deposition of bulk carbon species (ca. 9100 µmol [10]), which reveals lower hydrogenation activity compared to Fe carbides. However, catalytic activity is higher as referred to the core–shell samples possibly associated with complete carburization of the bulk catalyst. Indeed, faster carburization of bulk relative to nanosized Fe catalysts was recently reported [33].

Using methanation as a key reaction for determination of CO2 hydrogenation activity, the CH4 formation rate of the core–shell catalysts is correlated with the amount of carbidic carbon derived from TPH. As a result, Fig. 6 suggests the importance of the abundance of Fe carbides for the methanation of CO2. This interpretation is in good agreement with the literature demonstrating carbidic structures as active species in COx hydrogenation on Fe catalysts [9, 10].

Fig. 6
figure 6

Correlation of carbon quantity obtained from TPH and CH4 formation rate of the core–shell catalysts with different Fe core sizes (the line is just a guide for the eyes)

Full size image

Conclusion

In summary, core–shell catalysts with defined Fe core sizes were investigated as model catalysts for the conversion of CO2 with H2. The study shows that reverse water gas shift reaction, i.e. production of CO, represents the major pathway with substantial formation of CH4, while higher hydrocarbons are not observed. Due to the synthesis of unique core–shell catalysts with narrow particle size distributions a valid correlation of catalytic performance and catalyst structure (particle size) is obtained. More detailed, rising Fe particle sizes from 4 to 8 nm leads to clear enhancement of the CO and CH4 formation. Additionally, growing Fe core size results in stronger carburization and active Fe carbide entities, which are responsible for the increasing CH4 production. Furthermore, the carburization behavior for the core–shell catalysts differ significantly from that of bulk Fe catalysts, as recently reported in our previous work, as well [34]. This observation may be caused by a confinement effect of the encapsulated Fe nanoparticles within a silica shell [35] and requires further research. Despite the focus on fundamental studies in this work, Fe based core–shell catalysts might be of technical relevance, since they do not imply any pronounced susceptibility towards coke deposition as reported for bulk Fe catalysts [10].