Gas-Permeable Iron-Doped Ceria Shell on Rh Nanoparticles with High Activity and Durability.

Strong metal-support interaction (SMSI) is a promising strategy to control the structure of the supported metal catalyst. Especially, encapsulating metal nanoparticles through SMSI can enhance resistance against sintering but typically blocks the access of reactants onto the metal surface. Here, we report gas-permeable shells formed on Rh nanoparticles with enhanced activity and durability for the surface reaction. First, Fe species were doped into ceria, enhancing the transfer of surface oxygen species. When Rh was deposited onto the Fe-doped ceria (FC) and reduced, a shell was formed on Rh nanoparticles. Diffuse reflectance infrared Fourier-transform spectroscopy (DRIFTS) results show that the shell is formed upon reduction and removed upon oxidation reversibly. CO adsorption on the Rh surface through the shell was confirmed by cryo-DRIFTS. The reverse water gas shift (RWGS) reaction (CO2 + H2 → CO + H2O) occurred on the encapsulated Rh nanoparticles effectively with selective CO formation, whereas bare Rh nanoparticles deposited on ceria produced methane as well. The CO adsorption became much weaker on the encapsulated Rh nanoparticles, and H2-spillover occurred more on the FC, resulting in high activity for RWGS. The exposed Rh nanoparticles deposited on ceria presented degradation at 400 °C after 150 h of RWGS, whereas the encapsulated Rh nanoparticles showed no degradation with superior durability. Enhancing surface oxygen transfer can be an efficient way to form gas-permeable overlayers on metal nanoparticles with high activity and durability.

Introduction

Metal nanoparticles deposited on oxide supports are a classic type of heterogeneous catalysts, which play a key role in various surface reactions.[1,2] The size, shape, and composition of nanoparticles were controlled to enhance the activity and selectivity.[3] The interaction between the nanoparticles and the support also has received much attention. Especially, strong metal–support interaction (SMSI) has been frequently credited as a reason for improved catalytic property. SMSI can tune the electronic structure of surface-active sites, adjusting the adsorption strength of reaction intermediates.[4,5] SMSI often results in the oxide shell encapsulating metal nanoparticles, with enhanced durability by suppressing the aggregation of metal nanoparticles.[6−8]

The oxide shell is typically formed after the reduction process, through which sub-stoichiometric surface oxygen species are formed, sequentially migrating onto metal nanoparticles, finally forming thin overlayers.[9] However, the shell is often crystalline and impermeable to gas reactants, eventually resulting in poor catalytic activity once the overlayer is fully formed.[10,11] Because the shell is formed under the reducing environment at high temperatures, sometimes the metal nanoparticles are sintered into larger particles before the shell is formed.[12] Additionally, the oxide shell is usually degraded easily under humid conditions.[13,14] Therefore, it is desired for the oxide shell to have the following features; (1) the shell should be permeable for gaseous reactants and products to move through, (2) the shell should be formed without aggregation maintaining the small metal nanoparticle size, and (3) the shell should be durable even under humid conditions.

The formation of the oxide shell on metal nanoparticles, which partly satisfies the above conditions, has been recently reported. Christopher et al. reported CO2-induced overlayers on Rh/TiO2 catalysts formed via formate intermediates, which was gas-permeable for CO2 reduction.[15] Xiao et al. deposited Au nanoparticles on Mg–Al-layered double hydroxide and then calcined it under N2 flow, forming Mg–Al oxide overlayers partially encapsulating Au nanoparticles.[16] They also formed TiO overlayers on Au/TiO2 catalysts with enhanced activity and durability for CO oxidation.[17] They further found that CO2 can induce SMSI forming thin overlayers on Au/MgO catalysts by the reversible reaction, MgO + CO2 ↔ MgCO3, which was also gas-permeable for CO oxidation.[18]

Herein, we facilitated the surface redox property on the CeO2 support by doping Fe. The facile transfer of surface oxygen species enabled the formation of thin overlayers on Rh nanoparticles with a size of 2–4 nm. The overlayer was not formed on Rh/ceria prepared without Fe-doping. The reverse water gas shift (RWGS) reaction (CO2 + H2 → CO + H2O) was performed efficiently with high activity and durability, despite water formation, with selective CO production.

Experimental Detail

Catalyst Syntheses and Reactions

Fe-doped CeO2, which is denoted as xFC (x: 5, 10, and 20 showing atomic % of Fe) below, was synthesized with a precipitation method. Ce(NO3)3·6H2O (99.99%, Kanto Chemical) (1.5 g) and a proper amount of Fe(NO3)3·9H2O (≥98%, Sigma-Aldrich) for 5, 10, and 20 atomic % Fe were dissolved in 35 mL of deionized water. The solution was stirred for 30 min, then ammonia water (30%, Duksan) was added until the pH reached 8.5. The solution was stirred for 4 h, until its color turned into light brown. This brown slurry was filtered, washed with deionized water, and dried at 80 °C. The obtained cake was ground to fine powder and annealed at 500 °C for 4 h. The CeO2 support was also synthesized using the same protocol except adding Fe precursor. Various metals of Rh, Pt, Ru, Ir, and Pd were deposited on the FC and CeO2 supports using a wet impregnation method. The support powder (300 mg) was dispersed in 5 mL of deionized water. Metal precursors [RhCl3·xH2O (99.98%, Sigma-Aldrich), H2PtCl6·xH2O (≥99.9%, Sigma-Aldrich), RuCl3·xH2O (99.98%, Sigma-Aldrich), IrCl4·xH2O (≥99.9%, Sigma-Aldrich), and PdCl2 (≥99.9%, Sigma-Aldrich)] were dissolved in 1 mL of deionized water with a proper amount for their target weight %. The metal precursor solution was introduced dropwise into the support solution at 85 °C with stirring. The final solution was dried completely and calcined under air at 500 °C for 2 h. The RWGS reaction was carried out in a U-shaped quartz glass fixed-bed reactor. The catalyst (50 mg) was loaded onto the reactor. Typically, a pretreatment was performed at 500 °C for 2 h using 10% H2 flow (balance N2). All the reactions were performed at atmospheric pressure. The gas products were analyzed using an online gas chromatography instrument (Younglin GC 6500) equipped with a packed bed Carboxen 1000 column (75035, SUPELCO, 15 ft × 1/8 in. × 2.1 mm) and a thermal conductivity detector (TCD).

Characterizations

The crystalline structure was investigated with a powder X-ray diffractometer (XRD; SmartLab, RIGAKU). The peak shift in the XRD pattern by Fe-doping were monitored by Pohang light source (PLS) 8D beamline. X-ray absorption spectroscopy (XAS) of Rh K edge was investigated using 10C beamline of the PLS. The Rh K spectra were obtained in a fluorescence mode. The spectrum for Rh foil was also collected to calibrate each sample. The XAS data were processed with ATHENA software. In situ near ambient pressure X-ray photoelectron spectroscopy (NAP-XPS) was performed using 8A2 KASI-PAL AP-XPS beamline of the PLS. The X-ray energy source was 650 eV. The sample was mixed with carbon powder and then loaded onto Si wafer and heated under 1 Torr H2. High angle annular dark field-scanning transmission electron microscopy (HAADF-STEM) images and energy-dispersive X-ray spectroscopy (EDS) mapping images were obtained by a Titan-cubed G2 60-300 (FEI), and bright-field TEM images were collected by Tecnai G2 F30 S-Twin (FEI); both equipment were operated with an accelerating voltage of 300 kV. Raman spectra were collected using the FT-Raman spectrometer (Bruker). Brunauer–Emmett–Teller (BET) surface area analysis was performed using a Tristar II 3020 (Micromeritics).

Temperature-programmed reduction using H2 (H2-TPR) was performed on a BELCAT II (MicrotracBEL) equipped with a TCD. Before the measurement, the catalyst was pretreated with Ar at 150 °C for 1 h and cooled to room temperature. The metal dispersion was measured by a pulsed CO chemisorption using the same equipment. The catalyst (50 mg) was treated with a following sequence: (1) 5% H2/Ar at 500 °C (2 h); (2) 5% O2/He at various temperatures of 300, 400, 500, 600, or 700 °C (1 h); (3) He (1 h); (4) CO2 (30 min); (5) He (20 min); and (6) 5% H2/Ar (5 min). CO2 was injected to form surface carbonate suppressing CO spillover onto the CeO2 surface. After all these pretreatments, CO was pulse-injected onto the catalyst. Temperature-programmed desorption using CO at cryogenic temperature (cryo CO-TPD) was also performed with the same equipment. The catalyst was pretreated with Ar at 100 °C and cooled to −120 °C with liquid N2. 10% CO/He was injected for 30 min, and then, temperature was raised from −120 up to 800 °C.

Diffuse reflectance infrared Fourier-transform spectroscopy (DRIFTS; Nicolet iS-50, Thermo Scientific) was performed to observe CO adsorption, H2-spillover, and reaction intermediates. The catalyst (20 mg) was mixed with 180 mg of KBr powder, ground well, loaded in a sample cup, and set into a DRIFTS cell with a KBr window. 5% CO/Ar was flowed at room temperature for 10 min, and DRIFT spectra were collected under evacuation. Cryo-DRIFTS was also performed with a ZnSe window by cooling to −150 °C by liquid N2. The thermodynamic equilibrium composition was calculated at various temperatures with HSC Chemistry software (Metso Outotec).

Results and Discussion

Fe-doped ceria, which is denoted as FC, was first prepared by co-precipitation, then Rh was deposited on the FC support. When 1 wt % Rh was deposited on the ceria support without Fe-doping then reduced (1Rh/CeO2), DRIFTS peaks were clearly observed upon CO adsorption, as shown in Figure a. However, when 5 atomic % of Fe was doped into ceria (1Rh/5FC), the CO adsorption peaks became much smaller. No CO adsorption peaks were observed when 10 or 20 atomic % of Fe was doped into ceria (1Rh/10FC or 1Rh/20FC). For the 1Rh/10FC catalyst, reduction and oxidation were performed repeatedly at 500 °C, then CO-DRIFT spectra were obtained (Figure b). No CO peak was observed after the first reduction, but the peaks appeared after subsequent oxidation. When this catalyst was reduced again, the peaks disappeared, then re-appeared after subsequent oxidation. The following reduction also made the CO peaks disappear. These results clearly show that overlayers were formed on Rh nanoparticles upon reduction, and the overlayers can be removed by oxidation (Figure c). The formation and removal of the shell were reversible.

Figure 1

(a) CO-DRIFTS results of 1Rh/CeO2, 1Rh/5FC, 1Rh/10FC, and 1Rh/20FC obtained after reduction. (b) CO-DRIFTS results after repetitive reduction with H2 and oxidation with O2 at 500 °C for 1Rh/10FC. (c) Scheme of a Rh nanoparticle encapsulated with the FC shell. Red, beige, and brown atoms denote Rh, Ce, and Fe, respectively. (d) EXAFS results of the Rh K edge. Each DRIFT spectrum was collected after 10 min of evacuation.

Extended X-ray absorption fine structure (EXAFS) and X-ray absorption near-edge structure (XANES) results of the Rh catalysts are shown in Figures d and S1, respectively.The 1Rh/CeO2 catalyst showed an Rh–O peak at 1.6 Å and an Rh–Rh peak at 2.4 Å, but a new peak appeared at 2.0 Å when the FC support was used in EXAFS. This peak became larger as the Fe content increased. This peak probably results from the presence of Fe or support O in the first coordination shell of Rh. Similar behaviors, showing an additional EXAFS peak in the intermediate range, were reported previously for Pt/TiO2 or RhFe/TiO2, when there was SMSI between metal (Pt or Rh) and the support (TiO2).[19−21] Rh on FC supports had more metallic state than Rh on ceria supports with lower white line intensity in XANES.

The removal of the shell in 1Rh/10FC upon oxidation was tested by CO-DRIFTS at various temperatures of 300–700 °C, as shown in Figure S2. The area of CO peaks was the maximum at 500 °C and then decreased at higher temperatures. CO chemisorption was also performed using the method of Takeguchi et al.,[22] and the size of Rh nanoparticles was estimated from the CO uptake (Table S1). When 1Rh/10FC was reduced at 500 °C, no CO chemisorption was obtained, being consistent with CO-DRIFTS result. As the oxidation temperature increased, the CO uptake increased, indicating that more Rh sites are exposed. The CO uptake was the largest for the oxidation at 500 °C, then the CO uptake decreased at higher temperatures. When the shell was removed the most, the Rh size was estimated as 2.4 nm. The Rh nanoparticles seemed to aggregate upon the oxidation at higher temperatures.

Figures and S3 show TEM images of 1Rh/10FC catalysts. The amorphous shell surrounding Rh nanoparticle is observed in Figures a and S3b–d. HAADF-STEM and the corresponding EDS mapping images confirm the existence of Rh nanoparticles with sizes of <5 nm. In the EDS mapping images, Fe and Ce signals were collected as well at the areas where Rh signals were obtained, supporting the overlayer formation.

Figure 2

(a) Bright-field TEM image, (b) HAADF-STEM image, and the (c,d) corresponding EDS mapping images of the 1Rh/10FC catalyst.

The FC supports were characterized in more detail. XRD patterns in Figure a show that Fe addition onto ceria did not produce additional phase, except 20FC showing very small peaks of Fe2O3. High-resolution XRD patterns in Figure b clearly show the shifts of the CeO2(111) peak upon Fe addition, indicating that Fe was successfully doped into the ceria lattice. HAADF-STEM and the corresponding EDS mapping images of 10FC support show that Fe was finely distributed on the ceria (Figure S4). The BET surface area of the support was 84.2, 102.9, 94.8, and 81.1 m2/g for CeO2, 5FC, 10FC, and 20FC, respectively (Table S2). The surface area of the Rh catalyst was 47.7, 67.1, 69.3, and 64.3 m2/g for 1Rh/CeO2, 1Rh/5FC, 1Rh/10FC, and 1Rh/20FC, respectively.

Figure 3

(a) XRD pattern, (b) high-resolution XRD pattern obtained from synchrotron X-ray source, (c) H2-TPR, and (d) Raman spectroscopy results for the supports of CeO2, 5FC, 10FC, and 20FC.

H2-TPR results in Figure c show that a surface reduction peak appeared at lower temperature upon Fe-doping. While bare CeO2 had a broad peak at 535 °C, the reduction peak was shown at 430, 410, and 405 °C for 5FC, 10FC, and 20FC, respectively. The 20FC support also showed a small peak at 500 °C, which results from Fe2O3 species. The surface oxygen can be removed more easily upon Fe-doping. In Raman spectroscopy results (Figure d), the peak at 465 cm–1 indicates an octahedrally symmetrical vibration mode of ceria (F2g), and the peak at 600 cm–1 indicates a defect-induced mode(D). The ratio of peak intensities of these peaks (ID/IF) was estimated to compare the amount of oxygen vacancy.[23] The ratio increased significantly after Fe-doping with a maximum value in 10FC. Fe-doping onto the ceria support enhanced the oxygen vacancy formation with more facile removal of surface oxygen species. As the oxidation state of Fe is between +2 and +3 and that of Ce is between +3 and +4, Fe-doping into CeO2 enriched the surface defects, leading to the improved oxygen transfer. This highly defective oxide enabled the formation of gas-permeable overlayers on Rh nanoparticles.

CO adsorption onto the Rh catalysts was investigated using DRIFTS. When the spectrum was obtained with flowing CO, the gaseous CO peaks appeared largely, but the peaks soon disappeared after evacuation in 10FC, as shown in Figure a. The CO molecules physically adsorbed on the surface was easily detached. The 1Rh/CeO2 catalyst, however, presented shoulders at 2090, 2015, and 1880 cm–1 with flowing CO (denoted as dotted boxes), and these peaks remained after evacuation, as shown in an inset of Figure b, which is typically observed in Rh metal nanoparticles. These shoulders indicate the CO adsorption on the Rh surface. In the case of 1Rh/10FC catalyst, interestingly, these shoulders were observed under CO flow, but they disappeared after evacuation (Figure c). The CO could be adsorbed on the Rh surface, but they were detached easily due to weak CO adsorption. In order to confirm the CO adsorption onto the Rh surface, cryo-DRIFTS were performed at −150 °C (Figure d). The 10FC support did not show the CO peak adsorbed at Rh sites, but 1Rh/CeO2 and 1Rh/10FC clearly showed the CO peak at 2063 cm–1, confirming the CO adsorption onto Rh sites. Surely, CO can pass through the overlayer, adsorbing on the Rh sites. CO adsorption on the Rh catalysts was further investigated with cryo CO-TPD (Figure S5). CO was adsorbed at −120 °C and then desorbed with increasing temperature. For both 1Rh/10FC and 1Rh/CeO2, CO desorption started at very low temperature. In the case of 1Rh/10FC, the last CO desorption peak appeared at 14 °C, but 1Rh/CeO2 had an additional peak at 67 °C. At room temperature, most CO was desorbed on 1Rh/10FC, whereas CO surely remained on 1Rh/CeO2.

Figure 4

DRIFTS results collected under CO flow and after evacuation for 10 min for (a) 10FC, (b) 1Rh/CeO2, and (c) 1Rh/10FC. (d) Cryo-DRIFTS results for 10FC, 1Rh/CeO2, and 1Rh/10FC obtained upon CO adsorption at −150 °C and subsequent evacuation.

The thermochemical reduction of CO2, which is typically called as RWGS reaction, was performed using the Rh/FC catalysts. As shown in Figure S6, thermodynamic equilibrium compositions were calculated by considering two cases; (1) CO only or (2) CO and methane together as products for various CO2/H2 ratios. Figure shows CO2 conversion and selectivity for RWGS with CO2/H2 = 1:1, which is the condition using H2 the least. As the Fe amount in the support increased, CO selectivity became higher. However, 1Rh/CeO2 presented poor CO selectivity of 55% at 350 °C, 1Rh/10FC and 1Rh/20FC presented CO selectivity of 100% throughout all the temperatures of 300–600 °C. Rh catalysts usually produce methane from RWGS reaction because strongly adsorbed CO on metallic Rh site is further hydrogenated into methane with low CO selectivity.[24] The weaker adsorption of CO on Rh through the overlayer enabled the selective CO production. The highest CO2 conversion, which is thermodynamically possible, could be achieved above 400 °C for 1Rh/10FC. Figure S7 shows CO2 conversion and selectivity for CeO2 and FC supports without Rh. The CO2 conversion was much poorer, but CO selectivity was 100%.

Figure 5

(a) CO2 conversion and (b) CO selectivity for RWGS reaction for 1Rh/CeO2, 1Rh/5FC, 1Rh/10FC, and 1Rh/20FC catalysts. Reactions were performed with a feed of CO2/H2 = 1:1 and 48,000 mL/gcat·h.

The effects of CO2/H2 ratio and weight hour space velocity (WHSV) were evaluated for the 1Rh/10FC catalyst, as shown in Figure S8. As H2 is provided more, methane is produced more, but CO selectivity was still above 90% even for CO2/H2 = 1:4. Even at very high WHSV, CO selectivity remained 100% for CO2/H2 = 1:1. The CO production rate was compared with the literature values in Table S3. CO could be produced with a very high production rate with 100% selectivity although the smallest ratio of CO2/H2 = 1:1 was used in a wide range of temperatures. Other types of metals, Pt, Ir, Ru, and Pd, were deposited on 10FC or bare CeO2, and their performances for RWGS reaction were tested and are shown in Figure S9. The CO2 conversion and CO selectivity were greatly enhanced when 10FC was used instead of CeO2. Fe-doping on CeO2 probably facilitated the overlayer formation on various types of precious metal nanoparticles.

The formation of overlayer on 1Rh/10FC at various conditions and its effect on RWGS reaction were examined and are shown in Figure S10. When the reduction was not performed, CO-DRIFTS peaks adsorbed on Rh nanoparticles were clearly observed. Christopher et al. and Xiao et al. reported that CO2 flow can induce the shell formation on Rh/TiO2 or Au/MgO, respectively,[15,18] so we also tested whether the shell is formed when 20% CO2 and 2% H2 were flown on the 1Rh/10FC catalyst at 250 °C for 4 h. Unlike the literature, the shell was not formed on 1Rh/10FC, instead, the CO-DRIFTS peaks were observed. When the reduction was performed at 200 °C, the CO-DRIFTS peaks were shown, and RWGS reaction activity was hardly changed from the case without reduction, indicating that the shell was not formed. When the reduction was performed at 500 or 700 °C, no CO-DRIFTS peaks were observed, indicating the shell formation. The 1Rh/10FC obtained after the reduction at 500 °C presented high CO2 conversion in RWGS reaction, but the 1Rh/10FC after the reduction at 700 °C showed much poorer CO2 conversion. The CO2 conversion at 400 °C was 21.6% after the reduction at 500 °C and 6.0% after the reduction at 700 °C (thermodynamically highest CO2 conversion was 22.5%). Figure S11 shows CO-DRIFTS results of the 1Rh/10FC after the reduction at 700 °C under CO flow, indicating that the overlayer formed by the reduction at 700 °C was not gas-permeable. The reduction at 500 °C was the best to form a gas-permeable overlayer on Rh nanoparticles. The overlayer formation was also investigated for Rh/Fe2O3. When 1 wt % Rh was deposited on Fe2O3 and reduced at 500 °C, the shell was observed as shown in Figure S12. However, its activity and selectivity for RWGS were very similar to bare Fe2O3, indicating that the shell was not gas-permeable. CO-DRIFT spectra also presented no CO peak even at −150 °C.

In situ DRIFTS results on 1Rh/10FC and 1Rh/CeO2 were obtained under gas flow of CO2/H2 = 1:3, as shown in Figure a,b. The formation of surface formates and CO intermediates, which are further reduced to methane,[25,26] was observed on the 1Rh/CeO2 catalyst, but these peaks were not observed on the 1Rh/10FC catalyst. Interestingly, the baseline lift was observed clearly in 1Rh/10FC, whereas the lift was not significant in 1Rh/CeO2. This lift was previously reported in TiO2-supported metal catalysts when H2-spillover occurs fast.[27−29] H2 molecules dissociate into H atoms on metal sites and spill-over to lattice O sites on the oxide surface. The electrons moved into shallow trap states and thermally excited into conduction bands forming continuous electronic states. This highly delocalized surface shows broad IR absorbance with baseline lift. Upon fast H2-spillover, the electrons transferred from metal to support, reducing π back-donation from a Rh atom to an adsorbed CO molecule, weakening the CO binding.[30] When H2-spillover does not occur fast, the electrons are accumulated on the Rh surface, enabling strong π back-donation, which strengths Rh–C bonding with weaker C–O bonding. H2-spillover occurs on 1Rh/10FC faster than 1Rh/CeO2, resulting in weaker CO adsorption.

Figure 6

In situ DRIFTS spectra when exposed to CO2 and H2 (CO2/H2 = 1:3) for (a) 1Rh/10FC and (b) 1Rh/CeO2. Injection of the feed gas started at 25 °C, and temperature was raised to 300 °C with a rate of 5 °C/min. The gas flow maintained at 300 °C for 10 min. Each spectrum was collected every minute, and the arrows indicate time flow. In situ Raman spectra when exposed to H2 (5% H2/Ar) for (c) 1Rh/10FC and (d) 1Rh/CeO2. Injection of the feed gas started at 25 °C, and temperature was raised to 500 °C with a rate of 20 °C/min.

The changes upon H2 flow was further monitored with in situ Raman spectroscopy. 5% H2/Ar was flown, and the temperature increased up to 500 °C. Figure c,d compares the ratio of the peak intensities (ID/IF) for 1Rh/10FC and 1Rh/CeO2. In Rh/10FC, the ratio of ID/IF increased more, indicating that surface oxygen defects truly increased more upon H2 exposure at elevated temperatures. Figure S13 also shows NAP-XPS Ce 3d and O 1s results of 1Rh/10FC and 1Rh/CeO2 in 1 Torr of H2. The fraction of Ce3+ and Oads, which typically represents surface oxygen vacancy, was estimated at different temperatures. 1Rh/10FC presented higher fractions of Ce3+ and Oads than 1Rh/CeO2, confirming that the surface oxygen species are activated more easily by H2-spillover.

The durability of 1Rh/10FC was tested and compared with 1Rh/CeO2 and is shown in Figure . When the RWGS reaction was performed at 400 °C for 150 h, the 1Rh/10FC showed no change in CO2 conversion (∼21%) and CO selectivity (100%), whereas CO2 conversion decreased and CO selectivity increased in the 1Rh/CeO2. HAADF-STEM images of 1Rh/10FC and 1Rh/CeO2 were obtained after the reaction and are shown in Figure S14. The distinct Rh nanoparticles could be easily found in 1Rh/CeO2, but distinguishing Rh nanoparticles was more difficult in 1Rh/10FC. The sintering of Rh nanoparticles was not severe, but the ceria domain became much larger in 1Rh/CeO2. XRD patterns obtained after the reaction (Figure S15) show that the ceria domain increased slightly from 9.5 to 10.7 nm in 1Rh/FC, whereas the ceria domain increased more from 12.2 to 17.2 nm in 1Rh/CeO2. When the RWGS reaction occurred repeatedly in 100–600 °C, the 1Rh/10FC presented no change, whereas the 1Rh/CeO2 presented a decrease in CO selectivity. Figure S16 shows the RWGS reaction result performed at 600 °C for 150 h on the 1Rh/10FC. At this condition, the equilibrium composition is estimated as 15% CO2, 15% H2, 10% CO, 10% H2O, and 50% N2. Although the overlayer formed by SMSI is often known to be susceptible to steam,[13] the reaction was performed stably for 150 h. Even after 150 h of RWGS at 600 °C, no CO-DRIFTS peaks were observed, as shown in Figure S17, indicating that the overlayer still existed after the long-term reaction.

Figure 7

Long-term durability test results of (a) 1Rh/10FC and (b) 1Rh/CeO2 obtained at 400 °C for 150 h. (c) Repeated reaction results of 1Rh/10FC and 1Rh/CeO2. The catalysts after each cycle were cooled to room temperature in pure N2 flow. Reactions were performed in a feed flow of CO2/H2 = 1:1 and 48,000 mL/gcat·h.

Conclusions

FC presented more facile oxygen transfer than bare ceria. When Rh was deposited and reduced, an overlayer was formed surrounding Rh nanoparticles on FC by SMSI (Rh/FC) while the Rh surface was exposed on bare ceria (Rh/CeO2). The overlayer was formed upon reduction and removed upon oxidation; this change was reversible. This overlayer was gas-permeable, as confirmed by DRIFTS. The RWGS reaction was performed on these Rh catalysts. Rh/CeO2 produced methane significantly, but Rh/FC produced CO with 100% selectivity, due to weak CO adsorption. Especially, 1Rh/10FC (1 wt % Rh, 10 atomic % Fe addition into ceria) could produce CO selectively with a very high production rate at all temperature ranges when the smallest amount of H2 was used with a feed flow of CO2/H2 = 1:1. This catalyst also showed high durability for 150 h of RWGS reaction at 600 °C or repeated reactions at 100–600 °C. The formation of gas-permeable shells on metal nanoparticles by enhanced oxygen transfer can be an efficient way to increase the activity and durability simultaneously for high-temperature gas-phase reactions.