Introduction

Physicochemical interfaces between the metal and support are important in heterogeneous catalysis1. Construction of catalytically active interfaces between metals and supports has attracted extensive attentions. The metal support interaction leads to high dispersion2,3 or even over-encapsulation of active metals4. Therefore, it is crucial to design and control suitable metal-support interaction and thus create abundant and effective interfacial active sites for molecule transformation. In particular, the metal interaction with third row (MgO5, Al2O36 and SiO27) oxides and fourth row 3d metal oxides (TiO21,4,8,9,10,11,12, FeOx13,14,15, Co3O416,17) have been widely explored and discussed in terms of electron transfer process18,19, oxygen vacancies1,11,12,20 and surface wetting4. In comparison, fifth or sixth row oxides, such as rare earth oxides, are seldom used as catalysts support except for CeO22,3,20,21,22,23,24,25. These early transition rare earth cations are strong Lewis acid and have strong binding to Lewis base such as H2O and NH3. The metal-rare earth oxide interface should have very different properties to those third and fourth row metals. It is a complete surprise that they have been rarely investigated in the H2O or NH3 based reactions, such as water-gas shift (WGS) reaction11,21−23.

WGS is an irreplaceable reaction to produce H2 from CO and H2O, which is responsible for ammonia synthesis, hydrogenation reactions and hydrogen fuel cells26. On one hand, WGS is exothermic and limited by thermodynamic equilibrium at high-temperature (HT). On the other hand, H2O is a very weak oxidant for CO and has very sluggish kinetics at low-temperature (LT). In comparison, medium-temperature (MT) WGS can achieve rapid kinetics with high CO conversion equilibrium27,28. Therefore, developing effective and stable catalytic interfaces in the MT region is a practical route to replace the existing cascade HT-LT WGS process. This is usually achieved with Ni-based catalysts11,12,29,30,31,32,33,34,35,36, which is mostly active between 300 and 400 °C. Existing research focus on CeO2 based support30,31,32,33,34,35 mostly due to its abundant oxygen vacancy. However, at MT region, such oxygen vacancy may not be necessary, whereas the surface adsorption and activation of H2O is the key. This is an area that rare earth support, such as Y2O3, has good competitivity. In addition, Y2O3 also offers good thermal stability, high chemical durability and high mechanical strength, promoting the stability of the metal oxide interface.

Here we reported an in situ formed Ni-NiOx-Y2O3 interface during MT-WGS reaction, achieving 140.6 μmolCO gcat−1 s−1 of CO conversion at 300 °C, which was 4 times to the existing literature reports. Such unique interface was only formed with a 9:1 molar ratio of Ni to Y. The addition of only 10% Y2O3 played a key role in catalytic activity improvement and prevented the catalyst from sintering. In addition, through theoretical simulation we have demonstrated the electron transfer at the interface, which indicated the existence of the interaction intrinsically. The Ni-NiOx-Y2O3 interface was clearly characterized with aberration corrected electron microscopy and in situ Raman spectra. Combined with DFT calculation, H2O molecules dissociated effectively on the Ni-NiOx-Y2O3 interfacial site was proved. And the calculation of the whole reaction process also proved that the Ni-NiOx-Y2O3 interface was very efficient for WGS reaction. The excellent catalytic performance of Ni-NiOx-Y2O3 interface suggested the potential of studying and applying rare earth elements in catalysis, creating new metal-support interface with unconventional surface behaviors.

Results

Catalytic performance of the Ni9Y1Ox catalyst in WGS reaction

Series of NiaYbOx (a: b was the molar ratio of Ni to Y) samples were synthesized via the ultrasonic spray method21,37,38. The simultaneous decomposition of Ni and Y precursors in the same sprayed droplet ensured the formation of maximized interface between the two metals (Supplementary Fig. 1). The droplets were sent into a high temperature tube furnace under the purging of N2 to sufficiently undergo pyrolysis reaction, forming the NiaYbOx catalyst. Supplementary Figs. 2, 3a and Fig. 1a showed the activities of Ni9M1Ox (M = Y, Ce, Sm and Al) catalysts and the NiaYbOx catalysts with different Ni/Y ratios for the WGS reaction. The CO conversion of the Ni9Y1Ox catalyst was much higher than other catalysts, reaching more than 90% at 250 °C (pretreatment in the H2 atmosphere, the gas hourly space velocity (GHSV) was 42,000 cm3 gcat−1 h−1). This catalytic activity transferred the HT to MT WGS reaction for Ni-based catalyst. The different catalytic performance of NiaYbOx catalysts correlated well with their reducing ability (Supplementary Fig. 4). Catalyst with low reduction temperature and large H2 consumption in the H2 temperature-programmed reduction (H2-TPR) profile showed high CO conversion. The Ni9Y1Ox has a small reaction peak at 220 °C and a major one at 400 °C. In comparison, both NiO and Y2O3 have only one feature at 350 and 575 °C, respectively (Supplementary Fig. 5). The peak at 220 °C was therefore related to the Ni-Y2O3 interface. As shown in Fig. 1b, pretreatment in the H2 atmosphere shifted the CO conversion to lower temperature range for almost 150 °C, compared to the air-pretreated one. This result suggested that the oxidation state of the Ni played an important role. The formation of CH4 was only detected at 250 and 300 °C with less than 3% selectivity. Compared to other catalyst system in the literature11,15,21,30,35,39,40,41, the Ni9Y1Ox catalyst achieved the highest reaction rate and turnover frequency (TOF) of Ni-based catalysts, even higher than some Cu-based and noble metal catalysts between 250 and 350 °C (Fig. 1c, Supplementary Table 1). It also maintained solid time-on-stream stability at high temperature and with an ultra-high GHSV, in which the conversion decayed by only 2% in 100 h (Fig. 1d). Even in the stream also contained CO2 and H2 (5% CO, 23.3% H2O, 10% H2, 5% CO2, N2), Ni9Y1Ox could still maintain similar catalytic performance comparable to that of commercial Cu-Zn-Al (Supplementary Fig. 3b). The addition of Y2O3 significantly improved the activity and stability of the catalyst, compared to pure Ni sample (with NiO as precursor).

Fig. 1: Catalytic performance test of the NiaYbOx catalysts.
figure 1

a Temperature-dependent activities of the catalysts (Ni1Y9Ox, Ni9Y1Ox, pure Ni catalyst, the reaction gas content was 2%CO, 10%H2O, and the rest was N2); b temperature-dependent activity and selectivity of Ni9Y1Ox under different pretreatment conditions; c comparison of reaction rates with other Ni-based catalysts; d the long-term stability at a high GHSV (Ni9Y1Ox: 300 °C, GHSV = 420,000 cm3 gcat−1 h−1; pure Ni catalyst: 280 °C, GHSV = 168,000 cm3 gcat−1 h−1).

Full size image

Structural change of the Ni9Y1Ox catalyst

The specific surface areas (SBET) increased from 12.9 to 55.5 m2 g−1 when the Ni content was increased from 5.4% to 67.3% (Supplementary Table 2). Combined with the transmission electron microscope (TEM) and scanning electron microscope (SEM) characterizations (Supplementary Fig. 6), it could be observed that the Ni9Y1Ox catalyst presented spherical morphology (250–1500 nm). Aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images showed that the sample composed of uniformly dispersed nano-particles (Fig. 2). The inter-planar spacing of the fresh Ni9Y1Ox (Fig. 2d) sample was the same to the lattice fringe of NiO (111), which was consistent with the X-ray diffraction (XRD) results (Fig. 3a). In addition, no lattice fringe of Y2O3 was observed, indicating that Y2O3 was poorly crystallized due to the low calcination temperature. In comparison, Ni9Y1Ox-used (Fig. 2h, i) mainly showed the lattice fringe of metallic Ni, indicating the reduction of Ni2+ to Ni0 during the H2 pretreatment. In addition, lattice fringes of Y2O3 were found between Ni particles (Fig. 2h) indicated the partial crystallization of Y2O3. The structural changes of the catalyst were observed more intuitively based on the results of elemental mapping. As shown in Fig. 2e, j, the initial uniform distribution of Ni, Y and O in the fresh Ni9Y1Ox catalyst segregated after reaction. For Ni9Y1Ox-used, the Ni particles grew larger to about 8 nm, and the Y and O were mainly distributed in the surroundings of Ni particles. This structure prevented the further growth of Ni particles and provided abundant interfaces between Ni and Y2O3. Comparing with TEM images of pure Ni sample (Supplementary Fig. 7), it was clear that Y2O3 played the crucial role in forming interfaces and stabilizing the catalyst structure.

Fig. 2: The aberration-corrected HAADF-STEM images and elemental mappings.
figure 2

a−d Ni9Y1Ox-fresh; f−i Ni9Y1Ox-used; electron energy-loss spectroscopy (EELS) elemental mapping results of (e) Ni9Y1Ox-fresh or (j) Ni9Y1Ox-used.

Full size image
Fig. 3: Phase change and surface electronic structure.
figure 3

a XRD patterns of the fresh and used Ni9Y1Ox catalysts; b, c XPS results of the fresh and used Ni9Y1Ox catalysts; d, e quasi in situ XPS results of Ni9Y1Ox-used; f Y K edge X-ray absorption near edge spectra profiles and g Y K edge EXAFS of the fresh and used Ni9Y1Ox catalysts for experiment data (solid) and fitted lines (dotted).

Full size image

We further explored the chemical state of the catalyst surface by X-ray photoelectron spectroscopy (XPS). The fitting results of XPS data were integrated in Supplementary Table 3. From Ni 2p XPS spectra (Fig. 3b, d), it was seen that the peak of Ni9Y1Ox-fresh mainly corresponded to Ni2+ species, while that of Ni9Y1Ox-used sample mainly corresponded to Ni0 species12,30,31,32,33. This observation was consistent well with the observed lattice fringes (Fig. 2h). Trace amount of Ni2+ species were detected for Ni9Y1Ox-used, which was caused by the partial oxidation when exposed in air. From the result of quasi in situ XPS, only Ni0 species were detected on the surface (Fig. 3d). The Y 3d (Fig. 3c, e) spectra showed the Y3+ species for both samples42,43,44. The main difference was that Y3+ in Y2O3 was mainly detected from the fresh sample; but for the used samples, it was mainly Y carbonate. The catalyst surface change might result from the WGS reaction containing CO2, which was easily absorbed by Y2O3 on the surface. The bulk Y3+ was still in the form of Y2O3 as confirmed in XRD results (Fig. 3a). The local structure of Y3+ was further examined in the Y K edge X-ray absorption fine structure (XAFS). At the near edge, pure Y2O3 crystal showed peaks after the edge (Fig. 3f). Literature suggested that there were two Y3+ sites inside Y2O3 crystal45. Site 1 at 17054.4 eV had the same Y-O bonding length, whereas site 2 at 17063.0 eV had different Y-O lengths. Both fresh and used Ni9Y1Ox only had the site 1 feature, suggesting weak crystallinity and uniform Y-O bonding length. The near edge spectra also confirmed the Y3+ states for both fresh and used catalysts (Fig. 3f). The Y-O coordination numbers (C.N.) decreased from 7.671 ± 2.033 Å to 5.155 ± 0.392 Å after catalysis (Fig. 3g and Supplementary Table 4). This was associated with the formation of Y-Y coordination at 3.613 ± 0.098 Å. The change in the coordination situation was mainly related to the better crystallization of Y2O3 after the catalytic reaction, which was consistent with the XRD results (Fig. 3a). The combination of XRD, TEM, XPS and XAFS results suggested that Y3+ doped into the NiO lattice at the fresh state. During the reduction with H2, the NiO was reduced to Ni, leaving no space for Y3+. Therefore, Y3+ was partially converted into crystalline Y2O3 and formed the Ni-Y2O3 interface.

Then we explored the nature of the interfacial interactions. Firstly, the theoretical and actual hydrogen consumption of the catalysts was quantitatively calculated according to the H2-TPR result (Supplementary Fig. 4, Supplementary Table 2). The actual H2 consumption was higher than the theoretical value, which indicated that the Y2O3 species in the catalyst were also reduced due to the formation of the interface. Besides, DFT calculations demonstrated the nature of the interaction was an electrostatic attraction (Supplementary Fig. 8). The charge density calculations suggested that the combination Y3O4/NiOx/Ni{111} from Y3O4, NiOx and Ni{111} appeared noticeable electronic migration (NiOx gained 1.643 electrons from Y3O4 and 0.187 electron from Ni{111}).

The determination of Ni-NiOx-Y2O3 interfaces

The phase change of Ni9Y1Ox catalyst was further examined with in situ XRD under 5% H2/Ar atmosphere (Fig. 4a). It was found that NiO phase dominated from room temperature to 300 °C. The formation of metallic Ni appeared at 350 °C, which was in good agreement with the ex situ XRD results (Fig. 3a). This also explained the low activity of the air-activated catalyst below 350 °C because NiO was not reduced below this temperature under reaction condition. In addition, the crystalline dominate size of metallic Ni increased gradually with the temperature rise (Supplementary Table 5), 5.5 nm at 400 °C and 9.6 nm at 500 °C, according to the calculation using Scherrer formula.

Fig. 4: In situ characterizations to monitor the interfacial structure changing.
figure 4

a In situ XRD patterns in 5% H2/Ar for the Ni9Y1Ox catalyst; in situ Raman spectra under the WGS reaction conditions for (b, c) Ni9Y1Ox catalyst and (d) pure Ni sample.

Full size image

We further verified microscopic structure change of the interface during the WGS reaction by in situ Raman spectra. The vibration peak around ~505 cm−1 which corresponds to NiO46,47,48,49,50 was observed in the Ni9Y1Ox and pure NiO sample before reaction (Fig. 4b, d, Supplementary Fig. 9). The change of temperature and particle size would cause the shift of NiO vibration peak50. In addition, the change of peak position could also be considered as the formation of other Ni-O coordination forms, resulting in the formation of NiOx (0 < x < 1) species. Pretreatment at 400 °C under 5%H2/Ar led to disappearance of NiO vibration peak (Fig. 4b), indicating that NiO was fully reduced to metallic Ni. This observation was consistent with the XRD results (Fig. 4a). Very interestingly, the vibration peak of NiO appeared again in the WGS reaction atmosphere, suggesting the formation of Ni-O bond on the surface. Furthermore, by switching the feeding gas between CO and H2O (Fig. 4c), it was found that the vibration peak of NiO disappeared when CO was introduced at 400 °C, and then it appeared again after feeding H2O vapor. This indicated that the O atoms around Ni were almost fully consumed when the catalyst was treated at high temperature under CO, which was evidenced by the in situ XRD results (Supplementary Fig. 10). The formation of NiOx was not observed with pure Ni sample obtained from ultrasonic spray method (Fig. 4d), regardless of oxidation or reduction conditions. Therefore, without the Ni-Y2O3 interface, Ni alone could not directly react with H2O to formed NiOx. The in situ Raman and XRD experiments suggested that when exposed to H2O vapor, active oxygen species were formed over the Ni-Y2O3 interface and subsequently converted into Ni-NiO-Y2O3 interface as the active states for WGS reaction. Therefore, the roles of Y2O3 were: (I) restricting the growth of the Ni particles under the reduction conditions; (II) forming the Ni-NiOx-Y2O3 interfacial sites under the WGS condition.

Reaction pathway with Ni-NiOx-Y2O3 interface participation

At present, there were mainly two recognized mechanisms for the WGS reaction: redox mechanism and associative mechanism51,52,53,54. For the redox mechanism, CO reacted with the surface oxygen to form CO2 and vacancies, where H2O dissociated to form H2. Following this mechanism, the produced H2 could be detected during the H2O dissociated process. For the associative mechanism, CO and H2O adsorbed on the surface of the catalyst to form intermediate, and then the intermediate further decomposed to produce CO2 and H2 simultaneously. In that case, H2O tended to form surface hydroxyl species, which could react with CO20,21. Here, the H2O dissociation experiment was conducted (Supplementary Fig. 11). No H2 was produced in either the thermostatic test or the temperature-programmed heating test, ruling out the existence of redox mechanism. Furthermore, the temperature programmed surface reaction (TPSR) experiment was performed. The results showed a wide range of desorption signals of CO2 (Supplementary Fig. 12), which was caused by the oxidation of CO with surface O. The strong adsorption of CO was also found in the temperature-programmed desorption (TPD) using Ar as the carrier gas (Supplementary Fig. 13) and CO kinetic order (Supplementary Fig. 14c).

From the results of transient isothermal surface reaction (250 °C, Fig. 5a) after WGS reaction, it could be observed that the concentration ratio of CO2/H2 was ~2 when the reaction reached a stable state. It suggested that H2O might be dissociated to OH* first, and then OH* reacted with CO to generate CO2 and H*, following the associative mechanism20,21 according to 2 OH + 2 CO → H2 + 2 CO2. In addition, the cyclic CO temperature-programmed reduction (CO-TPR) was used to verify the production of hydroxyl36. As shown in Fig. 5b, the catalyst was treated in the WGS reaction atmosphere at room temperature between two CO-TPR experiments. For the CO-TPR-1, there were mainly three reduction peaks, among which the one at lower temperature (182 °C) was the reduction of small-size NiO, and the one at higher temperature (413 °C) should be the reduction of large-size NiO. The reduction peak at 253 °C was accompanied by the generation of H2, which resulted from the reaction between CO and hydroxyl species20,21,36. The formation of hydroxyl was caused by the dissociation of H2O in the WGS reaction before CO-TPR-1. For CO-TPR-2, two main reduction peaks could be observed, among which the higher temperature reduction peak was considered as the reduction of NiOx species formed in the reaction atmosphere. This reduction was similar to the H2-TPR result of the used sample (Supplementary Fig. 15a). Accordingly, the reduction peak at 252 °C was also accompanied by the generation of H2 signal, indicating the generation of surface hydroxyl was recyclable. The same result was also observed for the catalyst treated in the H2O/Ar atmosphere between CO-TPR-1 and CO-TPR-2 (Supplementary Fig. 15b). In order to ensure that pure Y2O3 did not contribute to the dissociation of H2O, we demonstrated this by activity test, in situ Raman and H2O dissociation experiments of pure Y2O3 (Supplementary Fig. 16). It was believed that the interface played an important role in the dissociation of H2O.

Fig. 5: The WGS mechanism study and DFT calculation of the Ni9Y1Ox catalyst.
figure 5

a Surface reaction on the Ni9Y1Ox catalyst at 250 °C under 2%CO/Ar atmosphere; b cyclic CO-TPR experiments for CO consumed and CO2 evolved against temperature for the Ni9Y1Ox catalyst (Between CO-TPR-1 and CO-TPR-2, the catalyst was treated in the WGS atmosphere at room temperature); c the ∆G of H2O adsorption and dissociation process at three Y atom sites in the Y3O4/NiOx/Ni{111} model, the reaction temperature was set to 300 °C and the partial pressure of water vapor was set to 10 kPa, i.e., the inlet pressure; d the simulated WGS reaction pathways (the reaction temperature: 300 °C, the partial pressure of CO, H2O, CO2 and H2: 1, 8000, 2000 and 2000 Pa, i.e., the outlet pressure of the reactor).

Full size image

We further calculated the activation of the H2O (300 °C) at different surface locations to verify the hypothesis of interfacial active sites. H2O adsorbed weakly on metallic Ni surface (Ni{111}). When the partial pressure of H2O was equal to 10 kPa, the ∆G of adsorbing on the top position and the bridge position were 0.516 eV and 0.758 eV, respectively. (Supplementary Fig. 17). However, the adsorption of H2O molecules in the Y3O4/NiOx/Ni{111} model was completely different. The ∆G of H2O adsorption at the three Y sites were preponderant (−0.428, −0.468 and 0.219 eV) because Y2O3 was a stronger Lewis acid than Ni (Fig. 5c). The coordination numbers of the Y atom at these three sites were 5, 4, and 4 in turn, and the Bader charges value were 1.84, 2.01 and 2.01, respectively. This was agreement with the EXAFS result of the used sample (the C.N. for Y-O was 5.155 ± 0.392). This result indicated that the lower coordination number (the C.N. of Y atom in Y2O3 was 6) and higher positive charge of Y atoms led to its stable binding with H2O molecules. Besides, two models (the Ni{111} and Y3O4/NiOx/Ni{111}) for H2O dissociation were also different. For the Ni{111}, although the adsorbed H2O molecules dissociated to *OH and *H were the process of decreased ∆G, the whole process of H2O adsorption and dissociation was still a non-spontaneous process (∆G > 0) (Supplementary Fig. 17). For the Y3O4/NiOx/Ni{111} model, although the H2O adsorption at site I (−0.428 eV), site II (−0.468 eV) and site III (0.219 eV) were various, the Gibbs free energies of the decomposition products (IMI-2, IMII-2 and IMIII-2) were significantly lower than those of the Y3O4/NiOx/Ni{111} plus vapor, i.e., the ∆G were down around 0.6 eV (Fig. 5c). In general, the Ni-NiOx-Y2O3 interface was much more favorable for the activation of H2O than pure Ni. The results of H2O reaction order measurements also supported the same conclusion (Supplementary Fig. 14a, b). The H2O reaction order of the pure Ni sample (0.57 at 270 °C) was obviously larger than that of the Ni9Y1Ox sample (0.32 at 270 °C), indicating that the latter was easier to adsorb and dissociate H2O. The whole reaction path of the WGS reaction was simulated and presented in Fig. 5d and Supplementary Table 7. H2O was adsorbed on the atom Y at the first step and CO was adsorbed on the Ni atom at third step. The rate-determining step was the fourth elementary reaction, i.e., *OH migration from atom Y to atom C, and the most probable cause was the bonding force of CO were too strong. As a whole, the WGS reaction was spontaneous because the sign of ∆G was negative (−1.476 eV at 300 °C, the partial pressure of each gas species was set to the value at the outlet of reactor, shown as Supplementary Table 6). In addition, the ∆G at the inlet of reactor were also provided and the larger spontaneous tendency was presented for the lower product concentrations (shown as Supplementary Fig. 18 and Supplementary Table 8).

From the conclusion of the reaction pathway study, we proposed the associative reaction mechanism involving the NiOx species at the interface (Supplementary Fig. 19). The adsorbed H2O molecules first dissociated at the interface NiOx site to generate two OH* molecules, and then the active hydroxyl species further reacted with the adsorbed CO to generate CO2 and H2. Therefore, the reactive Ni-NiOx-Y2O3 interfacial site enabled the catalyst to efficiently catalyze the WGS reaction and largely improved its reaction efficiency.

Discussion

In summary, the addition of Y2O3 was essential for the active Ni-NiOx-Y2O3 interface in WGS reaction. Compared to pure Ni sample, the presence of Y2O3 tremendously improved the catalytic activity and stability, enabling efficient WGS reaction at medium temperature range. Based on experimental investigation and DFT calculation, the Ni-NiOx-Y2O3 interface effectively dissociated H2O molecules and facilitated the reaction with associative mechanism. Y2O3 and other rare earth metal oxide system should play an important role in catalytic reactions, in particular those with Lewis base molecules such as H2O and NH3. This will be a golden era for the adventure of rare earth catalysis.

Methods

Synthesis of catalysts

In a typical synthesis of NiaYbOx, 4 mmol nitrates (99%, Tianjin Kermal Factory) were added to 60 mL anhydrous ethanol. The added mass of Ni(NO3)2·6H2O and Y(NO3)3·6H2O was according to the molar ratio of Ni to Y (1: 9, 3: 7, 5: 5, 7: 3, 9: 1 and 9.5: 0.5). The mixture was stirred at room temperature for 10 minutes and then dispersed by ultrasound for 10 minutes. Then, N2 flow was used to take small droplets atomized by the atomizer into the high-temperature tubular furnace (450 °C) for decomposition reaction, and the samples were pumped to the receiver for collection. The collected samples were dried overnight in an oven at 60 °C, and then calcined at 400 °C (2 °C min−1 of ramping rate). The obtained sample was named NiaYbOx, where a: b was the molar ratio of Ni and Y. Other composite catalysts (Ni9Ce1Ox, Ni9Sm1Ox and Ni9Al1Ox) and oxides (NiO and Y2O3) followed the same synthetic method with NiaYbOx, where only nitrates were altered.

N2 adsorption-desorption measurements

The N2 adsorption-desorption measurements was proceeded on a Builder SSA-4200 surface area analyzer at 77 K after degased the NiaYbOx catalysts at 200 °C for 6 h under vacuum. The specific surface area (SBET) was calculated from Brunauer-Emmett-Teller (BET) method.

Transmission electron microscopy (TEM)

The TEM images were taken on a JEOL JEM-2100F microscope operating at 100 kV. The High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images were obtained on a JEOL ARM200F microscope equipped with a probe-forming spherical-aberration corrector and Gatan image filter (Quantum 965). The element mapping results were derived from the electronic energy-loss spectroscopy (EELS) analysis equipped with STEM model.

Scanning electron microscopy (SEM)

The SEM images were taken on a Zeiss SUPRA 55 scanning microscope with the acceleration voltage of 5.0 kV.

X-ray diffraction (XRD)

Both ex situ and in situ modes were carried out on a PANalytical X’pert3 powder diffractometer (40 kV, 40 mA) using Cu Kα radiation (λ = 0.15406 nm). The powder samples were placed inside a quartz holder and the diffraction angles (2θ) ranged from 10 to 90°. For the in situ XRD experiments, 100 mg catalysts were loaded into an Anton Paar XRK900 in situ chamber. The XRD data was measured from 25 to 450 °C in 5% H2/Ar mixture (30 mL min−1) or 2% CO/Ar mixture (30 mL min−1).

X-ray photoelectron spectroscopy (XPS)

XPS characterization was carried out on a Thermo scientific ESCALAB Xi+ XPS spectrometer with Al Kα radiations, and with the C 1 s peak at 284.8 eV as an internal standard for all the spectra. For the quasi in situ XPS experiments, 20 mg catalysts were loaded into an HPGC 300 reaction chamber made by Fermion instruments (Shanghai) Co., LTD. The XPS data was measured after the in situ WGS reaction (400 °C, 2%CO/~3%H2O in Ar, 15 mL min−1).

X-ray absorption fine structure (XAFS)

Y K-edge (17.038 keV) XAFS was performed at the BL01B1 beamline of Super Photon ring − 8 GeV (SPring-8, Japan). XAFS spectra (16.714–18.499 keV) were measured in transmission mode using Si (311) DCM and ion chambers. Y2O3 was used for energy shift calibration. XAFS data was analyzed using Demeter software package (including Athena and Artemis, version 0.9.26). Athena was used for data normalization and Artemis was used for EXAFS fitting.

Temperature-programmed reduction

The temperature-programmed reduction by H2 (H2-TPR) was carried on a Builder PCSA-1000 instrument. 30 mg catalysts (20 − 40 mesh) were pretreated in air at 300 °C for 0.5 h and then purged with Ar. The final test was carried out in 5%H2/Ar (30 mL min−1) from room temperature to 700 °C. The temperature-programmed reduction by CO (CO-TPR) was carried on an online mass spectrometer (TILON LC-D200M). 100 mg catalysts (20 − 40 mesh) were pretreated by 5%H2/Ar (30 mL min−1) at 400 °C for 0.5 h and then purged with Ar. After the WGS reaction at 300 °C for 2 h, the catalyst was heated from room temperature to 800 °C in the 2%CO/Ar (30 mL min−1) to obtain the result of CO-TPR-1. Next, the sample were cooled to room temperature and switched to the retreatment gas (2%CO/~3%H2O in Ar at RT, 30 mL min−1; or ~3%H2O in Ar at 300 °C, 30 mL min−1) for 0.5 h. Then, the catalyst was heated from room temperature to 800 °C to obtain the result of CO-TPR-2.

In situ Raman spectroscopy

All the Raman spectra were collected from a LabRAM HR800 spectrometer (HORIBA JY) with 632.8 nm laser. Samples were placed on the micro-Raman reaction cell (Xiamen TOPS) and Raman shift ranged from 200 to 800 cm−1 with a spectral resolution of 2 cm−1. For the first mode, the samples were pretreated by 5%H2/Ar (30 mL min−1) at 400 °C for 0.5 h, and then cooled to room temperature. Next, WGS reaction gas (2%CO/~3%H2O in Ar, 30 mL min−1) was fed and the samples were heated from room temperature to 400 °C. The second mode was the switch experiment. After pretreated in 5%H2/Ar (30 mL min−1) at 400 °C, the samples were purged with two kinds of feeds (2%CO/Ar, 30 mL min−1; or ~3%H2O in Ar, 30 mL min−1) at 400 °C in turn.

Temperature-programmed surface reaction (TPSR)

The TPSR experiment was measured by the self-constructed device, using an online mass spectrometer (TILON LC-D200M) as the detector. The experiments were mainly divided into two parts. In the first part, the H2O dissociation experiment, 100 mg catalysts (20 − 40 mesh) were pretreated by 5%H2/Ar (30 mL min−1) at 400 °C for 0.5 h and then purged with Ar. After WGS reaction at 300 °C for 2 h, the samples were heated from room temperature to 600 °C under the ~3%H2O/Ar atmosphere to get the heating process H2O dissociation result. The constant temperature H2O dissociation experiment process was carried out after the WGS reaction. ~3%H2O/Ar was fed at a constant temperature (300, 350 and 400 °C) to collect the results. The second part was the reaction of CO with surface OH. After the same pretreatment and in situ reaction process, Ar purge was carried out at 300 °C for 2 h. Then, the samples were cooled to room temperature and switched to the 2%CO/Ar. Next, the results were collected at constant temperature (250 °C).

The density functional theory (DFT) calculations

The heterogeneous catalysis simulations were carried out using the Vienna ab Initio simulation package55,56,57,58. The Perdew−Burke−Ernzerhof (PBE) exchange−correlation functional and the projector augmented-wave (PAW) pseudopotential were applied to spin-unrestricted geometry optimizations58,59,60. The cutoff energy for the plane wave basis was set to 400 eV. The convergence threshold of the electronic self-consistency was specified as 1.0×10−6 eV, and the total energy change of the whole catalyst system between two ionic relaxation steps was designated as less than 0.02 eV. The cleaved Ni {111} surface comprised a 5-layer slab (the bottom 2 layers were fixed) as well as a vacuum layer of 15 Å. The lattice parameters of the Ni {111} surface were a = b = 12.4592 Å, c = 23.1383 Å, α = β = 90° and γ = 120°, and the number of Ni atoms was 125. Four oxygen atoms were positioned on the top layer of Ni to simulate NiOx, then Y3O4 cluster was loaded on NiOx layer, shown as Supplementary Fig. 8. As a group, the formula of the Y3O4/NiOx/Ni{111} model was Ni125O8Y3. The binding force between Y3O4 cluster and NiOx was Y-O or Ni-O ionic bonds, and the NiOx layer and Ni{111} was bound by metallic bond.

Catalytic performance measurement

The WGS activities of the NiaYbOx catalysts were tested in a self-constructed fixed-bed flow reactor. The temperature controller (UDIAN, XIAMEN YUDIAN AUTOMATION TECHNOLOGY CO., LTD.) was used in the reactor temperature control system. Both the actual temperature of the furnace temperature and the location of the catalyst were detected. The gas path system consisted of two parts: CO was the standard gas, and H2O was generated by vaporization after quantitative pumping by water injection pumps. The actual reaction gas content was 2%CO, 10%H2O, and the equilibrium gas was N2. During the test, 100 mg catalysts (20 − 40 mesh) were filled with gas hourly space velocity (GHSV) was 42,000 cm3 gcat −1 h−1. The catalysts were pretreated in a 5% H2/N2 at 400 °C. After a cooling system, the dry gas entered the gas analyzer (Gasboard-3100, Wuhan Sifang Corp), and then the real-time CO, CO2 and CH4 contents were obtained. The CO conversion and CH4 selectivity were calculated through Eq. (1) and Eq. (2), respectively.

$${X}_{{{{{{\rm{CO}}}}}}}=\frac{{n}_{{{{{{\rm{CO}}}}}}}^{{{{{{\rm{in}}}}}}}-{n}_{{{{{{\rm{CO}}}}}}}^{{{{{{\rm{out}}}}}}}}{{n}_{{{{{{\rm{CO}}}}}}}^{{{{{{\rm{in}}}}}}}}\times 100\%$$
(1)
$${S}_{{{{{{{\rm{CH}}}}}}}_{4}}=\frac{{n}_{{{{{{{\rm{CH}}}}}}}_{4}}^{{{{{{\rm{out}}}}}}}}{{n}_{{{{{{{\rm{CO}}}}}}}_{2}+}^{{{{{{\rm{out}}}}}}}{n}_{{{{{{{\rm{CH}}}}}}}_{4}}^{{{{{{\rm{out}}}}}}}}\times 100\%$$
(2)
$${{{{{\rm{TOF}}}}}}=\frac{{{{{{\rm{CO}}}}}}\; {{{{{\rm{converted}}}}}}\,({{{{{\rm{mol}}}}}})}{{{{{{\rm{Ni}}}}}}\; {{{{{\rm{atom}}}}}}\; {{{{{\rm{at}}}}}}\; {{{{{\rm{interface}}}}}}\,\left({{{{{\rm{mol}}}}}}\right)\times {{{{{\rm{time}}}}}}\,({{{{{\rm{s}}}}}})}$$
(3)

The stability test of the Ni9Y1Ox catalyst was conducted at 300 °C (GHSV = 420,000 cm3 gcat −1 h−1) for 100 h. The reaction rate was measured using 5 mg catalysts mixed with 200 mg quartz sand, and the CO conversion was controlled below 15%. The turnover frequency (TOF) value were calculated according to the Eq. (3). The kinetic order of CO and H2O was measured by adjusting the content of CO and H2O in the mixed gas at constant temperature (240−280 °C).