Overcoming chromium poisoning in solid oxide cells through multiscale perovskite engineering Authors: Min Li, Huixian Liu, Yunfei Bu, Xiang Wang, Haoran Wang, Jiawei Chen, Jiawei Li, Yantao Zhao, Zhibin Yang Published: 2026 ABSTRACT Chromium poisoning and sluggish oxygen kinetics limit oxygen electrodes in reversible solid oxide cells. We introduce a triple barrier design that integrates lattice stabilization by Ba to suppress SrO segregation, BaCoO3 nanodomains that trap chromium, and surface acidity tuning via trace Mo at the B site in La0.6Sr0.1Ba0.35Co0.2Fe0.78Mo0.02O3₋δ. The electrode reaches 0.058 Ω·cm2 polarization resistance at 750 °C, 70.4 percent lower than LSCF, delivers 1.352 W·cm⁻2 at 800 °C, and 2.08 A·cm⁻2 at 1.5 V. Under 0.5 A·cm⁻2 with a chromium source, single cells operate stably for about 1000 hours. Quantitative characterization and thermodynamic analysis show suppressed SrCrO4 formation and shallow chromium ingress, consistent with BaCoO3 mediated trapping and Mo moderated surface basicity. Density functional theory at operating temperature and pressure indicates Ba and Mo together weaken CrO3 adsorption. The design reconciles high activity with chromium tolerance and is transferable to related oxygen electrodes. Introduction Reversible solid oxide cells (RSOCs) are high-efficiency, fuel-flexible electrochemical devices that interconvert electricity and chemical fuels1,2. When operating in fuel-cell (FC) mode, RSOCs electrochemically oxidize H2 or carbon-containing fuels to generate electricity3. In electrolysis-cell (EC) mode, they store surplus electrical energy as H2 or synthetic fuels4. This operational flexibility positions RSOCs as key components for balancing temporal mismatches between renewable energy generation and demand5,6. In both modes, the oxygen electrode governs a large fraction of the losses and lifetime7. Modern deployment scenarios demand operation at high current density ( > 0.5 A·cm⁻2) for tens of thousands of hours under gas streams that often contain multiple reactive impurities8. These demanding conditions expose the oxygen electrode to volatile contaminants, particularly chromium (Cr) species from metallic interconnects, which accelerate structural degradation and interfacial deactivation9. Consequently, stability metrics under combined thermal, chemical, and electrochemical stresses remain insufficient for grid-scale RSOC systems. Cr-induced degradation in perovskite oxygen electrodes occurs through multiple coupled pathways. In benchmark compositions such as La0.6Sr0.4Co0.2Fe0.8O3₋δ (LSCF), volatile CrO3 species react with surface-segregated SrO to form insulating SrCrO4 precipitates10. This process causes pore blockage, reduces the active surface area, and contracts the triple-phase boundary region where the gas, electrode, and electrolyte phases meet. The resulting suppression of oxygen reduction and evolution kinetics drives rapid performance decay, with degradation rates often exceeding 1% per 100 hours under practical operating conditions. Current mitigation strategies include the use of protective coatings on metallic interconnects to suppress Cr volatilization11,12, the use of Cr getter layers to sequester volatile species13, and the development of intrinsically Cr-resistant electrode materials3. While each approach offers partial solutions, none adequately address the fundamental coupling between surface chemistry, bulk stability, and catalytic activity that governs long-term electrode performance. Recent advances in oxygen-electrode materials have explored three primary directions. Surface modification strategies employ protective layers such as BaO14, BaCoO315, or Gd0.1Pr0.1Ce0.8O2₋δ16 to physically or chemically inhibit Cr deposition. These approaches, however, introduce additional processing complexity and create new interfaces susceptible to degradation under thermal cycling. High-entropy oxide compositions, such as (La0.25Pr0.25Nd0.25Sm0.25)Ba0.5Sr0.5Co1.5Fe0.5O5+δ7, and SrCo0.5Fe0.2Ti0.1Ta0.1Nb0.1O3₋δ17, increase the configurational entropy to stabilize the lattice structure but often dilute the catalytically active sites, compromising the intrinsic activity. Exsolution engineering produces reactive nanodomains such as PrNi0.5Mn0.5O3–PrOₓ18 or Ba1₋xCe0.8Gd0.2O3₋δ–BaCO319 for in situ Cr trapping but typically lacks sufficient bulk stabilization and surface chemistry control. These single-mechanism strategies fail to address the interconnected nature of degradation pathways, limiting their effectiveness under RSOC-related stresses. Here, we introduce a triple-barrier design that integrates structural, interfacial, and chemical defenses within a single, scalable perovskite platform, as shown in Fig. 1. The rationale is to simultaneously address lattice instability, volatile Cr ingress, and acid–base imbalance, which are intrinsically coupled in conventional LSCF oxygen electrodes. Starting from the LSCF base composition, partial substitution of Sr to Ba produces La0.6Sr0.1Ba0.3Co0.2Fe0.8O3₋δ (LSB0.3CF), which suppresses oxide segregation and stabilizes the lattice, providing an initial structural safeguard against degradation. Further increasing the A-site Ba excess to surpass the lattice solubility limit (A/B ≈ 1.05) generates La0.6Sr0.1Ba0.35Co0.2Fe0.8O3₋δ (LSB0.35CF), facilitating thermodynamically driven in situ exsolution of BaCoO3 nanodomains. These exsolved phases preferentially capture Cr species and simultaneously limit contaminant ingress while promoting oxygen-surface kinetics, creating an external barrier. The final optimization involves trace Mo substitution at the B-site, yielding La0.6Sr0.1Ba0.35Co0.2Fe0.78Mo0.02O3₋δ (LSB0.35CFM0.02). We restrict the Mo doping level to 2 mol% because the strong surface enrichment allows for effective acidity tuning without higher concentrations compromising the bulk electrical conductivity20,21. This design provides a practical window where surface chemistry is modulated to neutralize the surface basicity responsible for Sr-chromate formation without sacrificing bulk transport. Consequently, this modification provides a reliable chemical barrier against acidic contaminants such as CrO3, complementing the structural and exsolved barriers. Together, these three complementary mechanisms form synergistic triple protection: lattice stabilization resists bulk degradation, exsolved-phase trapping intercepts Cr species at the surface, and acidity regulation mitigates the chemical driving forces for poisoning. A comprehensive comparison of this integrated strategy against conventional anti-Cr approaches, including interconnect coatings and single-mechanism electrode modifications, is summarized in Supplementary Table 2. As demonstrated below, this integrated strategy delivers improve oxygen exchange kinetics while maintaining the durability under severe Cr exposure. As a transferable multilevel design, the triple-barrier concept reconciles activity with Cr tolerance under RSOC-relevant loads and offers practical guidance for oxygen electrode engineering.Fig. 1Schematic illustrating the design concept of the La0.6Sr0.1Ba0.35Co0.2Fe0.78Mo0.02O3₋δ (LSB0.35CFM0.02) electrode.The graphical elements in Fig. 1 were created using Microsoft PowerPoint. Schematic illustrating the design concept of the La0.6Sr0.1Ba0.35Co0.2Fe0.78Mo0.02O3₋δ (LSB0.35CFM0.02) electrode. The graphical elements in Fig. 1 were created using Microsoft PowerPoint. Results Structure and surface exchange coefficient X-ray diffraction (XRD) analysis of the LSBxCFMy (x = 0, 0.3, 0.35; y = 0, 0.02) powders revealed perovskite structure formation in all the compositions (Fig. 2a). The diffraction peaks move systematically toward lower angles with higher Ba content, in line with lattice expansion arising from the larger radius of Ba2⁺ (1.61 Å) relative to that of Sr2⁺ (1.44 Å)22,23. The LSB0.35CF and LSB0.35CFM0.02 compositions exhibit an additional peak at ~31.5°, indicating BaCoO3 phase formation. We deliberately create a slight A-site Ba supersaturation so that a high density of BaCoO3 nanodomains precipitates at or near the surface. These domains are stable within the operating window and act as operationally active getters during Cr-contaminant exposure. Rietveld refinement of LSB0.35CFM0.02 (Fig. 2b, Supplementary Table 3) suggests a two-phase system composed of 97.48 wt% orthorhombic perovskite (space group Pbnm) and 2.52 wt% hexagonal BaCoO3 (space group P63/mmc). Transmission electron microscopy (TEM) revealed distinct lattice fringes at 3.90 Å and 4.90 Å, which were indexed to the (100) planes of LSB0.35CFM0.02 and BaCoO3, respectively (Fig. 2c)24. The phase boundary between these domains appears well defined, indicating a strong structural connection typical of exsolved phases. Energy-dispersive X-ray spectroscopy mapping reveals a homogeneous distribution of constituent elements (La, Sr, Ba, Co, Fe, Mo, and O) throughout the sample, with localized Ba–Co–O enrichment (Fig. 2d). These regions are indicative of Ba–Co–O enriched nanodomains, possibly related to a BaCoO3-like phase. The La/Fe signals likely stem from the matrix background due to the interaction volume of the electron probe exceeding the nanodomain scale. This suggests that these domains are not isolated pure phases but are coherently integrated with the host lattice. These observations are consistent with the presence of Ba–Co–O enriched regions, although the exact phase composition may involve mixed or heterogeneous local structures. High-temperature XRD analysis suggests the phase stability from ambient temperature to 1000 °C (Supplementary Fig. 1).Fig. 2Structural characterization and oxygen surface exchange properties of oxygen electrodes.a XRD pattern of LSBxCFMy (x = 0, 0.3, 0.35; y = 0, 0.02) powders. b Rietveld refinement profiles c TEM micrograph and d EDX mapping of the LSB0.35CFM0.02 powder. Normalized conductivity relaxation curves for LSBxCFMy at 750 °C e before and f after Cr poisoning. g Comparison of oxygen surface-exchange coefficients (kchem, bars, left axis) and the corresponding retention ratios (kret, symbols, right axis) before and after Cr poisoning; error bars denote the standard deviations (n = 3). Retention values (kret) are shown alongside absolute k values to clearly distinguish relative durability from initial activity. Source data for Fig. 2 are provided as a Source Data file. Structural characterization and oxygen surface exchange properties of oxygen electrodes. a XRD pattern of LSBxCFMy (x = 0, 0.3, 0.35; y = 0, 0.02) powders. b Rietveld refinement profiles c TEM micrograph and d EDX mapping of the LSB0.35CFM0.02 powder. Normalized conductivity relaxation curves for LSBxCFMy at 750 °C e before and f after Cr poisoning. g Comparison of oxygen surface-exchange coefficients (kchem, bars, left axis) and the corresponding retention ratios (kret, symbols, right axis) before and after Cr poisoning; error bars denote the standard deviations (n = 3). Retention values (kret) are shown alongside absolute k values to clearly distinguish relative durability from initial activity. Source data for Fig. 2 are provided as a Source Data file. Electrical conductivity relaxation (ECR) measurements provide a direct assessment of oxygen surface exchange kinetics25. The normalized conductivity transients for the LSBxCFMy (x = 0, 0.3, 0.35; y = 0, 0.02) compositions following a pO2 step change from 0.10 to 0.21 atm at 750 °C are shown in Fig. 2e. The extracted surface exchange coefficient (kchem) for the LSCF is 6.78 × 10⁻4 cm·s⁻1, which is consistent with the literature values26. Progressive Ba and Mo incorporation systematically enhances kchem, with LSB0.35CFM0.02 reaching 2.01 × 10⁻3 cm·s⁻1. This value shows competitive performance against those of comparable perovskite electrodes27–30. To evaluate Cr tolerance, we repeated the ECR measurements following exposure to Cr2O3 at 750 °C for 120 h (Fig. 2f, g). All the compositions exhibit prolonged relaxation times after Cr2O3 exposure, indicating degraded surface kinetics. This degradation stems from SrCrO4/BaCrO4 formation through reactions between volatile CrO3 and segregated alkaline earth oxides, which block active sites and typically reduce k by an order of magnitude19. Despite the widespread degradation after Cr2O3 exposure, quantitative analysis revealed distinct durability profiles. As plotted in Fig. 2g (right axis), compared to LSCF (kret ≈ 0.071), LSB0.35CFM0.02 has a higher retention ratio (kret = kafter/kbefore ≈ 0.105). Consequently, the modified LSB0.35CFM0.02 maintains improved retention of surface exchange kinetics, kafter = 2.10 × 10−4 cm s−1, approximately 4.4 times higher than that of the LSCF (4.79 × 10⁻5 cm·s⁻¹) under the same conditions (mean ± s.d., n = 3). This sustained absolute activity of LSB0.35CFM0.02, rather than the absolute degradation magnitude, supports the effectiveness of the optimized composition against Cr poisoning. Activity and Cr tolerance of the oxygen electrodes To elucidate the oxygen reduction reaction (ORR), the electrochemical impedance spectroscopy (EIS) results of LSB0.35CFM0.02 (Fig. 3a) at various pO2 values were analyzed using the distribution of the relaxation time (DRT). The resulting curves (Supplementary Fig. 2) reveal three distinct processes categorized as low-frequency (LF), mid-frequency (MF), and high-frequency (HF) features. Their possible physicochemical origins were evaluated using characteristic capacitance and pO2-dependent reaction order (n, Rp = k(pO2)n31, Fig. 3b) as commonly adopted descriptors. Detailed parameter ranges and evaluation rationale are provided in “Methods” section and Supplementary Table 1. These responses are consistent with surface- or interfacial-related coupled processes32–37 within commonly used capacitance-based categorizations. The assignments are intended for trend comparison rather than unique mechanistic identification. The sum of the circuit resistances agrees with the DRT integral, with a relative error of <3%, indicating internal consistency of the fit within the adopted model framework, while noting that EIS-based interpretations are inherently non-unique.Fig. 3Electrochemical activity and Cr tolerance of the oxygen electrodes.a EIS curves of half-cells with LSB0.35CFM0.02 under different pO2 values. Arrhenius plot of (b) RLF, RMF, and RHF versus pO2 for LSB0.35CFM0.02 and c Rp versus T for LSBxCFMy (x = 0, 0.3, 0.35; y = 0, 0.02). The reported apparent activation energy (Ea) values are presented as means ± s.d. (n = 3). d, e Rp evolution of half-cells with LSBxCFMy at 750 °C and an open circuit voltage (OCV)/0.5 A·cm⁻2 with Cr2O3 exposure. f–j EIS and h, k DRT curves of half-cells with LSCF/LSB0.35CFM0.02 at 750 °C and an OCV/0.5 A·cm⁻2 with Cr2O3 exposure. The impedance spectra were deconvoluted using DRT with Tikhonov regularization, where the regularization parameter was optimized using the L-curve or generalized cross-validation. Data validation was performed by fitting representative spectra to the equivalent circuit model (L-Ro−3(R-CPE)) shown in the inset (a). Note that the ohmic resistance (Ro) has been subtracted from the Nyquist plots for visual clarity. The @Cr symbol refers to a sample after Cr2O3 exposure. The structure of the symmetrical half-cell is LSB0.35CFM0.02/GDC/SSZ/GDC/LSB0.35CFM0.02. Source data for Fig. 3 are provided as a Source Data file. Electrochemical activity and Cr tolerance of the oxygen electrodes. a EIS curves of half-cells with LSB0.35CFM0.02 under different pO2 values. Arrhenius plot of (b) RLF, RMF, and RHF versus pO2 for LSB0.35CFM0.02 and c Rp versus T for LSBxCFMy (x = 0, 0.3, 0.35; y = 0, 0.02). The reported apparent activation energy (Ea) values are presented as means ± s.d. (n = 3). d, e Rp evolution of half-cells with LSBxCFMy at 750 °C and an open circuit voltage (OCV)/0.5 A·cm⁻2 with Cr2O3 exposure. f–j EIS and h, k DRT curves of half-cells with LSCF/LSB0.35CFM0.02 at 750 °C and an OCV/0.5 A·cm⁻2 with Cr2O3 exposure. The impedance spectra were deconvoluted using DRT with Tikhonov regularization, where the regularization parameter was optimized using the L-curve or generalized cross-validation. Data validation was performed by fitting representative spectra to the equivalent circuit model (L-Ro−3(R-CPE)) shown in the inset (a). Note that the ohmic resistance (Ro) has been subtracted from the Nyquist plots for visual clarity. The @Cr symbol refers to a sample after Cr2O3 exposure. The structure of the symmetrical half-cell is LSB0.35CFM0.02/GDC/SSZ/GDC/LSB0.35CFM0.02. Source data for Fig. 3 are provided as a Source Data file. Supplementary Figure 3 presents the EIS plots of half-cells at 650–800 °C corrected for ohmic resistance (Ro). The baseline LSCF exhibits an Rp slightly higher than some reported values37,38, which is likely attributed to microstructural variations from the screen-printing process. However, the relative enhancement achieved by our modified electrode is substantial. Rp greatly decreases with increasing Ba/Mo doping, with LSB0.35CFM0.02 manifesting the lowest Rp (0.058 Ω·cm2) and a 70.4% reduction versus the LSCF (0.196 Ω·cm2) at 750 °C. (Fig. 3c) The activation energy (Ea) decreases to its lowest value of 1.567 eV for LSB0.35CFM0.02, aligning with the reduced polarization resistance. The ~0.1 eV reduction in Ea corresponds to the nearly 3.1-fold increase in the intrinsic exchange rate at 750 °C, matching the 2.96-fold enhancement of the surface exchange coefficient in Fig. 2e. The improvement is ascribed to the synergistic interplay of multiphase engineering and cation doping. First, for LSB0.3CF, the rational incorporation of Ba is associated with improved oxygen reaction kinetics, consistent with a reduced LF contribution in DRT curves39 (Supplementary Fig. 4). Second, in LSB0.35CF, the A-site Ba excess triggers the exsolution of BaCoO3 nanodomains. These domains not only provide excellent oxygen adsorption/dissociation capability15,40 but also multiply the active sites by forming abundant heterointerfaces, thereby potentially enhancing surface exchange activity41. Finally, the Mo-doping in LSB0.35CFM0.02 further modulates the B-site electronic structure within the Co–O–Fe/Mo framework. This substitution facilitates electron polaron hopping and lowers the energy barrier for the rate-determining step, leading to the fastest surface exchange kinetics among all the compositions42,43. The Rp evolution of half-cells with LSBxCFMy (x = 0, 0.3, 0.35; y = 0, 0.02) at 750 °C and OCV with Cr2O3 exposure are shown in Fig. 3d. The baseline LSCF decreases by 91.3%, with Rp increasing from 0.196 to 0.375 Ω·cm2 within 120 h (Fig. 3f). This deterioration is consistent with chromate-related species formation (e.g., SrCrO4), associated with reactions between volatile CrOx and surface-enriched SrO. By substituting Sr with Ba, LSB0.3CF establishes the first structural barrier, demonstrating a mitigated degradation of 63.6% (Rp from 0.088 to 0.144 Ω·cm2; Supplementary Fig. 5). This improvement arises from the larger radius of Ba2+ (1.61 Å vs. 1.44 Å for Sr2+)22,23, which increases the ionic radius and increases the Goldschmidt tolerance factor44. A higher tolerance factor theoretically drives the perovskite lattice toward ideal cubic symmetry, thus minimizing local strain and stabilizing A–O coordination39,45. To further assess this structural stability, we performed ab initio molecular dynamics (AIMD) simulations at 750 °C. The AIMD results quantitatively support improved stability under the simulated conditions: (Fig. 4a) the LSB0.3CF model exhibits limited atomic displacement with a root-mean-square deviation (RMSD) of <2.2 Å and fast energy convergence ( ± 0.06 eV per atom) after 10 ps. In contrast, the LSCF model shows substantial structural distortion (RMSD > 3.2 Å) and larger energy fluctuations ( ± 0.12 eV per atom) after 10 ps (Supplementary Fig. 6). These results suggest Ba-induced lattice stabilization as the first structural barrier against Cr-induced degradation. We evaluated the intrinsic surface stability of the Ba/Mo-containing electrode in Cr-free air at 750 °C (Supplementary Fig. 7). LSCF exhibited a ~ 0.02 Ω·cm2 increase over 200 h, which is commonly associated with spontaneous SrO segregation10, whereas LSB0.35CFM0.02 remained stable. This behavior is consistent with suppressed A-site surface enrichment enabled by bulk lattice stabilization.Fig. 4Mechanistic insights into the structural and chemical barriers of oxygen electrodes.a Snapshots of LSB0.3CF along the AIMD trajectory up to 10 ps at 750 °C. b Acidity scale of relevant oxides49. c CO2-TPD and d NH3-TPD profiles of LSCF, LSB0.35CF and LSB0.35CFM0.02. Source data for (a, c, d) are available in Supplementary Data 1 and the Source Data file, respectively. Mechanistic insights into the structural and chemical barriers of oxygen electrodes. a Snapshots of LSB0.3CF along the AIMD trajectory up to 10 ps at 750 °C. b Acidity scale of relevant oxides49. c CO2-TPD and d NH3-TPD profiles of LSCF, LSB0.35CF and LSB0.35CFM0.02. Source data for (a, c, d) are available in Supplementary Data 1 and the Source Data file, respectively. Note: CO2-TPD and NH3-TPD profiles were baseline corrected, mass normalized and integrated over predefined temperature windows ( < 200 °C and 200–500 °C) under identical ramp rates and flow conditions. a CO2-TPD: The amount of chemisorbed CO2 corresponds to the decomposition of surface carbonates at basic sites46; data are presented as the means ± s.d. (n = 3). b NH3-TPD: NH3 desorption in the weakly acidic region corresponds to coordination with surface Lewis acid centers (exposed transition metal cations)47,48; data are presented as the means ± s.d. (n = 3). Source data for Table 1 are provided as a Source Data file.Table 1Quantitative analysis of surface adsorption properties derived from TPD profilesSampleCO2 Desorption(μmol·g−1)NH3 Desorption(μmol·g−1)Physisorbed( < 200 °C)Chemisorbeda (200–500 °C)Weak Acidb( < 200 °C)Strong Acid(200–500 °C)LSCF9.627.521.250.0LSB0.35CF10.255.718.446.2LSB0.35CFM0.0212.99.465.113.3 Table 1Quantitative analysis of surface adsorption properties derived from TPD profilesSampleCO2 Desorption(μmol·g−1)NH3 Desorption(μmol·g−1)Physisorbed( < 200 °C)Chemisorbeda (200–500 °C)Weak Acidb( < 200 °C)Strong Acid(200–500 °C)LSCF9.627.521.250.0LSB0.35CF10.255.718.446.2LSB0.35CFM0.0212.99.465.113.3 Quantitative analysis of surface adsorption properties derived from TPD profiles Further increasing the A-site Ba excess in LSB0.35CF activates the exsolved-phase trapping barrier, leading to a much more stable performance with only a 20.8% increase in Rp (from 0.072 to 0.087 Ω·cm2; Supplementary Fig. 8). This is attributed to the in situ exsolution of BaCoO3 nanodomains, which preferentially capture Cr species (Supplementary Table 4) and likely impede their inward diffusion. Finally, LSB0.35CFM0.02 achieves the stability and displays almost no change in Rp (0.058 − 0.060 Ω·cm2; Fig. 3g). This high durability is unlocked by the Mo-induced chemical barrier, which modulates the surface acid–base equilibrium to suppress the driving force for Cr poisoning (Fig. 4b)49. Despite the low nominal doping level (2 mol%), Mo strongly affects the surface chemistry. This is attributed to the surface enrichment of high-valence Mo6+42,50 and its strong electronic inductive effect51. Specifically, the Mo–O interaction downshifts the oxygen 2p band center (from −2.50 eV in LSCF to −3.18 eV in LSBCFM, Supplementary Fig. 9) and reduces the electron density on the surface oxygen (the Bader charge shifts from −1.25 |e| to −1.13 |e | ), which effectively lowers the basicity of neighboring lattice sites and mitigates the affinity for acidic Cr species. Since volatile Cr species (e.g., CrO3) act as strong gaseous acids, their deposition mechanism is chemically analogous to the adsorption of CO2 (an acidic probe). Quantitative TPD analysis (Fig. 4c, d and Table 1) reveals that Ba doping increases the carbonate-related CO2 integral (200–500 °C) from 27.5 (LSCF) to 55.7 μmol·g−1 (LSB0.35CF), indicating stronger basicity and a greater likelihood of carbonate formation46. Introducing 2 mol% Mo reduces the CO2 integral to 9.4 μmol·g−1 while increasing the NH3 weak-acid window ( < 200 °C) from 18.4 (LSB0.35CF) to 65.1 μmol·g−1 (LSB0.35CFM0.02), which is consistent with decreased basicity and increased Lewis acidity at the surface47. This drastic suppression of carbonation implies a dual functional benefit: it not only mitigates chemically analogous CrO3 poisoning but also resolves the intrinsic vulnerability of Ba-based electrodes to CO2 carbonation, as supported by stable operation for 50 h under 1% CO2 at 750 °C and OCV (Supplementary Fig. 10). Crucially, these acid–base shifts align with the temperature- and pressure-corrected adsorption free energies for acidic Cr species and with the improved retention after Cr2O3 exposure, suggesting that surface chemistry, rather than geometric area alone, governs durability. The Rp variation of half-cells with LSCF/LSB0.35CFM0.02 at 750 °C and 0.5 A·cm⁻2 current density with Cr2O3 exposure for 1000 h is compared in Fig. 3e. Under high current loading, the LSCF decreases rapidly (1.45% h⁻1), whereas LSB0.35CFM0.02 exhibits a negligible Rp increase with the rate of 0.25% h⁻1, indicating enhanced Cr tolerance under polarization. To clarify the electrochemical-driven degradation mechanism of Cr poisoning, DRT analysis was performed (Fig. 3h, k). Notably, under OCV conditions, the LF peak increases more significantly than the MF/HF peaks, whereas under a 0.5 A·cm⁻2 load, the increase in the HF peak is more pronounced. This suggests that, in addition to oxygen adsorption and transport, the interfacial reactions between the electrode and electrolyte become increasingly vulnerable under the current load36,37. The pronounced growth of the HF peak indicates aggravated interfacial charge-transfer resistance. Under cathodic polarization, the buried interface experiences a decrease in local pO2 and an increase in oxygen vacancy concentration52. This drives A-site cation migration toward the interface, facilitating the capture of volatile Cr species53. A Cr-enriched region is observed near the electrode–electrolyte interface under polarization, consistent with an increase in the high-frequency contribution observed in EIS. As shown by SEM-EDS line scans (Supplementary Fig. 11c, d), the LSCF cell exhibits prominent Cr concentration near the electrode–electrolyte interface, identified as fine SrCrO4 deposits by HR-TEM (Supplementary Fig. 11e, f)54. These observations support a working model that Cr ingress contributes to interfacial deactivation under polarization. In contrast, the LSB0.35CFM0.02 cell shows suppressed Cr signal across the entire cross-section. Characterizations of Cr tolerance The morphological evolution of the electrodes provides the first evidence for enhanced stability. As revealed by SEM, both the fresh LSCF and LSB0.35CFM0.02 electrodes initially exhibit porous, interconnected skeletons (Supplementary Fig. 12). However, after Cr2O3 exposure at 750 °C and OCV for 120 h, a striking difference in both electrodes emerges. The LSCF surface is severely compromised by coarse and irregular deposits (Fig. 5a), identified as SrCrO4 originating from the reaction between segregated SrO and volatile CrO326. Conversely, the LSB0.35CFM0.02 surface remains clean and structurally intact (Fig. 5b). To decipher the mechanism behind this protection, we investigated the phase interactions. Thermodynamically, the BaO + CrO3 → BaCrO4 reaction is more favorable than the SrO + CrO3 → SrCrO4 reaction (e.g., −431.7 versus −368.8 kJ mol⁻1 at 750 °C, Supplementary Table 4), which indicates a stronger drive for BaCoO3 nanodomains to capture the Cr species first and limits Cr ingress into the perovskite lattice. Electrically, the resulting BaCrO4 product is less resistive than SrCrO4 (4.5 × 10⁻3 S cm⁻1 versus 1.8 × 10⁻4 S cm⁻1 at 800 °C)14, helping preserve oxygen surface exchange and charge transfer within the oxygen electrode. The XRD pattern of the sintered LSB0.35CFM0.02–Cr2O3 mixture (Supplementary Fig. 13) clearly exhibits distinct diffraction peaks corresponding to BaCrO4, suggesting that volatile Cr species are intercepted by the Ba-rich domains before they attack the perovskite backbone. This trapping effect is further corroborated by Raman spectroscopy and mapping (Fig. 5c–e). The posttest LSCF spectrum shows intense peaks at 850–990 cm⁻¹ (SrCrO4)55 and 340–430 cm⁻¹ (Cr2O3)7, accompanied by distinct Co3O4 peaks (600–700 cm⁻¹)56,57 that signal severe B-site segregation concomitant with lattice decomposition. In contrast, LSB0.35CFM0.02 displays suppressed peak intensities for Cr-related species. More importantly, the absence of Co3O4 signals indicates that the perovskite backbone remains structurally intact under Cr attack. The Raman mappings provide spatial visualization of the degradation. While LSCF exhibits widespread and intense CrO42⁻ signals across the surface (Fig. 5d), signifying extensive poisoning and that the signal for LSB0.35CFM0.02 (Fig. 5e) is weak and spatially localized. Crucially, these detected chromate bands colocalize with the Ba-rich nanodomains, consistent with active trapping that limits Cr penetration under operating conditions.Fig. 5Microstructural and surface chemical evolution of the oxygen electrodes after Cr exposure.SEM micrograph of the (a) LSCF and b LSB0.35CFM0.02 electrode (a, b) after Cr2O3 exposure at 750 °C and OCV for 120 h. c Raman spectra of LSCF/LSB0.35CFM0.02 after Cr2O3 exposure at 750 °C and OCV for 120 h and d, e corresponding Raman intensity mappings of the chromate (CrO42⁻) peak at 859 cm⁻1 over a 20 × 20 μm2 area. XPS spectra of LSCF/LSB0.35CFM0.02 before and after Cr2O3 exposure at 750 °C and OCV for 120 h: f, g Sr and i, j Cr. h Quantitative evolution of the degree of Sr segregation (Srsurf/Srlat) of LSCF/LSB0.35CF/LSB0.35CFM0.02 before and after the test; data are presented as the means ± s.d. (n = 3). The light-green shaded region denotes the stable regime threshold of Sr segregation ratio. k Atomic concentration of accumulated Cr species on the after-test LSCF/LSB0.35CF/LSB0.35CFM0.02 electrode surfaces; data are presented as the means ± s.d. (n = 3). The peak fits used constrained doublets as described in the “Methods” section. The O 1 s spectra are presented in Supplementary Fig. 15 only and are not used for quantitative interpretation. The @Cr symbol refers to a sample after Cr2O3 exposure. Source data for Fig. 5 are provided as a Source Data file. Microstructural and surface chemical evolution of the oxygen electrodes after Cr exposure. SEM micrograph of the (a) LSCF and b LSB0.35CFM0.02 electrode (a, b) after Cr2O3 exposure at 750 °C and OCV for 120 h. c Raman spectra of LSCF/LSB0.35CFM0.02 after Cr2O3 exposure at 750 °C and OCV for 120 h and d, e corresponding Raman intensity mappings of the chromate (CrO42⁻) peak at 859 cm⁻1 over a 20 × 20 μm2 area. XPS spectra of LSCF/LSB0.35CFM0.02 before and after Cr2O3 exposure at 750 °C and OCV for 120 h: f, g Sr and i, j Cr. h Quantitative evolution of the degree of Sr segregation (Srsurf/Srlat) of LSCF/LSB0.35CF/LSB0.35CFM0.02 before and after the test; data are presented as the means ± s.d. (n = 3). The light-green shaded region denotes the stable regime threshold of Sr segregation ratio. k Atomic concentration of accumulated Cr species on the after-test LSCF/LSB0.35CF/LSB0.35CFM0.02 electrode surfaces; data are presented as the means ± s.d. (n = 3). The peak fits used constrained doublets as described in the “Methods” section. The O 1 s spectra are presented in Supplementary Fig. 15 only and are not used for quantitative interpretation. The @Cr symbol refers to a sample after Cr2O3 exposure. Source data for Fig. 5 are provided as a Source Data file. Finally, X-ray photoelectron spectroscopy (XPS) was employed to quantify the surface chemical evolution (Fig. 5f–k). With respect to the LSCF (Fig. 5f), Cr exposure triggered a substantial increase in the surface Sr component (Srsurf: 134.2/135.5 eV)58, indicating massive SrO segregation. This is mirrored by the dominant Cr⁶⁺ signals (578.7/588.3 eV)59 characteristic of chromates (Fig. 5i). In the case of LSB0.35CFM0.02 (Fig. 5g), although an increase in Srsurf is observed, the degree of segregation is strictly limited. XPS was used to quantify Sr segregation by the Sr 3 d surface-to-lattice ratio (Srsurf/Srlat), which was obtained from constrained doublet fits (reported as the mean ± s.d., n = 3). Quantitative analysis revealed a striking contrast (Fig. 5h and Supplementary Table 5): after testing, the Srsurf/Srlat ratio for LSCF markedly increased from 0.65 to 1.70, indicating severe surface reconstruction. To further isolate the role of Mo doping, we compared the surface evolution of LSB0.35CF and LSB0.35CFM0.02 before and after Cr exposure. While both samples show similarly low initial Srsurf/Srlat ratios ( ~ 0.45–0.48), consistent with the stabilizing effect of Ba substitution. However, after Cr exposure, LSB0.35CF shows a clear increase in surface Sr segregation (0.77), whereas Mo-doped LSB0.35CFM0.02 remains nearly unchanged (0.56). Consequently, the surface chromium fraction after exposure is markedly reduced on the optimized electrode (Fig. 5k). The atomic concentration of detrimental Cr species is as high as 27.43 at.% (dominated by Cr⁶⁺) for LSCF and 12.49 at.% for LSB0.35CF, whereas it is drastically reduced to 3.51 at.% on LSB0.35CFM0.02. This reduced surface Cr accumulation is consistent with the smaller resistance increase observed in EIS. This controlled comparison indicates that the improved Cr tolerance of LSB0.35CFM0.02 does not arise solely from intrinsically reduced A-site surface availability, but also from an additional Mo-related suppression of Cr-induced surface segregation. These results support a synergistic mechanism involving Ba-enabled lattice stabilization and Mo-enabled surface protection against Cr-containing species. We do not use the O 1 s Osurf as a descriptor because of its sensitivity to adsorbed water60. The O 1 s spectra are provided in Supplementary Fig. 15 for completeness. Performance and durability of RSOC The single-cell architecture (Ni-YSZ/Ni-SSZ/SSZ/GDC/LSB0.35CFM0.02) with an oxygen electrode thickness of ~18 μm is shown in Fig. 6a. As shown in Fig. 6b and Supplementary Fig. 16, the LSB0.35CFM0.02 cell delivers enhanced peak power densities of 0.598–1.352 W·cm⁻2 at 700–800 °C, exhibiting competitive performance relative to the LSCF cell (0.248–0.673 W·cm⁻2). In the EC mode, the cell with LSB0.35CFM0.02 also exhibited enhanced catalytic activity compared with that with LSCF (Fig. 6c and Supplementary Fig. 17). Under a voltage of 1.5 V with 40% H2O–60% H2 fuel at 800 °C, the LSB0.35CFM0.02 cell achieves a current density of 2.08 A·cm⁻2, which is a 39.6% increase over that of the LSCF cell (1.49 A·cm⁻2). Durability tests against Cr poisoning were first conducted in FC mode at 750 °C and 0.5 A·cm⁻2 for 500 h (Fig. 6d, e and Supplementary Fig. 18). The Ro of the LSCF cell increases at 0.158% h⁻1, whereas that of the LSB0.35CFM0.02 cell remains virtually constant (Supplementary Fig. 19). More critically, the Rp of the LSCF cell increases rapidly at 0.781% h⁻1, whereas the LSB0.35CFM0.02 cell shows only a negligible increase of 0.029% h⁻1 (Supplementary Fig. 20). A detailed description of the impedance features, supported by characteristic capacitance analysis and equivalent circuit validation, is provided in “Methods” section and Supplementary Table 6. This analysis identifies the mid-frequency P3 and high-frequency P4 peaks as the oxygen surface exchange and bulk ion transport processes, respectively. Consequently, as shown in Supplementary Fig. 18b, the marked growth of these specific peaks in the postexposure DRT curves indicates that Cr poisoning severely deteriorates the surface exchange and bulk transport kinetics within the oxygen electrode61. This degradation is consistent with enhanced A-site surface enrichment and the formation of chromate-related species under Cr2O3 exposure. Such Cr-containing deposits can reduce accessible active surface area and introduce additional transport constraints62,63, in agreement with the observed deterioration in ORR performance. Furthermore, the LSB0.35CFM0.02 cell indicates high robustness under long-term Cr2O3 exposure (Fig. 6g). Under a 0.5 A·cm⁻2 load, both cells initially present a voltage of ~0.95 V. Upon Cr2O3 injection, the LSCF cell voltage decreases to 0.53 V, whereas the LSB0.35CFM0.02 cell maintains a stable output at 0.95 V. Finally, the RSOC with LSB0.35CFM0.02 demonstrates excellent reversibility and stability at ±0.5 A·cm⁻2 for 1000 h. To elucidate the microstructural origin of this stability, time-of-flight secondary-ion mass spectrometry (ToF-SIMS) was then performed. The depth profiles (Fig. 6f) reveal severe and homogeneous CrO4– permeation throughout the LSCF cell. In sharp contrast, only weak and superficial Cr signals are detected in LSB0.35CFM0.02 cells. This is consistent with the engineered triple-barrier, which effectively intercepts Cr species at the surface, preventing deep intrusion. This strategic mechanism is schematically illustrated in Fig. 7: the synergy of lattice stabilization, exsolved-phase trapping, and acidity regulation successfully reconciles high electrochemical activity with Cr tolerance, providing a general design paradigm for designing durable oxygen electrodes under RSOC-relevant loads.Fig. 6Performance and durability of RSOC under Cr exposure.a SEM micrograph of a single cell for tests. I–V plots of LSB0.35CFM0.02 cells in (b) FC mode and c EC mode at 700–800 °C. d EIS and e DRT curves of the SOFCs with LSCF/LSB0.35CFM0.02 at 750 °C and 0.5 A·cm⁻2 under Cr2O3 exposure. Data validation was performed by fitting representative spectra to the equivalent circuit model (L-Ro-5(R-CPE)) shown in the inset (d). f ToF-SIMS depth profiles comparing the normalized chromium intensities (CrO4–/LaO–) versus the equivalent depth for the LSCF (after 500 h) and LSB0.35CFM0.02 (after 1000 h) cells. g voltage evolution of the RSOCs with LSCF/LSB0.35CFM0.02 at 750 °C and ±0.5 A·cm⁻2 under Cr2O3 exposure. All presented I–V curves and cell voltages are reported without iR correction. The data are presented as the means ± s.d. (n = 3). The @Cr symbol refers to a sample after Cr2O3 exposure. The structure of the anode-supported single cell is Ni-YSZ/Ni-SSZ/SSZ/GDC/LSB0.35CFM0.02. Source data for Fig. 6 are provided as a Source Data file.Fig. 7Schematic of the triple-barrier mechanism against Cr corrosion for the RSOC oxygen electrode.(1) Inner layer: Lattice stabilization via Ba doping, verified by HT-XRD for phase stability and supported by AIMD/DFT simulations. (2) Middle layer: Acidity regulation via Mo doping, verified by CO2/NH3-TPD shifts in probe uptake and XPS surface speciation. (3) Outer layer: Exsolved-phase trapping via BaCoO3 nanodomains, verified by XRD phase signatures, Raman mapping of colocalized chromates, ToF-SIMS depth profiling, and thermodynamic calculations. The graphical elements in Fig. 7 were created using Microsoft PowerPoint and Adobe Illustrator. Performance and durability of RSOC under Cr exposure. a SEM micrograph of a single cell for tests. I–V plots of LSB0.35CFM0.02 cells in (b) FC mode and c EC mode at 700–800 °C. d EIS and e DRT curves of the SOFCs with LSCF/LSB0.35CFM0.02 at 750 °C and 0.5 A·cm⁻2 under Cr2O3 exposure. Data validation was performed by fitting representative spectra to the equivalent circuit model (L-Ro-5(R-CPE)) shown in the inset (d). f ToF-SIMS depth profiles comparing the normalized chromium intensities (CrO4–/LaO–) versus the equivalent depth for the LSCF (after 500 h) and LSB0.35CFM0.02 (after 1000 h) cells. g voltage evolution of the RSOCs with LSCF/LSB0.35CFM0.02 at 750 °C and ±0.5 A·cm⁻2 under Cr2O3 exposure. All presented I–V curves and cell voltages are reported without iR correction. The data are presented as the means ± s.d. (n = 3). The @Cr symbol refers to a sample after Cr2O3 exposure. The structure of the anode-supported single cell is Ni-YSZ/Ni-SSZ/SSZ/GDC/LSB0.35CFM0.02. Source data for Fig. 6 are provided as a Source Data file. Schematic of the triple-barrier mechanism against Cr corrosion for the RSOC oxygen electrode. (1) Inner layer: Lattice stabilization via Ba doping, verified by HT-XRD for phase stability and supported by AIMD/DFT simulations. (2) Middle layer: Acidity regulation via Mo doping, verified by CO2/NH3-TPD shifts in probe uptake and XPS surface speciation. (3) Outer layer: Exsolved-phase trapping via BaCoO3 nanodomains, verified by XRD phase signatures, Raman mapping of colocalized chromates, ToF-SIMS depth profiling, and thermodynamic calculations. The graphical elements in Fig. 7 were created using Microsoft PowerPoint and Adobe Illustrator. Origin of enhanced Cr tolerance in Ba/Mo-codoped LSCF electrodes Density functional theory (DFT) calculations were performed to elucidate the origin of the ability of the Ba/Mo-codoped LSCF to enhance the resistance to Cr erosion. The adsorption process of CrO3 on the LSCF, LSBCF, and LSBCFM surfaces was investigated, with the resulting configurations at various sites shown in Fig. 8. When CrO3 is adsorbed on the three surfaces, Cr and O in CrO3 tend to bond with O and M (La/Sr/Ba) sites on the surface, respectively. The adsorbed Gibbs free energies (Gads) of CrO3 at different sites on the three surfaces are summarized in Table 2. Specifically, on the LSCF and LSBCF surfaces, the adsorption configurations of CrO3 were optimized at all possible sites, namely, La–O–La, Sr–O–Sr, La–O–Sr, and La–O–Ba, and the corresponding adsorption energies were calculated. Energetically, CrO3 preferentially adsorbs at the La–O–La and Sr–O–Sr sites of the LSCF and LSBCF surfaces, with Gads values of –3.62 and –3.74 eV, respectively. On the LSBCFM surface, similar to LSBCF, the adsorption of CrO3 has also been attempted at four sites (La–O–La, Sr–O–Sr, La–O–Sr, and La–O–Ba). After structural optimization, the adsorption configurations at La–O–La, Sr–O–Sr, and La–O–Sr converged to the same stable structure, in which the three oxygen atoms of CrO3 bind to one La and two Sr, with a distance of 2.72 Å and a Gads of –2.15 eV. Moreover, the Gads of CrO3 at the La–O–Ba site on the LSBCFM surface is –3.58 eV. The computational analysis at 750 °C suggests a trend toward reduced CrO3 affinity, noting the typical uncertainty associated with calculated adsorption energies, and the corresponding relative adsorption rate constants are provided in Supplementary Table 7.Fig. 8Optimized configurations of key adsorption structures.Detailed illustration of CrO3 adsorption on different surface sites of LSCF (a–c: La–O–La, Sr–O–Sr, La–O–Sr), LSBCF (d–g: La–O–La, Sr–O–Sr, La–O–Sr, La–O–Ba), and LSBCFM (h, i: La–O–Sr, La–O–Ba) surfaces. Source data for Fig. 8 are provided as a Supplementary Data 1 file.Table 2Summarized adsorption energies of CrO3 on various sites on the LSCF, LSBCF, and LSBCFM surfacesGads (eV)La–O–LaSr–O–SrLa–O–SrLa–O–BaLSCF–3.62–2.56–3.28/LSBCF–3.72–3.74–2.28–3.58LSBCFM//–2.15–3.58 Optimized configurations of key adsorption structures. Detailed illustration of CrO3 adsorption on different surface sites of LSCF (a–c: La–O–La, Sr–O–Sr, La–O–Sr), LSBCF (d–g: La–O–La, Sr–O–Sr, La–O–Sr, La–O–Ba), and LSBCFM (h, i: La–O–Sr, La–O–Ba) surfaces. Source data for Fig. 8 are provided as a Supplementary Data 1 file. Table 2Summarized adsorption energies of CrO3 on various sites on the LSCF, LSBCF, and LSBCFM surfacesGads (eV)La–O–LaSr–O–SrLa–O–SrLa–O–BaLSCF–3.62–2.56–3.28/LSBCF–3.72–3.74–2.28–3.58LSBCFM//–2.15–3.58 Summarized adsorption energies of CrO3 on various sites on the LSCF, LSBCF, and LSBCFM surfaces Overall, Ba/Mo codoping is associated with weaker CrO3 adsorption at selected sites in our calculations, consistent with experimentally observed mitigation of Cr-related degradation. The Bader charge was analyzed to elucidate the mechanism through which Ba/Mo codoping weakens CrO3 adsorption. In all the cases, net charge transfer occurs from the perovskite surface to the CrO3 species during adsorption. As illustrated in Supplementary Fig. 21, substantial electron accumulation is observed between Cr in CrO3 and the surface oxygen sites, as well as between O in CrO3 and La/Sr sites. This redistribution of electron density corresponds to chemical bonds between CrO3 and the surface upon adsorption. In summary, Ba/Mo codoping effectively reduces the adsorption strength of CrO3 on the LSBCFM surface, particularly at the La–O–Sr sites. This trend can be rationalized by reduced electron density at oxygen sites within La–O–Sr motifs, which may weaken interactions with CrOx species. Discussion In this work, we establish a multiscale design principle for overcoming the fundamental trade-off between catalytic activity and chemical stability in SOC electrodes. Through systematic integration of lattice engineering, interfacial modification, and surface chemistry control, we developed La0.6Sr0.1Ba0.35Co0.2Fe0.78Mo0.02O3₋δ as a high-performance oxygen electrode with exceptional chromium tolerance. The material achieves an Rp of 0.058 Ω·cm2 at 750 °C, 70.4% lower than that of standard LSCF, while maintaining degradation rates below 0.25% h⁻¹ under a 0.5 A·cm⁻2 load with Cr2O3 exposure for 1000 h. Three complementary mechanisms contribute to the enhanced Cr resistance. First, Ba substitution at the A site increases the tolerance factor toward unity, stabilizing the perovskite lattice and suppressing the segregation of Sr, which serves as the primary precursor for chromate formation. Second, controlled Ba excess triggers in situ exsolution of BaCoO3 nanodomains that function as sacrificial Cr scavengers while providing additional catalytic sites. Third, the incorporation of trace Mo modulates the surface acid–base properties, creating unfavorable thermodynamics for CrO3 adsorption. DFT calculations support the proposed trend, suggesting reduced electron density at La–O–Sr motifs and weakened Cr–O interactions. Comprehensive characterization through impedance spectroscopy, surface analytical techniques, and computational modeling provides mechanistic insights extending beyond empirical performance metrics. DRT analysis revealed that current-induced degradation preferentially affects electrode/electrolyte interfacial processes, whereas ToF-SIMS analysis demonstrated minimal Cr penetration in the optimized composition. These findings identify critical degradation pathways and support the effectiveness of the integrated protection strategy. Beyond the specific LSB0.35CFM0.02 composition, this work suggests a generalizable multilevel design principle: coupling intrinsic lattice stabilization (via A-site substitution), thermodynamic getter formation (via A-site excess), and surface acidity modulation (via B-site doping). This strategy is readily applicable to other Sr/Ba-containing perovskite electrodes (e.g., La1-xSrxCoO3₋δ and Ba1-xSr1-xCo1-yFeyO3₋δ), offering a generalizable blueprint for developing durable, contamination-resistant electrode materials. The implementation of this design strategy could substantially extend RSOC operational lifetimes, advancing their deployment for grid-scale energy storage and sustainable fuel production. Methods Powder synthesis The electrode powders were synthesized using the sol-gel process. Taking La0.6Sr0.1Ba0.35Co0.2Fe0.78Mo0.02O3₋δ (LSB0.35CFM0.02) as a representative example, stoichiometric amounts of metal nitrates—La(NO3)3·6H2O (99.9%, Macklin), Sr(NO3)2 (99.5%, Macklin), Ba(NO3)2 (99.5%, Macklin), Co(NO3)2·6H2O (99.9%, Macklin), Fe(NO3)3·9H2O (99.9%, Macklin), and (NH4)6Mo7O24·4H2O (99.9%, Macklin)—were dissolved in deionized water to form a precursor solution. Citric acid (CA, 99.5%, Macklin) and ethylenediaminetetraacetic acid (EDTA, 99.5%, Macklin) were added as complexing agents, with the molar ratio of total metal ions/EDTA/CA fixed at 1/1/1.5. The pH of the solution was adjusted to ~9 using an aqueous ammonia solution (25–28 wt.%, Macklin). Under continuous stirring and heating at 90 °C for 10 h, the solvent evaporated to form a viscous gel. Finally, the dried gel was calcined at 1000 °C for 5 h in air to obtain the crystallized perovskite powder. Cell fabrication Commercial anode-supported half-cells (Ni-YSZ/Ni-SSZ/SSZ/GDC, with a dense SSZ electrolyte thickness of ~10 μm), dense SSZ electrolyte pellets (thickness of ~250 μm, diameter of ~12 mm), and GDC powder were purchased from SOFCMAN Energy. The electrolyte pellets and half-cells were stored in a desiccator at room temperature prior to use. To fabricate the symmetric cells, a GDC buffer layer ink—prepared by mixing the commercial GDC powder with a terpineol-based binder containing 6 wt.% ethylcellulose (Macklin) at a weight ratio of 2:1—was screen-printed onto both sides of the dense SSZ electrolytes and sintered at 1300 °C for 5 h. Subsequently, the LSB0.35CFM0.02 ink, prepared by dispersing the as-synthesized powders into the same terpineol-based binder system, was symmetrically screen-printed onto the sintered GDC interlayers and sintered at 1000 °C for 2 h in air to obtain the LSB0.35CFM0.02/GDC/SSZ/GDC/LSB0.35CFM0.02 symmetric cells. For single cells, the electrode ink was applied onto the GDC layer of the commercial anode-supported half-cells and sintered under identical conditions to yield Ni-YSZ/Ni-SSZ/SSZ/GDC/LSB0.35CFM0.02 single cells. The geometric area of the functional oxygen electrode is 0.25 cm2. Electrochemical testing EIS measurements were carried out in the frequency range of 0.01–100 kHz with an AC amplitude of 10 mV. EIS was recorded under identical gas flow (50 mL min–1), temperature (750 °C), and open-circuit/current-bias conditions across all compositions. The DRT analysis employed a logarithmic time grid with Tikhonov regularization, with the regularization strength determined by L-curve or generalized cross-validation analysis. For validation, representative spectra were fitted to an equivalent circuit model (L-Ro-n(R-CPE)) using nonlinear least-squares using the ZView3 software. The current collection was achieved by brushing silver paste (95%, Macklin) onto the electrode surfaces and attaching silver wires (99.9%, Macklin). The assembly was connected to an Autolab PGSTAT302 workstation for electrochemical characterization, including current-voltage (I–V) curves and EIS under various operating conditions. The Ro was obtained from the high-frequency intercept of the Nyquist plots with the real axis. The Rp was determined from the difference between low-frequency and high-frequency intercepts. All presented I–V curves and cell voltages are reported without iR correction. The data are presented as the means ± s.d. (n = 3). For the Cr-poisoning tests on symmetric cells, the electrochemical durability was evaluated under two distinct conditions: open-circuit voltage and a cathodic polarization current density of 0.5 A·cm⁻2 using a standard three-electrode method. In this configuration, one side of the symmetric electrode served as the working electrode, the opposite side as the counter electrode, and a silver wire attached near the electrolyte edge acted as the reference electrode. ~10 g of Cr2O3 powder (99.9%, Macklin) placed 10 mm upstream in the gas delivery tube was utilized as the Cr source. To evaluate the CO2 tolerance, the symmetric cells were exposed to an atmosphere containing 1 vol.% CO2 balanced with N2 at 750 °C, where the impedance evolution was monitored over time. The impedance features resolved by DRT analysis were evaluated using characteristic capacitance values derived from peak frequencies according to Eq. (1)64.1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C\approx \frac{1}{2\pi {f}_{max }R}$$\end{document}C≈12πfmaxR Characteristic capacitance (C) and the pO2 dependence reaction order (n) were used as commonly adopted descriptors to estimate the possible physicochemical origin and spatial scale of each response. The resulting categorization framework is summarized in Supplementary Table 1. Because impedance responses generally arise from coupled physicochemical processes, these categorizations are used for comparative trend analysis rather than unique mechanistic identification. ECR measurements The ECR technique is employed to quantify the oxygen surface exchange kinetics. The underlying principle is based on the defect chemistry of mixed ionic-electronic conductors, where the electrical conductivity (σσ) is directly proportional to the charge carrier concentration. Consequently, the re-equilibration of oxygen stoichiometry following an abrupt change in pO2 can be monitored in real-time by recording the conductivity relaxation65. The synthesized powders were first pressed (5 MPa) and sintered (1150 °C, 5 h) to attain dense bar-shaped samples. For the Cr-poisoning tolerance tests (ECR@Cr), these dense sticks were pre-treated by sintering at 750 °C for 120 h, placed upside-down on an alumina boat containing 10 g Cr2O3 powder (99.9%, Macklin) to ensure exposure to volatile Cr species. During the measurement, the electrical conductivity was measured at 750 °C using a four-probe DC method. The experiment involves subjecting a dense, bar-shaped sample to a step change in pO2 (e.g., from 0.10 to 0.21 atm). The normalized conductivity relaxation, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${g}(t)$$\end{document}g(t), is defined as Eq. (2)66.2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$g(t)=\frac{\sigma (t)-\sigma (0)}{\sigma (\infty )-\sigma (0)}$$\end{document}g(t)=σ(t)−σ(0)σ(∞)−σ(0)where σ (0)σ(0), σ (t)σ(t), and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\sigma (\infty )$$\end{document}σ(∞) represent the conductivity at the initial state, at time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${t}$$\end{document}t, and at the new equilibrium state, respectively. The relaxation kinetics are generally governed by two competing processes: the surface exchange reaction (kchem) and bulk chemical diffusion (Dchem). The rate-determining step is dictated by the relationship between the sample thickness (L) and the critical length (Lcrit), which is defined as Eq. (3)67.3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${L}_{{crit}}=\frac{{D}_{{chem}}}{{k}_{{chem}}}$$\end{document}Lcrit=Dchemkchem When the sample thickness is significantly smaller than the critical length (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$L\ll {L}_{{crit}}$$\end{document}L≪Lcrit), the oxygen transport across the sample is sufficiently fast, and the overall relaxation process becomes limited by the surface exchange reaction. For the dense bar samples used in this work, the characteristic thickness ( ~ 2 mm) is well below the typical Lcrit reported for cobalt-based perovskites (often 5–10 mm under these conditions). Therefore, the measured relaxation curves were fitted using the surface-limited model to extract the surface exchange coefficient (kchem), as the diffusion process is considered negligible in this regime. To rigorously evaluate the Cr tolerance, durability was quantified by the retention metric kret = kafter/kbefore. We ensured strict comparability by maintaining identical exposure conditions (10 g Cr2O3 source for 120 hours), temperature (750 °C), pO2 modulation (0.10 to 0.21 atm), and sample thickness ( ~ 2 mm) across all compositions, and by performing the relaxation analysis using the same fitting workflow. Characterization details Phase and Structural Analysis: The phase compositions of the synthesized powders and sintered pellets were analyzed by XRD using a Rigaku SmartLab diffractometer equipped with Cu Kα radiation (λ = 1.5418 Å). The data were collected in the 2θ range of 20–80° with a scanning speed of 5° min⁻¹. High-temperature XRD patterns were recorded using the same diffractometer equipped with an in-situ heating chamber (Anton Paar, HTK 1200 N) from ambient temperature to 1000 °C in air. Microstructural Characterization: The microstructural morphology of the electrodes before and after testing was characterized using a field-emission scanning electron microscope (FE-SEM, Hitachi SU8010, Japan). All SEM-EDS line-scan measurements were performed using the same instrument settings: acceleration voltage (3 kV), working distance (8.1 mm), magnification (3 k), and acquisition time (60 s per line scan). Detailed crystal structures and lattice fringes were investigated using TEM and high-resolution TEM on a JEOL JEM-2100F microscope operated at an accelerating voltage of 200 kV. Elemental distribution maps were acquired using an EDX detector integrated with the TEM system. Surface Chemical Analysis: XPS measurements were performed on a Thermo Scientific ESCALAB 250Xi spectrometer using monochromatic Al Kα radiation (1486.6 eV). XPS spectra were acquired at the same take-off angle (90°) and pass energy (30 eV) for all samples. All binding energies were charge-referenced to the adventitious C 1 s at 284.8 eV. Sr 3 d fitting employed two doublets for surface (Srsurf) and lattice (Srlat) species with fixed doublet separations and branching ratios, shared FWHM within a state, and a Shirley background. Cr 2p fitting used analogous constraints for low- (Cr3⁺) and high-oxidation (Cr⁶⁺) components. The degree of segregation was quantified by the ratio of integrated doublet areas (Srsurf/Srlat). Each reported value represents the mean ± standard deviation (s.d.) derived from three independently prepared electrodes (n = 3). Raman spectroscopy and mapping were performed using a Horiba LabRAM HR Evolution system with a 532 nm excitation laser. Single-point spectra were collected to identify the phase evolution and surface deposits. High-resolution Raman mapping was conducted over a 20 × 20 μm2 area to visualize the spatial distribution of chromium species. The maps were constructed based on the integrated intensity of the characteristic chromate (CrO42⁻) band centered at 859 cm⁻1 peak. Depth profiling of elemental distribution was conducted using time-of-flight secondary-ion mass spectrometry (ToF-SIMS, IONTOF M6). A pulsed 30 keV Bi3+ primary ion beam was used for analysis (scanning area: 30 × 30 μm2), while a 2 keV Cs+ analysis beam was used for sputtering (150 × 150 μm2) to prevent crater edge effects. Secondary ions were collected in negative polarity. To ensure rigorous comparability, the intensity of the chromium contaminant (CrO4–) was normalized to the stable lattice matrix signal (LaO–). The depth scale was calibrated by converting the sputter time using an estimated sputter rate of ~0.15 nm s−1. TPD of CO2/NH3 was performed on a Micromeritics AutoChem II 2920 chemisorption analyzer. To ensure strict comparability, identical sample mass ( ~ 100 mg), heating rate (10 °C min⁻¹), and carrier gas flow (50 mL min⁻¹) were applied across LSCF, LSB0.35CF, and LSB0.35CFM0.02 samples. Prior to adsorption, samples were pre-annealed at 500 °C in dry He for 1 h to remove pre-adsorbed carbonates and other surface impurities. After cooling to 50 °C, the sample was saturated with 5% CO2/He or 5% NH3/He for 1 h and subsequently purged with He to remove physisorbed species. Desorption profiles were collected up to 800 °C using a thermal conductivity detector. Quantification of desorbed amounts (μmol g⁻¹) was achieved by external calibration, in which a series of standard gas pulses (typically 50–200 μL) were injected to build a linear calibration curve (R2 > 0.995) covering the full experimental signal range. Thermodynamic Calculations: The Gibbs free energy (ΔG) for the reactions between alkaline earth metal oxides (SrO/BaO) and Cr species was calculated using HSC Chemistry 6.0 software to evaluate the thermodynamic feasibility of the getter mechanism. DFT calculations DFT calculations were performed using the Vienna Ab initio Simulation Package (VASP)68–70. The AO-terminated (100) surfaces of LSCF (La0.6Sr0.4Co0.2Fe0.8O3₋δ) were modeled using a supercell size of 7.71 Å × 19.27 Å in the xy-plane with eight atomic layers in the z-direction and a 20 Å vacuum gap to prevent spurious interactions. Given the substantial substitution level in LSBCF (La0.6Sr0.1Ba0.3Co0.2Fe0.8O3₋δ), wherein three-quarters of the Sr in LSCF is replaced by Ba, its structural model was constructed by introducing this substitution into the bulk LSCF. The corresponding surface configuration of LSBCF was subsequently generated through cleavage of bulk LSBCF. Furthermore, the LSBCFM (La0.6Sr0.1Ba0.3Co0.2Fe0.78Mo0.02O3₋δ) surface was further generated by substituting one Fe atom with Mo on the LSBCF surface. The top and side views of the three surface models are shown in Supplementary Fig. 22. All spin-polarized periodic DFT calculations were performed using the VASP with the Perdew–Burke–Ernzerhof functional and Grimme’s DFT-D3 dispersion correction68–70. A plane-wave cutoff of 400 eV and a (2 × 2 × 1) k-point mesh were used, with convergence criteria of 1 × 10⁻5 eV for energy and 0.03 eV/Å for forces. Effective Hubbard U values of 4, 6.5, 3, and 2 eV were applied to Fe71, Co72, Cr73, and Mo74, respectively. During structural optimization, the bottom four layers were fixed while the others were fully relaxed. The adsorption energy of CrO3 was defined as Eads = \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{{CrO}_{3}}$$\end{document}ECrO3+surface– \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{{CrO}_{3}}$$\end{document}ECrO3–Esurface, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{{CrO}_{3}}$$\end{document}ECrO3+surface, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$E_{{CrO}_{3}}$$\end{document}ECrO3, and Esurface denote the total energies of the adsorbed system, isolated CrO3 in vacuum, and clean surface, respectively. The adsorption Gibbs free energy was calculated as Gads = Eads + ΔEZPE – TΔS, where zero-point energy (ΔEZPE) and entropy (ΔS) contributions were included. All Gibbs free energies were evaluated at 750 °C. The rate of CrO3 adsorption was calculated by the Eqs. (4) and (5):4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{r}}}_{{{ads}}}={{{k}}}_{{{ads}}}{{C}}_{{site}}{{a}}_{{A}}{{{\theta }}}^{*}$$\end{document}rads=kadsCsiteaAθ*5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{{k}}}_{{{ads}}}=\frac{{{k}}_{{B}}{T}}{{h}}\exp \left(\frac{-\triangle {G}}{{{k}}_{{B}}{T}}\right)$$\end{document}kads=kBThexp−△GkBT Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Transparent Peer Review file Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Transparent Peer Review file Source data Source Data Source Data