Perovskite-triggered dual exsolution of oxygen-deficient CeO₂ matrix and NiFe nanoalloys for enhanced CO₂ electrolysis

Abstract

Perovskite cathodes for CO₂ electrolysis offer excellent redox stability but suffer from limited activity. Although in situ exsolution of B-site cations is a powerful strategy to alleviate this issue, this process often triggers co-segregation of insulating AO_x phases, which diminishes active site exposure and ionic-electronic conductivity. Here we addressed the critical issue of insulating phase segregation in conventional exsolved perovskite by developing a Ce-doping strategy in Sr_1.95Ce_0.05Fe_1.3Ni_0.2Mo_0.5O_6−δ (SCeFNM). This material, upon annealing in a reducing atmosphere, allowed in situ construction of nanoscale CeO₂–NiFe/oxide heterostructures (CeO₂–NiFe@SCeFNM) by co-exsolving oxygen-deficient CeO₂ phase and NiFe alloy nanoparticles (NPs) on the surface. The unique architecture achieved a high current density of 1.57 A cm⁻² at 1.5 V and 850 °C with a CO faradaic efficiency of over 96%, outperforming NiFe@SFNM and their counterparts. Combined results demonstrated that the superior activity mainly came from the synergy within the “three-in-one” heterostructure, where NiFe alloy NPs increased the electronic conductivity, and CeO₂ phase extended the O²⁻ migration channels capable of enhancing CO₂ adsorption and activation, while the perovskite backbone ensured structural integrity. This study establishes a universal paradigm for constructing advanced catalysts for diverse applications via co-exsolution of metal NPs and defective oxide phases from appropriately A-site-doped perovskites.

---

Green foundation

1. High-temperature CO₂ electrolysis within SOECs has emerged as a transformative pathway, which enables direct conversion of CO₂ into CO or syngas using renewable energy. This study innovatively addresses the critical issue of an insulating SrO phase inherent in conventional exsolution processes by strategically constructing heterostructured CeO₂–NiFe@SCeFNM as a SOEC cathode material, thereby advancing green chemistry and contributing to sustainable technologies.

2. We successfully achieved efficient conversion of CO₂ into industrially vital CO using renewable electricity in an SOEC, which delivered a high current density of 1.57 A cm⁻² at 850 °C and 1.5 V, together with a high CO faradaic efficiency of over 96%, therefore significantly enhancing carbon utilization efficiency.

3. Future optimization to avoid AO phase segregation of perovskites via lowering operating temperatures or substituting rare-earth elements at A-sites could further advance their sustainability and scalable implementation of electrocatalytic carbon-neutral technologies.

1. Introduction

The increasing consumption of fossil fuels has led to significant carbon dioxide (CO₂) emissions, which profoundly impact global warming. An urgency to address climate change, therefore, has propelled CO₂ reduction technologies to the forefront of sustainable energy research, offering a dual solution for CO₂ utilization and renewable energy storage.^1,2 Among them, high-temperature CO₂ electrolysis within solid oxide electrolysis cells (SOECs) has emerged as a promising technology, enabling direct conversion of CO₂ into CO or syngas using renewable energy.^3,4 Despite the remarkable catalytic performance of Ni-yttrium stabilized zirconia (Ni-YSZ) cermet as an SOEC cathode, it suffers from intrinsic redox instability, agglomeration and severe coking issues, which critically limit its further utilization at larger industrial-scale application.^5–7 Alternatively, perovskite oxides (e.g., ABO₃, A₂BO₄) have been extensively investigated as promising catalysts for SOECs due to their mixed ionic-electronic conductivity and superior redox stability.^8,9 Nevertheless, a major limitation lies in their relatively inferior catalytic activity, which leads to an insufficient electrolysis performance, highlighting the imperativeness of exploring smart design strategies for perovskite oxide electrodes.

Surface engineering via in situ exsolution has emerged as a powerful strategy to enhance catalytic activity by enabling controlled growth of nanoscale metallic nanoparticles (NPs) on perovskite oxide surfaces.^4,10–12 This strategy not only achieves a uniform dispersion of catalytic NPs across porous electrode architectures, but also generates abundant oxygen vacancies at the metal–oxide interfaces, thereby creating additional electrochemically active sites that improve CO₂ adsorption and activation capabilities. Furthermore, it ensures a high operational stability by mitigating active site loss caused by catalyst detachment and/or agglomeration, while maintaining structural integrity under harsh operating conditions.¹³ Among various metal NPs (e.g., Fe, Co, Ni, and Cu), in situ exsolution of Fe-/Ni-based metals/alloys has been widely employed since Fe shows exceptional CO₂ adsorption capacity and coking resistance, while Ni demonstrates a superior catalytic activity with the lowest activation energy barrier.^12,14 However, the exsolution of B-site reducible cations may induce the intergrowth of AO_x phases, which alters the functional properties of the material.^11,15 Previous studies have demonstrated that exsolution of B-site cations from perovskite oxides triggers SrO co-segregation on occasion—a process driving chronic electrochemical performance decay since the electronically and ionically insulating nature of surface-terminated SrO obstructs surface charge transfer and oxygen transport pathways.^16,17 Moreover, in cation-ordered LnBaCo₂O_5+δ perovskites (Ln = lanthanide), BaO segregation is also kinetically accelerated at temperatures exceeding 400 °C, resulting in a diminished surface coverage of Co active sites on the perovskite oxide framework.¹⁸

However, it is worth noting that the exsolution of A-site oxides is not always detrimental, since precise design of A-site oxide exsolution offers unprecedented opportunities to tailor CO₂ adsorption and oxygen exchange reactions, which serves as an alternative approach to enhancing SOEC performance. For instance, the exsolution of CaO and NiFe NPs from La_0.6Ca_0.4Fe_0.8Ni_0.2O_3−δ doubled its specific surface area.¹⁹ Likewise, Zhu et al. synthesized self-generated Fe/MnO_x NPs by regulating A-site deficiency, and the interfacial charge redistribution of Fe/MnO_x optimized CO₂ dissociation pathways and promoted chemical adsorption and dissociation of CO₂ molecules on the cathode surface.²⁰ Moreover, the exsolution of oxides can also increase oxygen vacancy content and boost O²⁻ migration. Previous studies have reported that Bi-doped La_0.4Sr_0.6Co_0.2Fe_0.7Nb_0.1O_3−δ exsolved Bi₂O₃ and CoFe alloy NPs, which raised the oxygen vacancy concentration by 40%,²¹ while oxygen vacancy-enriched PrO_x NPs enhanced O²⁻ migration capability, which accelerated O²⁻ exchange and migration processes on the electrode surface.²² Therefore, precise engineering of oxide-metal/oxide heterointerfaces holds the potential to enhance CO₂ electrolysis activity and stability. Cerium oxide (CeO₂), due to its exceptional oxygen storage capacity and redox cycling between Ce³⁺ and Ce⁴⁺, possesses a desirable level of O²⁻ conductivity, and the redox couple of Ce^4+/3+ on the surface accelerates O²⁻ transfer between CO₂ and CO. Meanwhile, NiFe alloy NPs, characterized by their high electronic conductivity and robust thermal stability, are promising candidates for CO₂ dissociation and CO desorption. However, individual NiFe alloy NPs suffer from insufficient O²⁻ migration, while pure CeO₂ lacks the electronic conductivity required for efficient charge transfer in CO₂ electrolysis.

Recognizing this complementary potential between CeO₂ and NiFe alloy NPs, herein we designed and synthesized a novel Ce-doped heterostructured catalyst by annealing Sr_1.95Ce_0.05Fe_1.3Ni_0.2Mo_0.5O_6−δ (SCeFNM) in a reducing atmosphere to strategically construct metallic NPs with oxygen-vacancy-rich CeO₂ on SCeFNM (CeO₂–NiFe@SCeFNM), realizing dual exsolution of NiFe alloy and CeO₂ NPs with enhanced electronic conductivity, O²⁻ migration, and CO₂ adsorption and activation. Experimental results demonstrated that a cell with a CeO₂–NiFe@SCeFNM cathode achieved a comparable current density (j) of 1.57 A cm⁻² at 850 °C and 1.5 V for pure CO₂ atmosphere, together with a high CO faradaic efficiency (FE_CO) of over 96% and a stability of more than 120 hours, showcasing an exceptional catalytic activity for CO₂ electrolysis. The strategic integration of dual phases through doping-induced co-precipitation offers a compelling approach to synergizing their complementary functionalities.

2. Experimental methods

2.1 Material synthesis

Sr₂Fe_1.3Ni_0.2Mo_0.5O_6−δ (SFNM), SCeFNM, and La_0.4Ce_0.6O₂ (LDC) were obtained via a sol–gel method.²³ Stoichiometric amounts of Sr(NO₃)₂, Fe(NO₃)₃·9H₂O, Ni(NO₃)₃·6H₂O, (NH₄)₆Mo₇O₂₄·4H₂O and Ce(NO₃)₃·6H₂O were dissolved in EDTA-NH₃·H₂O combined solution under continuous heating and stirring; citric acid was then introduced into the solution. The molar ratio of EDTA : citric acid : total cation was controlled to be around 1 : 1.5 : 1. Subsequently, NH₃·H₂O was added to adjust the pH value to around 8. The solution was stirred and heated on a hot plate at 80 °C until the formation of organic resins containing homogeneously distributed cations because of the slow evaporation of the solvent. The formed gels were heated at 300 °C for four hours to remove the organic components and the nitrates, followed by being annealed at 1100 °C for 10 hours to achieve the target powders. NiFe@SFNM and CeO₂-NiFe@SCeFNM were prepared by annealing SFNM and SCeFNM powders at 850 °C for 2 hours in a 5% H₂/Ar atmosphere with a flow rate of 50 mL min⁻¹ in a tubular furnace. The heating and cooling rates were both controlled at 2 °C min⁻¹. The La_0.8Sr_0.2Ga_0.8Mg_0.2O_3−δ (LSGM) electrolyte was prepared by a dry-press method, and then calcined at 1450 °C for 5 hours. The (La_0.6Sr_0.4)_0.95Co_0.2Fe_0.8O_3−δ (LSCF) and Ga_0.2Ce_0.8O_2−δ (GDC) powders were used as received.

2.2 Material characterization

The crystalline structure of all the synthesized powders was identified by X-ray diffraction (XRD) with Rigaku Rota flex Cu Kα radiation (40 kV, 44 mA), and the raw data were analyzed with JADE version 6.5. Microstructures were determined with a high-resolution Zeiss Sigma field emission scanning electron microscope (SEM). The low- and high-resolution transmission electron microscopy (TEM) images were acquired by using a Thermo Scientific Talos F200X STEM at an accelerating voltage of 200/300 kV. Energy-dispersive X-ray spectroscopy (EDX) combined with high-resolution TEM (HR-TEM) was applied to characterize the composition and element distribution. X-ray photoelectron spectroscopy (XPS, Kratos AXIS Ultra) was used to investigate the surface chemistry of SFNM, SCeFNM, and CeO₂–NiFe@SCeFNM with adventitious carbon (C 1s) at a binding energy (BE) of 284.8 eV as the reference. Temperature-programmed desorption of CO₂ (CO₂-TPD) was investigated with a Micromeritics chemisorption analyzer (Auto Chem II 2920) to study the CO₂ desorption. The CO₂ adsorption ability was also characterized by Fourier transform infrared (FTIR) spectroscopy (Nicolet iS50). The oxygen vacancy content was measured using an iodometric titration method under Ar atmosphere.

2.3 Cell fabrication and measurement

The LDC buffer layer was formed by screen-printing an α-terpinol/ethyl cellulose-bound slurry onto LSGM followed by 1350 °C calcination for 2 hours. Subsequently, CeO₂–NiFe@SCeFNM-GDC, NiFe@SFNM-GDC, SCeFNM-GDC, SFNM, and LSCF-GDC powders were separately mixed with a binder composed of α-terpinol and ethyl cellulose ether to form their corresponding slurries. GDC was incorporated into all cathodes primarily to enhance the oxygen-ion conductivity and extend the triple-phase boundaries. The electrode slurry was screen printed on the LSGM electrolyte, followed by calcining at 1100 °C for 2 hours, to obtain a cell configuration of CeO₂–NiFe@SCeFNM-GDC∥LDC∥LSGM∥LSCF-GDC. The effective area of the electrode was 0.5 cm² and Ag paste was used as the current collector. The as-fabricated single cell was mounted on one end of an alumina support for electrochemical measurements, with the LSCF-GDC anode exposed to ambient air. Before testing, the CeO₂–NiFe@SCeFNM cathode compartment was purged with 5% H₂/Ar. Then, pure CO₂ was fed to the cathode at a flow rate of 50 mL min⁻¹ for testing. A schematic diagram of the high-temperature CO₂ electrolysis setup is shown in Fig. S1. The electrochemical performance of the single cell was evaluated using an EnergyLab XM electrochemical workstation from OCV to 1.6 V at a scan rate of 20 mV s⁻¹. Electrochemical impedance spectra (EIS) were acquired with an applied AC amplitude of 10 mV in the frequency range of 0.1 Hz to 1.0 MHz. The CO₂ and CO contents of outlet gas were determined by online gas chromatography (Agilent 8860) using a thermal conductivity detector.

3. Results and discussion

The pristine perovskite oxides, SFNM and SCeFNM, were synthesized using a facile sol–gel method, as reported elsewhere.^24,25 The XRD patterns of SFNM and SCeFNM powders after calcination at 1100 °C were detected without impure phase (Fig. S2), but slight shifts of all diffraction peaks toward higher angles were observed for SCeFNM, indicating lattice volume shrinkage due to the partial substitution of a smaller ionic radius of Ce³⁺ (1.14 Å) to Sr²⁺ (1.34 Å). After annealing SFNM and SCeFNM powders at 850 °C in 5% H₂/Ar for 2 hours, the double perovskite oxide phase of SCeFNM was transformed into a Ruddlesden–Popper phase (Fig. 1a), together with dual exsolutions of A-site CeO₂ phase and B-site NiFe alloy NPs (CeO₂–NiFe@SCeFNM). In contrast, only NiFe alloy NPs were in situ exsolved from SFNM (NiFe@SFNM), accompanied with a slight phase transformation. To gain insights into the microstructural features, SFNM, SCeFNM, NiFe@SFNM, and CeO₂–NiFe@SCeFNM powders were subjected to SEM analysis. It was seen that SFNM and SCeFNM specimens showed smooth network structures (Fig. S3) whereas substantial NiFe alloy NPs appeared on the reduced SFNM and SCeFNM. Different from NiFe@SFNM, SCeFNM with Ce doping possessed a greater amount of exsolved NiFe alloy NPs, alongside a minor quantity of CeO₂ NPs (Fig. 1b). For further confirmation, HR-TEM was conducted, and it was found that lattice spacings of 0.202 and 0.313 nm were observed in HR-TEM images (Fig. 1c–d and S4) which corresponded to the (111) planes of CeO₂ and NiFe alloy, respectively. Accordingly, the high-angle annular dark-field scanning TEM (HADDF-STEM) image and EDX elemental mapping of CeO₂–NiFe@SCeFNM (Fig. 1e and f) demonstrated that the perovskite oxide surface was uniformly dispersed with CeO₂ and abundant NiFe alloy NPs with an average size of 17.5 ± 3.2 nm, while other elements (e.g., Sr, Mo, O) were homogeneously distributed within the perovskite oxide matrix. Moreover, the exsolved NPs were firmly embedded on the CeO₂–NiFe@SCeFNM substrate, indicating the presence of strong oxide-metal/oxide interactions at the heterointerfaces.

Fig. 1 (a) XRD patterns of NiFe@SFNM and CeO₂–NiFe@SCeFNM, together with an enlargement of the region at 29–30°. (b) SEM image of CeO₂–NiFe@SCeFNM. HR-TEM images and corresponding FFT patterns of (c) NiFe NPs and (d) CeO₂ NPs. (e) HADDF-STEM image of CeO₂–NiFe@SCeFNM with (f) EDX mappings of all elements.

Surface chemical variation of the as-synthesized perovskite oxides was characterized via XPS. The XPS full survey spectrum exhibited a Ce element signal in the CeO₂–NiFe@SCeFNM sample, further confirming the presence of Ce⁴⁺ (Fig. S5). Clearly, Fig. 2a and b showed that an increase in A-site Ce-doping led to a decrease in the intensities of Fe³⁺ and Ni³⁺ species instead of an increase in those of Fe²⁺ and Ni²⁺ species. This implied that increasing A-site Ce-doping reduced the oxidation states of Fe and Ni, and caused an enhanced electronegativity, which induced the formation of more oxygen vacancies in turn to maintain the charge neutrality. Notably, two additional peaks at 707 and 852 eV were observed for CeO₂–NiFe@SCeFNM, which were ascribed to the presence of Fe⁰ and Ni⁰. Moreover, the O 1s spectra were deconvoluted into three subpeaks, representing the lattice oxygen (O_Latt), the adsorption oxygen (O_Ads), and the surface carbonate/hydroxyl oxygen (CO₃²⁻/OH⁻) species, respectively.²⁶ Typically, the peak of O_Ads represented oxygen vacancies, and it is worth noting that CeO₂–NiFe@SCeFNM demonstrated a higher concentration of O_Ads as compared with SFNM and SCeFNM (Fig. 2c), confirming its oxygen vacancy concentration had a considerable enhancement, which aligned well with the iodine titration results in Fig. 2d, confirming its superiority for CO₂ adsorption. To examine the impact of oxygen vacancies on CO₂, its adsorption behaviors on SFNM, SCeFNM, and CeO₂–NiFe@SCeFNM powders were explored in the temperature range of 50–850 °C using CO₂-TPD. There were no desorption peaks observed at low temperatures (<300 °C) in Fig. 2e, while the peaks in the high-temperature range (>300 °C) were mainly associated with the desorption of chemisorbed CO₂.²⁷ Clearly, the peak area for CeO₂–NiFe@SCeFNM, benefitting from the exsolved CeO₂ phase with abundant oxygen vacancies, was significantly larger than those for SFNM and SCeFNM, indicative of a superior CO₂ adsorption capacity.²⁸ The elevated desorption temperature of 754 °C indicated an improved bonding ability between CO₂ molecules and perovskite oxides,¹⁶ which coincided well with the working temperatures of SOECs. Furthermore, since carbonates are critical intermediate species in the CO₂ electrolysis process,²⁹ CO₂ chemisorption was further characterized by FTIR spectroscopy with samples being exposed to CO₂ at 800 °C for 10 hours. As shown in Fig. 2f, the CO₃²⁻ absorption signals in the infrared region of ∼1440 cm⁻¹ for CeO₂–NiFe@SCeFNM were stronger as compared to SFNM and SCeFNM, confirming that CeO₂–NiFe@SCeFNM could significantly enhance CO₂ adsorption capacity attributed to the greater amount of oxygen vacancies.³⁰ In particular, in situ FTIR spectra for CeO₂–NiFe@SCeFNM all presented strong CO₃²⁻ signals from 750 to 850 °C (Fig. S6), demonstrating its robust CO₂ adsorption capability.

Fig. 2 (a) Fe 2p spectra, (b) Ni 2p spectra, (c) O 1s spectra, (d) oxygen vacancy concentration determined by iodometric titration, (e) CO₂-TPD plots and (f) FTIR spectra of SFNM, SCeFNM, NiFe@SFNM and CeO₂–NiFe@SCeFNM.

To unravel the impacts of exsolved CeO₂ phase and NiFe alloy NPs as well as their constructed oxide-metal/oxide heterointerface on CO₂ electrolysis activity, SFNM, SCeFNM, NiFe@SFNM, and CeO₂–NiFe@SCeFNM were fabricated as cathodes within SOECs supported by LSGM electrolyte, and were tested in pure CO₂ atmosphere from 750 to 850 °C. Based on the current–voltage (I–V) curves, a gradual increase in j was observed for all the samples with an order of SFNM < SCeFNM < NiFe@SFNM < CeO₂–NiFe@SCeFNM (Fig. 3a and S7), delivering a j of 1.57, 1.45, 1.35 and 1.23 A cm⁻² at 1.5 V and 850 °C, respectively (Table S1). This implied that as compared to conventional SFNM with only NiFe alloy NPs and pristine SFNM, CeO₂–NiFe@SCeFNM provided more reactive centers precisely due to the dual exsolution of precipitated CeO₂ and NiFe NPs, both of which synergistically facilitated the activity. Concurrently, the key reaction steps and kinetic behaviors during CO₂ electrolysis were investigated using EIS (Fig. S8) and the distribution of relaxation times (DRT) (Fig. S9). The Nyquist plots in Fig. 3b illustrate that the polarization resistance (R_p) values of SFNM, SCeFNM, NiFe@SFNM, and CeO₂–NiFe@SCeFNM cathodes under open-circuit voltage (OCV) at 850 °C were 1.025, 0.885, 0.857, and 0.792 Ω cm², respectively, in accordance with the j variation trend. Given that the contributions from the LSCF anode and LSGM electrolyte to the total resistance were quite close for all cells, the polarization resistance discrepancy stemmed from the different cathodes. Based on previous studies involving DRT analysis it was known that the peaks at low, intermediate, and high frequency (LF, IF, and HF) are associated with CO₂ adsorption and activation, migration of O²⁻ in the electrodes, and charge/ion transfer at the electrode–electrolyte interface, respectively.^13,31 Apparently, the dual exsolution treatment reduced the LF region area while leaving the IF and HF regions unaffected (Fig. 3c), suggesting that the diminished LF contribution for the CeO₂–NiFe@SCeFNM cathode optimized the CO₂ adsorption and activation kinetics. Furthermore, the R_p value of CeO₂–NiFe@SCeFNM decreased markedly with increasing applied potential (Fig. S8a and Table S2), reflecting an accelerated CO₂ electrolysis kinetics under high potentials. Moreover, increasing the applied potential could lower the LF region area, thus improving CO₂ adsorption and activation, and the ultimate CO₂ electrolysis activity (Fig. S9a). In contrast, raising the operating temperature lowered both the ohmic resistance (R_o) and R_p (Fig. S8e and Table S3), and the accordingly reduced IF and LF region areas in DRT plots revealed that the increase in temperature not only enhanced CO₂ adsorption and activation, but also accelerated O²⁻ migration at the electrode (Fig. S9e).³² This was further verified by the electrochemical activation energy, an important parameter to assess the CO₂ electrolytic process. The Arrhenius plots in Fig. 3d show that the CeO₂–NiFe@SCeFNM cathode delivered a smaller activation energy of 0.992 eV than its counterparts, indicative of a reduced cathodic energy barrier. As a result, benefiting from the synergy of NiFe alloy NPs and CeO₂ phase, CeO₂–NiFe@SCeFNM achieved the optimal CO₂ electrolysis performance among all the control materials, and surpassed that of most of the state-of-the-art cathode materials reported to date, as shown in Fig. 3e, again confirming its superiority for CO₂ electrolysis in SOECs.

Fig. 3 (a) I–V curves, (b) Nyquist plots and (c) DRT results of SFNM, SCeFNM, NiFe@SFNM and CeO₂–NiFe@SCeFNM at 850 °C. (d) Arrhenius plots of various cathodes from 750 to 850 °C. (e) Activity comparison of all samples in this study with previously reported state-of-the-art materials for CO₂ electrolysis at 1.5 V, where SR represents the reference in SI.

As crucial parameters to evaluate CO₂ conversion efficiency, CO production rate (PR_CO) and FE_CO are equally important for the development of SOECs. The potentiostatic CO₂ electrolysis measurements at varying applied potentials revealed that CeO₂–NiFe@SCeFNM achieved maximum j in a wide potential range from 1.0 to 1.5 V (Fig. 4a), which matched well with the I–V results. Additionally, the corresponding FE_CO and PR_CO under different applied potentials at 850 °C were also evaluated (Fig. 4b). The single cell with the CeO₂–NiFe@SCeFNM cathode achieved a high FE_CO of 96.03%, surpassing the 90.94% on SFNM and closely matching those of 95.72% on NiFe@SFNM and 95.35% on SCeFNM in CO₂-to-CO conversion, at 1.5 V and 850 °C. In addition, at 1.5 V, the PR_CO of CeO₂–NiFe@SCeFNM reached 10.32 mL min⁻¹ cm⁻², faster than those of 9.68, 8.68 and 8.42 mL min⁻¹ cm⁻² over NiFe@SFNM, SCeFNM, and SFNM, respectively, which further confirmed that except for j, co-exsolution of CeO₂ phase and NiFe alloy NPs also improved CO₂ conversion rate. The long-term stabilities of the cells with different cathodes were also investigated, and Fig. 4c shows j and FE_CO as a function of time under a constant potential of 1.2 V at 850 °C. The CeO₂–NiFe@SCeFNM-assembled cell could maintain a good performance with a negligible j decay rate of 0.793 mA h⁻¹ during continuous operation of 120 hours, and retained a steady FE_CO of 90%, superior to that of the cells with NiFe@SFNM, SCeFNM, and SFNM. For a better holistic view, a radar map presenting j (A cm⁻²), 1/R_p (Ω⁻¹ cm⁻²), PR_CO (mL min⁻¹ cm⁻²), FE_CO and stability retention rate per 100 hours (R_sr = j₁₀₀/j_initial)²⁵ is plotted in Fig. 4d for comparison. It illustrated clearly that dual exsolution was more beneficial for CO₂ electrolysis, which also highlighted the critical role of CeO₂ phase in tuning CO₂ electrolysis activity. In addition, no extra phase and impurity were observed in the XRD pattern post-test for the CeO₂–NiFe@SCeFNM cathode (Fig. 4e), indicating a good structural stability under harsh electrolysis conditions. Meanwhile, the cathode-electrolyte interface remained closely stacked, and no delamination was detected after long-term CO₂ electrolysis operation (Fig. 4f and S10a, b). Importantly, the perovskite oxide matrix still retained the uniformly anchored in situ exsolved NPs without severe agglomeration for the CeO₂–NiFe@SCeFNM cathode (Fig. 4g). However, there were no noticeable in situ exsolved NPs on the NiFe@SFNM cathode (Fig. S10c), clarifying a good structural resilience under high-temperature harsh conditions in the mixture of CO₂/CO, as also confirmed by the Raman spectrum of the CeO₂–NiFe@SFNM cathode after stability testing in Fig. 4h. Notably, no obvious peaks associated with SrCO₃ (1050 cm⁻¹), D-bond (1340 cm⁻¹) and G-bond (1580 cm⁻¹) were observed, evidencing the good coking resistance and the strong suppression to Sr segregation of CeO₂–NiFe@SCeFNM.²⁷

Fig. 4 (a) The potentiostatic measurements for CO₂ electrolysis and (b) PR_CO and corresponding FE_CO under different applied potentials of all samples. (c) Long-term stability tests and the corresponding FE_CO at 1.2 V and 850 °C for NiFe@SFNM and CeO₂–NiFe@SCeFNM. (d) Radar map of SFNM, SCeFNM, NiFe@SFNM and CeO₂–NiFe@SCeFNM. (e) XRD patterns and SEM images of (f) a cross-section of a single cell and (g) CeO₂–NiFe@SCeFNM cathode as well as (h) Raman spectra of the CeO₂–NiFe@SCeFNM cathodes before and after stability testing.

To clearly elaborate the mechanism credited for the superior electrochemical activity on CeO₂–NiFe@SCeFNM, a schematic illustration of the possible reaction pathways over SFNM, NiFe@SFNM and CeO₂–NiFe@SCeFNM is shown in Fig. 5. It was seen that when SFNM was assembled as a cathode, its activity was limited to the slow gas–solid interface exchange process involving CO₂ adsorption, activation, and O²⁻ migration on the cathode surfaces (Fig. 5a). Nevertheless, the introduction of NiFe alloy NPs characterized by high electronic conductivity could promote charge transfer, and CO₂ adsorption and activation in CO₂ electrolysis, but individual NiFe alloy NPs suffered from insufficient O²⁻ migration (Fig. 5b). Thus, we proposed the “three-in-one” strategy in this study by integrating both NiFe alloy NPs and defective CeO₂ phase in situ exsolved from SCeFNM (Fig. 5c), which synergistically increased the ionic-electronic conductivity and the O²⁻ migration channels from CeO₂ to SCeFNM to promote CO₂ adsorption and activation. In addition, the intimate contact of SCeFNM, NiFe alloy NPs and CeO₂ at the nanoscale also effectively extended the triple-phase boundary length dedicated to CO₂ electrolysis, thus culminating in the exceptional CO₂ electrolysis activity.

Fig. 5 Schematic illustrations of possible reaction pathways for CO₂ electrolysis over (a) SFNM, (b) NiFe@SFNM and (c) CeO₂–NiFe@SCeFNM.

4. Conclusions

In summary, we synthesized Sr_1.95Ce_0.05Fe_1.3Ni_0.2Mo_0.5O_6−δ (SCeFNM) using a Ce-doping strategy, and strategically constructed CeO₂–NiFe/oxide heterostructures by in situ co-exsolving oxygen-vacancy-rich CeO₂ phase and NiFe alloy NPs from perovskite oxide (CeO₂–NiFe@SCeFNM) to avoid the co-segregation of insulating SrO phase. It was found that CeO₂–NiFe@SCeFNM cathode achieved a j of 1.57 A cm⁻² at 850 °C and 1.5 V, and a CO faradaic efficiency of 96%, outperforming NiFe@SFNM without oxygen-deficient CeO₂ phase, and pristine SFNM and SCeFNM. Crucially, the cathode demonstrated an exceptional operational stability over 120 hours without carbon deposition and particle agglomeration, overcoming the degradation limitations of traditional exsolution. Combined results and spectra collectively verified that the enhanced activity exhibited by CeO₂–NiFe@SCeFNM was mainly derived from the three-in-one structure, where the NiFe alloy NPs on CeO₂–NiFe@SCeFNM enhanced the electronic conductivity, and the CeO₂ phase facilitated O²⁻ migration as well as CO₂ adsorption and activation, while the perovskite oxide framework ensured the structural integrity. The strong interaction between the three phases in the nanodomain creates a “win–win” effect, and results in an optimal CO₂ electrolysis activity, which also presents a promising approach to enhancing SOEC performances by dual exsolution of A-site metal oxides and B-site metal NPs from perovskite oxides.

Author contributions

Biao Ouyang: conceptualization, investigation, writing – original draft preparation. Ming Yang: validation, data curation. Lin-Bo Liu: methodology. Shuo Liu: formal analysis. Yan Li: data curation. Xian-Zhu Fu: supervision. Yifei Sun: conceptualization. Subiao Liu: conceptualization, project administration, resources, supervision, writing – review and editing. Jing-Li Luo: supervision, validation.

Conflicts of interest

There are no conflicts of interest to declare.

Data availability

The data supporting this paper have been included as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d5gc04390e.

Acknowledgements

This work is supported by the Pilot Group Program of the Research Fund for International Senior Scientists (Grant No. 22350710789) and the Start-up Funding of Central South University (No. 206030104). This project was also supported by Shenzhen Science and Technology Program (JCYJ20240813145707010, JCYJ20220530143401002, JCYJ20241202124410015). The authors are grateful for resources from the High-Performance Computing Center of Central South University.

References

1. H. D. Matthews and S. Wynes, Science, 2022, 376, 1404–1409.
2. P. De Luna, C. Hahn, D. Higgins, S. A. Jaffer, T. F. Jaramillo and E. H. Sargent, Science, 2019, 364, eaav3506.
3. X. Hou, Y. Jiang, K. Wei, C. Jiang, T.-C. Jen, Y. Yao, X. Liu, J. Ma and J. T. S. Irvine, Chem. Rev., 2024, 124, 5119–5166.
4. J.-h. Myung, D. Neagu, D. N. Miller and J. T. S. Irvine, Nature, 2016, 537, 528–531.
5. T. L. Skafte, P. Blennow, J. Hjelm and C. Graves, J. Power Sources, 2018, 373, 54–60.
6. Y. Sun, Z. Liu, W. Zhang, X. Chu, Y. Cong, K. Huang and S. Feng, Small, 2019, 15, 1803513.
7. J. Ni, L. Chen, J. Lin and S. Kawi, Nano Energy, 2012, 1, 674–686.
8. R. Xu, S. Liu, M. Yang, G. Yang, Z. Luo, R. Ran, W. Zhou and Z. Shao, Chem. Sci., 2024, 15, 11166–11187.
9. M. Yang, L.-B. Liu, S. Liu, Y. Li, B. Ouyang, X.-Z. Fu, J.-L. Luo, Y. Sun and S. Liu, Chin. Chem. Lett., 2024, 110603.
10. Q. A. Islam, S. Paydar, N. Akbar, B. Zhu and Y. Wu, J. Power Sources, 2021, 492, 229626.
11. B.-W. Zhang, M.-N. Zhu, M.-R. Gao, X. Xi, N. Duan, Z. Chen, R.-F. Feng, H. Zeng and J.-L. Luo, Nat. Commun., 2022, 13, 4618.
12. H. Lv, T. Liu, X. Zhang, Y. Song, H. Matsumoto, N. Ta, C. Zeng, G. Wang and X. Bao, Angew. Chem., Int. Ed., 2020, 59, 15968–15973.
13. F. He, M. Hou, F. Zhu, D. Liu, H. Zhang, F. Yu, Y. Zhou, Y. Ding, M. Liu and Y. Chen, Adv. Energy Mater., 2022, 12, 2202175.
14. S. Jo, Y. H. Kim, H. Jeong, C.-h. Park, B.-R. Won, H. Jeon, K. T. Lee and J.-h. Myung, Appl. Energy, 2022, 323, 119615.
15. D. He, W. Ruan, J. Li, J. Ni and C. Ni, Chem. Eng. J., 2022, 433, 133632.
16. B.-W. Zhang, M.-N. Zhu, M.-R. Gao, J. Chen, X. Xi, J. Shen, R.-F. Feng, N. Semagina, N. Duan, H. Zeng and J.-L. Luo, Angew. Chem., Int. Ed., 2023, 62, e202305552.
17. B. Koo, K. Kim, J. K. Kim, H. Kwon, J. W. Han and W. Jung, Joule, 2018, 2, 1476–1499.
18. H. Téllez, J. Druce, J. A. Kilner and T. Ishihara, Faraday Discuss., 2015, 182, 145–157.
19. Y. Tian, Y. Liu, A. Naden, L. Jia, M. Xu, W. Cui, B. Chi, J. Pu, J. T. S. Irvine and J. Li, J. Mater. Chem. A, 2020, 8, 14895–14899.
20. J. Zhu, W. Zhang, Y. Li, W. Yue, G. Geng and B. Yu, Appl. Catal., B, 2020, 268, 118389.
21. E. Wang, C. Jin, L. Zhao, Z. Yang, C. Liu, S. Wang, X. Lei, M. Chao, H. Xu and R. Yang, Chem. Eng. J., 2024, 494, 153129.
22. Y. Yi, X. Xi, L. Huang, X. Liu, Y. Liao, J. Liu, J. Long, J. Zhang, X.-Z. Fu and J.-L. Luo, Adv. Funct. Mater., 2025, 35, 2412486.
23. S. Liu, H. Luo, Y. Li, Q. Liu and J.-L. Luo, Nano Energy, 2017, 40, 115–121.
24. L.-B. Liu, Y.-F. Tang, S. Liu, M. Yu, Y. Sun, X.-Z. Fu, J.-L. Luo and S. Liu, Adv. Energy Mater., 2025, 15, 2402967.
25. S. Liu, L.-B. Liu, M. Yang, Y. Li, Q. Pan, Y. Sun, X.-Z. Fu, S. Liu and J.-L. Luo, Chem. Eng. J., 2025, 519, 165404.
26. F. Hu, Y. Ling, S. Fang, L. Sui, H. Xiao, Y. Huang, S. Wang, B. He and L. Zhao, Appl. Catal., B, 2023, 337, 122968.
27. Y. Li, Y. Li, Y. Wan, Y. Xie, J. Zhu, H. Pan, X. Zheng and C. Xia, Adv. Energy Mater., 2019, 9, 1803156.
28. Y. Luo, X. Chang, J. Wang, D. Zhang, L. Fu, X.-K. Gu, Y. Wang, T. Liu and M. Ding, ACS Nano, 2025, 19, 1463–1477.
29. Y. Song, X. Zhang, K. Xie, G. Wang and X. Bao, Adv. Mater., 2019, 31, 1902033.
30. L. Zhang, C. Xu, W. Sun, R. Ren, X. Yang, Y. Luo, J. Qiao, Z. Wang, S. Zhen and K. Sun, Sep. Purif. Technol., 2022, 298, 121475.
31. X. Xi, Y. Fan, J. Zhang, J.-L. Luo and X.-Z. Fu, J. Mater. Chem. A, 2022, 10, 2509–2518.
32. L.-B. Liu, S. Liu, M. Yang, Y. Li, B. Ouyang, P. Zhang, X.-Z. Fu, Y. Sun, S. Liu and J.-L. Luo, Appl. Catal., B, 2025, 379, 125705.

---