FeO–CeO2 nanocomposites: an efficient and highly selective catalyst system for photothermal CO2 reduction to CO

Solar-driven catalysis is a promising strategy for transforming CO2 into fuels and valuable chemical feedstocks, with current research focusing primarily on increasing CO2 conversion efficiency and product selectivity. Herein, a series of FeO–CeO2 nanocomposite catalysts were successfully prepared by H2 reduction of Fe(OH)3-Ce(OH)3 precursors at temperatures (x) ranging from 200 to 600 °C (the obtained catalysts are denoted as FeCe-x). An FeCe-300 catalyst with an Fe:Ce molar ratio of 2:1 demonstrated outstanding performance for photothermal CO2 conversion to CO in the presence of H2 under Xe lamp irradiation (CO2 conversion, 43.63%; CO selectivity, 99.87%; CO production rate, 19.61 mmol h−1 gcat−1; stable operation over 50 h). Characterization studies using powder X-ray diffraction and high-resolution transmission electron microscopy determined that the active catalyst comprises FeO and CeO2 nanoparticles. The selectivity to CO of the FeCe-x catalysts decreased as the reduction temperature (x) increased in the range of 300–500 °C due to the appearance of metallic Fe0, which introduced an additional reaction pathway for the production of CH4. In situ diffuse reflectance infrared Fourier transform spectroscopy identified formate, bicarbonate and methanol as important reaction intermediates during light-driven CO2 hydrogenation over the FeCe-x catalysts, providing key mechanistic information needed to explain the product distributions of CO2 hydrogenation on the different catalysts. A nanomaterial that helps convert carbon dioxide to more useful chemicals has been developed by researchers in China. One potential method is to convert the carbon dioxide into carbon monoxide using a reaction known as reverse water-gas shift, and then use further reactions to convert this into fuel, or produce useful chemicals such as methanol or methane. This reaction normally requires high temperatures, and a catalyst is required to make the conversion efficient at lower, more practical temperatures. Tierui Zhang from the Technical Institute of Physics and Chemistry in Beijing and co-workers developed a nanocomposite based on iron and cerium with excellent performance in converting carbon dioxide into carbon monoxide with hydrogen only under light irradiation. This result indicates the potential of solar-driven catalysis for transforming carbon dioxide into fuels. A series of FeO-CeO2 nanocomposite catalysts (FeCe-x) were successfully fabricated by hydrogen reduction of hydroxide precursors at temperatures (x) between 200–600 °C. A FeCe-300 catalyst with a Fe:Ce ratio of 2-1 exhibited excellent performance for photothermal CO2 hydrogenation to CO (CO selectivity = 99.87%, CO production rate 19.61 mmol h−1 gcat−1, excellent stability). The FeO phase was effective in promoting the reverse water-gas shift (RWGS, CO2 + H2 → CO + H2O). Catalysts prepared at higher reduction temperatures contained both Fe0 and FeO, with the Fe0 catalyzing the Sabatier reaction (CO2 + 4H2 → CH4 + 2H2O) and thus lowering FeCe-x catalyst selectivity to CO.


Introduction
Modern societies are highly dependent on fossil fuel energy for electricity generation and transportation. Combustion of fossil fuels for energy releases CO 2 into the atmosphere, thereby causing global warming and a plethora of associated environmental problems [1][2][3] . To curb anthropogenic CO 2 emissions, researchers are now actively exploring different catalytic approaches for converting CO 2 into fuels and valuable commodity chemicals [4][5][6][7][8] . As a form of CO 2 sequestration, these catalytic approaches are particularly desirable, as they can generate economic value from CO 2 (thereby transforming CO 2 into a resource rather than an environmental problem requiring mitigation) 1,4,9 . However, transforming CO 2 into other useful compounds such as CO, CH 3 OH, HCOOH, CH 4 , and C 2+ hydrocarbons is challenging, requiring activation of strong C=O bonds 10 . Accordingly, the rates of CO 2 reduction via photocatalytic methods and electrocatalytic methods are currently too low to justify serious consideration from industry 11,12 . A further challenge with CO 2 reduction is achieving high selectivity for a specific product, which is highly desirable since it eliminates the need for separation of different reaction products. Most catalytic technologies for CO 2 reduction developed to date are based on hydrogenation approaches, which typically utilize the Sabatier reaction (CO 2 + 4H 2 → CH 4 + 2H 2 O) and/or the reverse water-gas shift (RWGS, CO 2 + H 2 → CO + H 2 O) 3,13 . While CO 2 can be converted into CH 4 and H 2 O via the Sabatier reaction, activation of CH 4 for further conversion into other higher-value chemicals requires a considerable energy input 14,15 ; thus, this approach for CO 2 conversion to value-added chemicals is not practical. In comparison, the RWGS reaction creates CO, which can be used to synthesize fuels such as methanol or various hydrocarbons via Fischer-Tropsch syntheses 6,16,17 . In addition, CO can be used as a reducing agent for smelting many metals, such as iron. Therefore, the efficient conversion of CO 2 into CO offers great economic and environmental benefits for future societies, motivating efforts to develop active and selective catalyst systems for this reaction.
Industrially, the RWGS reaction is a thermal catalytic process that utilizes transition metal-based catalysts, typically Fe, Co, Ni, Mn, Cu and combinations thereof, as the active phases [18][19][20][21] . Certain precious metal catalysts, such as Pt and Au, also show good activities for the RWGS reaction [22][23][24] , though they are not practical for industrial use due to their scarcity and susceptibility to poisoning. Similar to many other thermal catalytic processes, the RWGS reaction operates under relatively harsh reaction conditions (temperatures ranging from 400 to 700°C and high pressures) 25 , thus requiring significant energy input to achieve meaningful CO 2 conversion rates and yields of CO. Furthermore, at such high reaction temperatures, most RWGS catalysts show significant selectivity to CH 4 , thus yielding a product stream that contains CO and CH 4 , necessitating subsequent product separation 26 . However, lowering the reaction temperature generally improves the selectivity toward CO, though typically at the expense of a much lower CO 2 conversion 27 . Recently, researchers have begun to explore lightdriven photothermal processes as a means of driving the RWGS reaction under mild reaction conditions. In photothermal catalysis, strong absorption of solar photons by the catalyst results in catalyst heating to temperatures up to 450°C, thus allowing traditional thermal catalytic processes to occur at reasonable rates [28][29][30] . Recently, Ye and coworkers reported photothermal CO 2 hydrogenation to CH 4 with good efficiency and selectivity over a series of group VIII nanocatalysts 31 . Subsequently, Zhang and coworkers reported photothermal CO 2 hydrogenation to C 2+ hydrocarbons using an alumina-supported Co-Fe alloy nanoparticle catalyst 30 . Similarly, Garcia's group demonstrated that Co@C nanocomposites showed excellent photothermal catalytic performance for CO 2 hydrogenation to C 2+ hydrocarbons 32 . However, the photothermal RWGS reaction has received minimal attention to date, justifying a detailed investigation.
Herein, a series of novel FeO-CeO 2 nanocomposite catalysts were prepared by H 2 reduction of Fe(OH) 3 -Ce (OH) 3 precursors at temperatures between 200 and 600°C (the catalysts are denoted as FeCe-x, where x is the reduction temperature). The Fe:Ce molar ratio in the FeCe-x catalysts was varied from 1:2 to 3:1. The performance of the various FeCe-x catalysts for photothermal CO 2 hydrogenation under light excitation (Xe lamp) was then examined. An FeCe-300 catalyst with an Fe:Ce ratio of 2:1 exhibited very high selectivity for photothermal CO 2 hydrogenation to CO (Scheme 1). The catalyst, comprising FeO and CeO 2 nanoparticles, could be heated to 419°C under Xe lamp irradiation, resulting in a CO 2 conversion of 43.63%, a CO selectivity of 99.87%, and a CO production rate of 19.61 mmol h −1 g cat −1 . In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) revealed that the RWGS reaction (CO 2 + H 2 → CO + H 2 O) occurred on the FeO component, thus accounting for the high selectivity toward CO shown by the FeCe-300 catalyst. For FeCe-x catalysts synthesized at higher H 2 reduction temperatures, the appearance of metallic Fe 0 enabled the Sabatier reaction (CO 2 + 4H 2 → CH 4 + 2H 2 O) and thus decreased the photothermal CO 2 hydrogenation selectivity toward CO. To our knowledge, the CO yield of 19.61 mmol h −1 g cat −1 yield and CO selectivity of 99.87% obtained here for the FeCe-300 catalyst are the highest yet reported for solar-driven CO 2 hydrogenation. In addition, the FeCe-300 catalyst showed excellent stability during photothermal operation, with no loss in activity or change in CO selectivity detected over 50 h of testing. The results suggest that photothermal approaches can be very effective for solar-driven reduction of CO 2 to CO.

Catalyst preparation
The FeCe-x nanocomposite catalysts were prepared by H 2 reduction of Fe(OH) 3 -Ce(OH) 3 precursors at temperatures in the range of 200-600°C. Briefly, precursors were prepared via a simple precipitation reaction. Typically, 0.01 mol of Fe(NO 3 ) 3 ·9H 2 O, 0.005 mol of Ce (NO 3 ) 3 ·9H 2 O and 0.1 mol of urea were dissolved in 100 mL of deionized water. The resulting homogeneous solution was then refluxed at 110°C for 24 h under magnetic stirring. After cooling to room temperature, the product was collected by centrifugation and washed repeatedly with deionized water and finally dried at 60°C under vacuum. The product, a mixture of Fe(OH) 3 and Ce (OH) 3 , was then transferred to a tubular furnace and reduced at different temperatures between 200 and 600°C under a hydrogen atmosphere (H 2 /Ar = 10/90). A heating rate of 5°C min −1 was used in all experiments. The final products are denoted as FeCe-x, where x refers to the reduction temperature. For the FeCe-300 catalysts, the Fe: Ce molar was also varied to examine the effect of the Fe: Ce ratio on catalyst performance for photothermal CO 2 hydrogenation. Commercial CeO 2 with a particle size of 50 nm was purchased from a commercial supplier and used as a reference catalyst.

Characterization
Powder X-ray diffraction (XRD) patterns for the FeCe-x catalysts were collected on a Bruker D8 Focus X-ray diffractometer equipped with a Cu Kα radiation source (λ = 1.5405 Å) operating at 40 kV. X-ray photoelectron spectroscopy (XPS) data were obtained on a VGESCA-LABMKII X-ray photoelectron spectrometer using a nonmonochromatized Al-Kα X-ray source. Transmission electron microscopy (TEM) and high-resolution TEM (HRTEM) images were collected on a JEOL-2100F microscope operating at an accelerating voltage of 200 kV. The samples were dispersed on hydrophilic carbon films for the analyses. The instrument was also equipped for high-angle annular dark-field scanning TEM (HAADF-STEM) imaging and energy-dispersive X-ray (EDX) elemental mapping. Quantitative EDX analyses were performed on a Hitachi S-4800 scanning electron microscope. UV-vis diffuse reflectance spectra were collected on a Beijing PGENERAL TU-1901 spectrometer with an integrating sphere attachment. The spectra were recorded over the wavelength range of 200-2000 nm. DRIFTS data were recorded on a Vector 70v (Brucker) infrared spectrometer with a custom-built reaction cell (as described below).

CO 2 hydrogenation tests under light irradiation
The CO 2 hydrogenation tests were carried out in a flow reaction system consisting of a Teflon-lined stainless reaction chamber (50 cm −3 ) with a quartz window, a gas inlet and outlet, an electronic flowmeter to monitor the inlet gas flow rate and a thermocouple to measure the catalyst temperature ( Figure S1, Supporting Information). For each test, 50 mg of catalyst was uniformly dispersed on a light-permeable quartz fiber filter. Next, the reaction gas mixture (CO 2 /H 2 /Ar = 15/60/25, flow rate 15 mL min −1 ) was introduced into the reactor to achieve a pressure of 0.18 MPa. The system was then kept under these conditions for 1 h to eliminate any residual oxygen in the reactor. Then, the catalyst bed was irradiated with a 300 W Xe lamp (Beijing Perfect-light Co. Ltd, PLS-SXE-300 UV, light intensity 2.2 W cm −2 ) to initiate photothermal CO 2 conversion. After 1 h of irradiation, a gas sample was taken from the reactor by syringe and analyzed by a Shimadzu GC-2014 chromatograph (Shimadzu Co, Japan) equipped with a three-channel analysis system. The first channel analyzed hydrocarbons in an HP PLOT Al 2 O 3 column, with He as the carrier gas and a flame ionization detector (FID). The second channel analyzed CO 2 , N 2 , Ar, O 2 , CH 4 , and CO using a combination of micro packet Haysep Q, H-N and Molsieve 13× columns, employing He as the carrier gas and a thermal conductivity detector (TCD). The third channel analyzed H 2 using a micropacket HayeSep Q and Molsieve 5 Å column with N 2 as the carrier gas and a TCD detector. CO 2 conversion and product selectivity were determined by an internal standard method that used Ar. The CO yield was calculated from two standard curves with R 2 = 0.999.
Blank control experiments were also performed. Blank-1 tests were carried out under the same conditions as a normal test but without any catalyst. Blank-2 tests were carried out under the same conditions as a normal test but without the CO 2 /H 2 /Ar = 15/60/25 reaction gas. Both blank tests were performed after 1 h of illumination under a Xe lamp.

In situ DRIFTS studies of light-induced CO 2 hydrogenation
In situ DRIFTS was applied to study the surface intermediates formed during photothermal CO 2 hydrogenation on the different FeCe-x catalysts. A Vector 70v (Brucker) spectrophotometer equipped with a custombuilt in situ diffuse reflectance cell was used for the DRIFTS studies. A 300 W Xe lamp was used as the light source for photothermal experiments. The in situ cell was also equipped with an electric heating device to allow specific reaction temperatures to be reached (i.e., the same temperatures that were realized in the photothermal catalytic tests). For each test, 50 mg of catalyst was loaded into the reaction cell, and a background spectrum was collected under vacuum. Subsequently, a gas mixture containing CO 2 and H 2 (v/v = 1/4) was introduced to achieve an absolute pressure of 0.18 MPa. The reaction cell was maintained in this state for 30 min to allow adsorption-desorption equilibrium to be attained. Next, the 300 W Xe lamp and the electric heating device were applied to heat the catalyst and thus accurately simulate the temperature profiles realized in the photothermal catalytic tests. In situ DRIFTS spectra were collected every minute over a period of 1 h at 4 cm −1 resolution.

Composition and morphology of the FeCe-x catalysts
The FeCe-x nanocomposite catalysts were prepared by H 2 reduction of a mixed hydroxide precursor with an Fe:Ce atom ratio of 2:1 ( Figures S2 and S3, Supporting Information) at temperatures ranging from 200 to 600°C. The obtained FeCe-x catalysts (where x is the reduction temperature) were characterized in detail by XRD and XPS. When the reduction temperature was increased above 400°C, peaks due to metallic Fe 0 appeared (44.7 and 65.0°, JCPDS-65-4899), while the FeO peaks were attenuated, providing direct evidence for FeO reduction to Fe 0 . At all reduction temperatures (x), CeO 2 was the dominant cerium-containing phase. The Fe 2p XPS spectra in Fig. 1b (referenced against the C 1s signal of adventitious hydrocarbons at 284.7 eV) revealed changes similar to those seen by XRD with increasing H 2 reduction temperature. At reduction temperatures up to 400°C, only cationic Fe 2+ and Fe 3+ states with associated shake-up satellites were observed, consistent with the presence of FeO and Fe 2 O 3 , respectively. At reduction temperatures above 400°C, new peaks appeared at low binding energies relative to the Fe 2+ /Fe 3+ features, which can be readily ascribed to metallic Fe 0 (e.g., Fe 2p 3/2 = 706.7 eV), in perfect accord with the XRD results.
UV-vis absorption spectra were collected for the FeCe-x catalysts to gauge their light absorption properties. Strong absorption at UV, visible and NIR wavelengths was important to their function as photothermal catalysts for CO 2 hydrogenation (Fig. 1c). The FeCe precursor and FeCe-200 catalyst showed strong absorption below 500 nm due to the presence of Fe 3+ and Ce 4+ species. In contrast, the FeCe-x catalysts reduced at temperatures of 300°C or above absorbed strongly from 200 to 2000 nm, indicating that they possessed excellent light absorption properties (since CeO 2 only absorbs light weakly at visible wavelengths, the strong absorption by the FeCe-x catalysts with x = 300-600 can mainly be attributed to the presence of FeO and Fe 0 ). In addition, the light absorption characteristics of FeCe-300 catalysts with different Fe:Ce molar ratios were also studied ( Figure S4, Supporting Information). The FeCe-300 catalyst with the highest Fe: Ce molar ratio of 2:1 showed the strongest light absorption ability among the catalysts studied, again consistent with FeO rather than CeO 2 being the dominant light absorber in these nanocomposite catalysts. Next, the photothermal heating of the various FeCe-x catalysts under 300 W Xe lamp irradiation was examined. A thermocouple was used to monitor the surface temperature of the catalysts as a function of the irradiation time. As shown in Fig. 1d, the surface temperatures of FeCe-300, FeCe-400, FeCe-500 and FeCe-600 catalysts (Fe:Ce molar ratio 2:1) increased rapidly over the first 25 min of irradiation and then stabilized at 419, 425, 440, and 446°C, respectively. In comparison, the photothermal heating effects for the FeCe precursor and the FeCe-200 catalyst were obviously weaker, consistent with these two samples being less effective absorbers at UV-visible-NIR wavelengths than the other FeCe-x catalysts.

Photothermal CO 2 hydrogenation performance
The CO 2 hydrogenation activity of the FeCe-x catalysts and selected reference catalysts (FeCe precursor and commercial CeO 2 ) were evaluated in a flow system under light irradiation (0.18 MPa, 15 mL min −1 , CO 2 /H 2 /Ar = 15/60/25) without any external heating. The results of the CO 2 hydrogenation tests are summarized in Table 1. Blank experiments showed that no hydrocarbons or CO formed in the absence of a catalyst (Blank-1) or reactant gas (Blank-2). These experiments confirmed that the reaction device did not catalyze CO 2 hydrogenation and that the catalyst itself did not decompose to produce hydrocarbons or CO. As shown in Fig. 2b and Table 1, the FeCe hydroxide precursor and FeCe-200 showed minimal activity (conversion of CO 2 <15%) for CO 2 hydrogenation under Xe lamp irradiation, which can be rationalized by their relatively weak absorption properties in the 200-2000 nm range, indicating that photothermal heating was modest (i.e., not in the ideal range for CO 2 hydrogenation). In contrast, the FeCe-300 catalyst showed remarkable activity for CO 2 hydrogenation under Xe lamp irradiation, achieving a 44.33% CO 2 conversion with an excellent selectivity toward CO of 99.87% (Fig. 2a). Upon increasing the precursor reduction temperature to 400°C and then to 500°C, there was no significant change in the CO 2 conversion (42.05% and 40.89%, respectively), but the CO selectivity decreased progressively (95.65% and 92.01%, respectively). When the precursor reduction temperature was increased to 600°C, the CO 2 conversion dropped sharply to 22.75% (with CO selectivity 94.32%) due to sintering in the nanocomposite catalysts, which reduced the active area of the active Fe component. The decrease in the selectivity to CO from FeCe-300 to FeCe-500 was accompanied by a simultaneous increase in the selectivity to CH 4 , with the formation of CH 4 coinciding with the appearance of Fe 0 in the catalysts (see Fig. 1). The specific CO yield for the FeCe-300 catalyst was 19.61 mmol h −1 g cat −1 , higher than that obtained using any of the other FeCe-x catalysts. Based on this finding, we then examined the CO 2 hydrogenation performance of FeCe-300 catalysts prepared with different Fe:Ce molar ratios (Fig. 2c, Table 1). The Fe:Ce ratio of 2:1 delivered the best performance and highest CO 2 conversion among the FeCe-300 catalysts tested, reflecting its superior light absorption abilities ( Figure S4, Supporting Information). Interestingly, the selectivity to CO of the various FeCe- 300 catalysts was similar, consistent with all these catalysts containing the same active phase (FeO). On the basis of photothermal heating properties, CO 2 conversion, CO selectivity and CO yield, the FeCe-300 catalyst with an Fe:Ce molar ratio of 2:1 was identified as the best catalyst for CO 2 hydrogenation to CO among the various FeCe-x nanocomposites prepared in the current study. The results also demonstrate that the surface composition of the FeCe-x catalysts, especially the relative proportions of Fe 2 O 3 , FeO, and Fe 0 , strongly influences the product distribution for CO 2 hydrogenation. FeO is identified as the active site for the photothermal RWGS reaction and is thus responsible for the very high CO selectivity observed for the FeCe-300 catalyst. Metallic Fe 0 promotes photothermal CO 2 hydrogenation to CH 4 and H 2 O (i.e., the Sabatier reaction), thus explaining why the FeCe-400, FeCe-500, and FeCe-600 catalysts, which contain both FeO and Fe 0 , afforded lower selectivities to CO than the FeCe-300 catalyst (containing only FeO). Cycle tests of photothermal CO 2 hydrogenation were conducted on the FeCe-300 catalyst (Fe:Ce 2:1) to verify its stability. As shown in Fig. 2d, the FeCe-300 catalyst showed no significant change in CO 2 conversion (fluctuating between 40 and 52%), CO selectivity (97-99.9%) or CO yield (~20 mmol h −1 g cat −1 ) over 50 h of testing. The test confirmed that the FeCe-300 catalyst possessed excellent stability under the photothermal testing regime used here for CO 2 hydrogenation tests.

Morphological evolution of the FeCe-x catalysts
Since the activities and product selectivities of the FeCex catalysts for photothermal CO 2 hydrogenation were closely related to their surface Fe speciation, TEM and HRTEM were applied to further examine the morphologies and dispersion of FeO and Fe 0 in the catalysts. The FeCe-x catalysts all existed as nanoparticle clusters, with the mean size of the nanoparticles in the clusters increasing in size as the reduction temperature increased in the range 200-600°C ( Fig. 3a and Figure S6, Supporting Information). Obviously, increasing the reduction temperature caused nanoparticle sintering and grain growth, which was especially evident at the higher reduction temperature of 600°C, where large FeO/Fe 0 nanoparticles were observed to form (accounting for the Fig. 2 a Product selectivity toward CO and CH 4 for photothermal CO 2 hydrogenation over the FeCe hydroxide precursor and FeCe-x catalysts (Fe:Ce molar ratio = 2:1). b CO 2 conversion and product yields of CH 4 and CO for the FeCe hydroxide precursor and FeCe-x catalysts. c CO 2 conversion and product selectivity to CH 4 and CO for FeCe-300 catalysts with different Fe:Ce molar ratios. Data for commercial CeO 2 are also shown. d CO 2 conversion, product yields and product selectivities over 50 h of CO 2 hydrogenation tests. Reaction conditions: 50 mg catalyst, 300 W Xenon lamp, 5 cm irradiation distance, pressure of 0.18 MPa, reaction gas composition of CO 2 /H 2 /Ar = 15/60/25. loss of the sharp drop in CO 2 conversion observed for the FeCe-600 sample during photothermal CO 2 hydrogenation tests, see Table 1). The FeCe-300 catalyst with the best CO 2 hydrogenation performance was further characterized by HRTEM, HAADF-STEM and EDX elemental mapping (Fig. 3). HRTEM lattice fringes with spacings of 0.31 and 0.25 nm were observed, corresponding to the CeO 2 (111) and FeO(111) planes, respectively (Fig. 3b). EDX mapping established a very uniform distribution of Ce, Fe, O in the FeCe-300 catalyst (Fig. 3c, d), as was expected given the very small size of the FeO and CeO 2 crystallites formed at a reduction temperature of 300°C. The identification of discrete FeO and CeO 2 nanoparticles further confirmed that the sample was indeed a nanocomposite of FeO and CeO 2 (in agreement with XRD), as opposed to a ternary oxide phase. Since pure CeO 2 demonstrated negligible CO 2 hydrogenation activity under Xe lamp irradiation, it can be concluded that the function of CeO 2 in the FeCe-x catalysts was to (1) maintain a high dispersion of active FeO and Fe 0 components and (2) act as a cocatalyst to assist with the activation of CO 2 (e.g., under photothermal reaction conditions, CeO 2 will exist as CeO 2-x , thus activating CO 2 adsorption through surface V O ), as suggested by other reports [33][34][35] .

In situ DRIFTS studies of photothermal CO 2 hydrogenation
To rationalize the different selectivities of FeO and Fe 0 for photothermal CO 2 hydrogenation, in situ DRIFTS was used to identify key intermediates formed during photoinduced CO 2 hydrogenation on FeCe-300 (containing only FeO) and FeCe-500 (containing both FeO and Fe 0 ). For the FeCe-300 catalyst, peaks due to adsorbed CO 2  , 1338 cm −1 ) 36 appeared after reaction for 1 min (Fig. 4a). As the reaction proceeded, additional intermediate species, including bicarbonate (HCO 3 − , 1650 and 1423 cm −1 ) and CO 2 − (1681 and 1268 cm −1 ) 37,38 , were detected. Bicarbonate is a key intermediate in the RWGS reaction; thus, its detection here was not surprising since the FeCe-300 showed a very high selectivity for CO 2 hydrogenation to CO 38 . As the reaction continued, the characteristic peaks of adsorbed CO 2 and all the other intermediate species above decreased in intensity, indicating the consumption of these species as the reaction achieved a high steady-state rate. The development of a broad peak at 3250 cm −1 corresponding to adsorbed water as the reaction time approached 1 h was entirely consistent  with expectations for the RWGS (CO 2 + H 2 → CO + H 2 O). The DRIFTS data thus provide convincing evidence that the RWGS reaction was the dominant surface reaction on the FeCe-300 catalyst under Xe lamp irradiation, hence explaining the high selectivity of the FeCe-300 catalyst for photothermal CO 2 hydrogenation to CO (>99% selectivity). For the FeCe-500 catalyst containing FeO and Fe 0 , different intermediates emerged (Fig. 4b).
In the first few minutes of the reaction, three kinds of adsorbed CO 2 (1716, 1519, 1315, and 1338 cm −1 ), a bicarbonate species (1650 and 1423 cm −1 ), and CO 2 − (1681 and 1268 cm −1 ) were detected. In addition, bands associated with formate (1592 cm −1 ) and methanol (1050 cm −1 ) appeared, which are key intermediates of the Sabatier reaction (CO 2 + 4H 2 → CH 4 + 2H 2 O) 38 . With increasing reaction time, the peaks associated with adsorbed CO 2 , bicarbonate and CO 2 − lost intensity as was observed for FeCe-300, while the bands associated with formate and methanol were more persistent due to their continuous production and consumption. The results suggest that both the RWGS reaction and the Sabatier reaction occurred on the surface of the FeCe-500 catalyst, utilizing FeO and Fe 0 active sites, respectively. The product distributions of CO 2 hydrogenation over the FeCe-300 catalyst (RWGS pathway yielding only CO) and FeCe-500 catalyst (RWGS yielding CO and Sabatier pathway yielding CH 4 ) can be easily rationalized on this basis, highlighting the value of in situ DRIFTS for mechanistic assessment of working catalysts. To our knowledge, this is the one of the first occasions that DRIFTS has been successfully applied for the mechanistic investigation of a photothermal catalytic reaction.

Conclusion
In conclusion, a series of FeCe-x nanocomposite catalysts were successfully synthesized by a simple precipitation-reduction method. By varying the H 2 reduction temperature (x) in the range of 200-600°C, the valence state of Fe in the catalysts could be adjusted from Fe 2 O 3 to FeO and then Fe 0 at high temperatures (the Fecontaining phases were dispersed by CeO 2 at all temperatures). An FeCe-300 catalyst with an Fe:Ce molar ratio of 2:1 exhibited remarkable performance for photothermal CO 2 hydrogenation to CO (CO selectivity, 99.87%; CO production rate, 19.61 mmol h −1 g cat −1 ; and good cycling stability over 50 h). Detailed structural characterization studies and in situ DRIFTS measurements revealed that FeO in the FeCe-300 catalyst efficiently promoted the photothermal RWGS reaction, thus accounting for the very high selectivity to CO. FeCe-x catalysts prepared at higher reduction temperatures contained both FeO and Fe 0 , the latter catalyzing the Sabatier reaction to CH 4 and thus lowering the selectivity of CO 2 hydrogenation to CO. This work identifies the photothermal RWGS reaction as a new approach for harnessing solar energy to selectively produce CO from CO 2 under mild reaction conditions.