Anion Sublattice Engineering via Fluorine Doping to Enhance δ‐Bi₂O₃ Stability for Low‐Temperature Solid Oxide Electrochemical Cells

Abstract

Solid oxide electrochemical cells (SOCs) are promising next‐generation, eco‐friendly, and efficient energy conversion devices. However, their high operating temperatures hinder commercialization, primarily due to the lack of highly durable and active materials for low‐temperature operation. Herein, a highly stable and conductive δ‐Bi₂O₃‐based ionic conductor is introduced, in which unoccupied oxygen sites are mediated by F⁻ ions to enhance structural stability and conductivity. The optimized material exhibits an exceptional ionic conductivity of 0.228 S cm⁻¹ at 600 °C, representing a more than 70‐fold increase compared to conventional Y‐doped zirconia, while maintaining excellent long‐term stability. Density functional theory calculations reveal that F⁻ incorporation stabilizes the disordered anion sublattice, reinforcing the cation–anion bonding strength and enhancing the structural symmetry of the δ‐cubic fluorite structure. When integrated into a composite oxygen electrode, the developed ionic conductor enables superior electrochemical performances in SOCs, achieving 0.98 W cm⁻² in fuel cell mode and 0.63 A cm⁻² at 1.3 V in electrolysis mode at 600 °C. These findings provide insights into the rational design of stable and active materials for high‐performance SOCs, facilitating efficient operation at reduced temperatures and advancing their practical viability.

The highly stable and conductive δ‐Bi₂O₃‐based oxygen ionic conductors are developed through vacancy‐mediated stabilization of halogen ion doping for low‐temperature applications. The developed bismuth oxides exhibit excellent stability, with incorporated halogen ions reinforcing cation‐anion bonding strength and structural symmetry. By employing these bismuth oxides, solid oxide electrochemical cells achieve superior performance in both fuel cell and electrolysis cell modes, along with excellent durability.

1. Introduction

With increasing awareness of environmental protection and the accelerating depletion of fossil fuels, the demand for sustainable and efficient energy conversion systems has grown rapidly.^{[ 1 ]} Solid oxide electrochemical cells (SOCs) have attracted significant attention as promising energy conversion devices due to their outstanding properties, including high efficiency, eco‐friendliness, and operational flexibility between fuel cell (FC) mode and electrolysis cell (EC) modes.^{[ 2} , ³ , ⁴ , ⁵ , ^{6 ]} However, the high operating temperatures (>800 °C) of SOCs pose significant challenges, such as chemical degradation of component materials and mechanical failures, including delamination and cracking.^{[ 7} , ^{8 ]} Therefore, lowering the operating temperature of SOCs is crucial for their practical implementation.^{[ 9 ]} Unfortunately, at reduced temperatures, the electrochemical reactions in SOCs become less thermally activated, leading to increased polarization resistance in oxygen electrodes, which in turn impairs the electrochemical performance of low‐temperature SOCs (LT‐SOCs).^{[ 10} , ^{11 ]}

Cobaltite‐based mixed ionic and electronic conductors (MIECs) have been extensively investigated as oxygen electrodes for LT‐SOCs due to their superior electrocatalytic activity.^{[ 12} , ¹³ , ¹⁴ , ¹⁵ , ^{16 ]} However, their widespread application is hindered by chemical instability‐such as cation segregation (Sr, Co, etc.) and CO₂ poisoning‐as well as a thermal expansion coefficient (TEC) mismatch with commercial electrolytes, including scandia‐stabilized zirconia (ScSZ) and yttria‐stabilized zirconia (YSZ).^{[ 17} , ¹⁸ , ¹⁹ , ^{20 ]} As a cobalt‐free alternative, manganite perovskites have emerged as promising candidates for oxygen electrodes due to their high electronic conductivity and excellent mechanical and chemical compatibility with electrolyte materials.^{[ 21} , ²² , ^{23 ]} Moreover, incorporating manganite perovskites with various ionic‐conducting oxides enhances catalytic activity by increasing the density of triple‐phase boundary (TPB) sites, where gaseous, electronic, and ionic species interact.^{[ 24} , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ^{29 ]}

In particular, δ‐Bi₂O₃, known for its remarkably high ionic conductivity (e.g., 1 S cm⁻¹ at 800 °C), has demonstrated significant potential in composite oxygen electrodes for developing high‐performance LT‐SOCs. The outstanding ionic conductivity of δ‐Bi₂O₃ arises from its high concentration (25%) of disordered oxygen vacancies and the high oxygen ion mobility facilitated by the 6s ² lone pair electrons in Bi³⁺.^{[ 30} , ^{31 ]} However, pure δ‐Bi₂O₃ is thermodynamically stable only within a narrow temperature range (729—824 °C). To extend its stability, isovalent cations (e.g., Er³⁺, Dy³⁺, and Y³⁺) have been incorporated into Bi₂O₃ to induce lattice strain, with the goal of stabilization of the δ‐phase down to room temperature.^{[ 32 ]} Despite these efforts, below 650 °C, doped Bi₂O₃ undergoes a phase transition from a cubic to a low‐symmetry rhombohedral structure due to the metastability of the cubic phase, leading to a significant decrease in ionic conductivity.^{[ 33} , ^{34 ]} For instance, Er‐stabilized Bi₂O₃ (ESB) exhibited a sharp decline in its ionic conductivity from 0.15 to 0.006 S cm⁻¹ during 250 h of annealing at 600 °C, severely limiting its application for SOCs.^{[ 35 ]} While numerous studies have explored single or multiple cations doping to substitute Bi sites,^{[ 36} , ³⁷ , ^{38 ]} the inherently disordered and partially occupied oxygen sublattice remains insufficiently modified, leaving room for further improvement in phase stability.

Over the past decade, innovative strategies employing halogen anion doping (e.g., F⁻, Cl⁻) at oxygen sites have been explored to enhance the structural stability and long‐term durability of oxide materials.^{[ 39} , ^{40 ]} For example, Xiao et al. demonstrated that F⁻ doping in LiNi_0.8Co_0.15Al_0.05O₂ cathode materials for lithium‐ion batteries stabilizes the ordered layered structure and suppress metal ion migration by forming strong metal–F bonds.^{[ 41 ]} More recently, Ni et al. showed that F⁻ substitution induces lattice distortion and increases the energy barrier for oxygen vacancy formation, thereby enhancing the lattice rigidity of La_0.5Ba_0.5Cu_0.5Fe_0.5O_3‐δ in protonic solid oxide fuel cells.^{[ 42 ]} Furthermore, several studies have consistently reported that F⁻ doping in SOC electrode materials improves overall long‐term stability under various operating conditions.^{[ 43} , ⁴⁴ , ^{45 ]} Building upon these findings, we hypothesized that modulating the anion sublattice via F⁻ incorporation could enhance the phase stability of Bi₂O₃ at lower temperatures. While the disordered oxygen sublattice with abundant vacancies promotes fast ionic diffusion, it also contributes to phase instability. Strategic modification of oxygen sites with halogen ions may alter atomic interactions within cubic fluorite crystal symmetry, thereby enhancing stability without significantly compromising ionic conductivity.

In this study, we propose a highly durable ESB doped with F⁻ as a promising oxygen electrode component for LT‐SOCs. Comprehensive physicochemical and crystallographic analyses confirmed the successful formation of a solid solution with a cubic fluorite structure. Density functional theory (DFT) calculations provided insights into the energetically favorable site of F⁻ incorporation and its influence on atomic configuration, electronic structure, and phase stability. Additionally, a composite oxygen electrode composed of F⁻‐doped ESB and La_0.8Sr_0.2MnO₃(LSM) was developed, and a mechanistic analysis using the distribution of relaxation times (DRT) technique was conducted to elucidate the role of F⁻ doping in the oxygen‐related reactions. Finally, an SOC with the composite oxygen electrode containing F⁻‐doped ESB was fabricated to evaluate its electrochemical performance in both FC and EC modes along with its long‐term stability.

2. Results and Discussion

2.1. Physicochemical Characterization of F⁻‐Doped ESB

Figure 1a presents the X‐ray diffraction (XRD) patterns of a series of ESB powders with F⁻‐doping (Er_xBi_1‐xO_1.5‐0.75yF_1.5y, yFxESB) where Er³⁺ concentrations were varied in 10FxESB (x = 10, 15, and 20 mol%) to determine the minimum Er³⁺ dopant concentration required to stabilize the cubic fluorite phase in F⁻‐doped system. The results indicate that 15 mol% Er³⁺ doping (10F15ESB) successfully stabilizes the cubic phase while undoped Bi₂O₃ requires more than 20 mol% Er³⁺ to achieve δ‐cubic phase stabilization, as observed in Er_0.2Bi_0.8O_1.5 (20ESB). To investigate the solubility limit of F⁻ in the cubic phase, we synthesized yF15ESB (y = 2, 6, 10, and 14 mol%) powders with varying F⁻ dopant concentrations (Figure 1b). Samples with F⁻ doping concentrations exceeding 10 mol% exhibited the formation of the tetragonal β‐phase, which has lower ionic conductivity compared to the cubic δ‐phase. Based on these findings, 10F15ESB was selected for further investigation. The refined XRD pattern of 10F15ESB is shown in Figure 1c, while the corresponding pattern for 20ESB is presented in Figure S1 (Supporting Information). The detailed Rietveld refinement results, summarized in Table S1 (Supporting Information), confirm the formation of a cubic fluorite structure with Fm $\overline{3}$ m symmetry. The calculated lattice parameters for 20ESB and 10F15ESB were 5.508 and 5.515 Å, respectively. The slight lattice expansion in 10F15ESB is attributed to the reduced Er³⁺doping concentration, given that Er³⁺ (0.89 Å) has a smaller ionic radius than Bi³⁺ (1.03 Å). The inset of Figure 1c displays the lattice fringe pattern of 10F15ESB reconstructed via inverse fast Fourier transform (IFFT), with an interplanar spacing of 3.19 Å at (111) plane, which aligns well with the lattice parameter obtained from Rietveld refinement. Figure S2 (Supporting Information) provides the high‐resolution transmission electron microscopy (HR‐TEM) image and corresponding lattice fringe pattern of 20ESB. Figure 1d presents the elemental distribution in 10F15ESB, obtained through energy‐dispersive X‐ray spectroscopy (EDS), demonstrating the uniform distribution of Er, Bi, and F throughout the sample. Similarly, the homogeneous distribution in 20ESB is shown in Figure S3 (Supporting Information). As summarized in Table S2 (Supporting Information), the atomic fractions of fluorine in the xF15ESB samples (x = 2, 6, and 10) generally increased with the intended F⁻ doping level, and the Er‐to‐Bi ratio increased appreciably from 10F15ESB to 10F20ESB, indicating that the intended doping concentration variations were well reflected. To further analyze the elemental distribution, time‐of‐flight secondary ion mass spectrometry (TOF‐SIMS) depth profiles were obtained for 10F15ESB (Figure 1e) and 20ESB (Figure S4, Supporting Information) as a function of sputter time. The strong and persistent F⁻ signal detected in 10F15ESB confirms the successful incorporation of F⁻ ions into the material. The F 1s X‐ray photoelectron spectroscopy (XPS) spectrum in Figure S5 (Supporting Information) confirms the presence of F⁻ ions in 10F15ESB. Figure 1f presents the O 1s XPS spectra of both 10F15ESB and 20ESB. The O 1s peak was deconvoluted into two distinct subpeaks corresponding to chemisorbed oxygen species (O_ads) and lattice oxygen species (O_lat), located at ∼531 eV and ∼529 eV, respectively. The relative proportions of each oxygen species, summarized in Table S3 (Supporting Information), reveal that O_ads content in 10F15ESB (41.2%) is lower than in 20ESB (47.4%). Given that O_ads are associated with oxygen vacancies, this result suggests that F⁻ doping effectively reduces the concentration of oxygen vacancies. Figure S6 (Supporting Information) shows the XPS analysis of the O 1s peak for the 10F20ESB. The O_abs/(O_abs+O_lat) ratio of 10F20ESB was 41.75%, which is comparable to that of 10F15ESB and lower than that of 20ESB. These results indicate the oxygen vacancy concentration is more strongly influenced by the concentration of the F⁻ dopant than by that of the Er³⁺ dopant. The similar oxidation states of Er³⁺ and Bi³⁺ lead to a relatively lower influence on oxygen vacancy formation due to less pronounced charge compensation effects. Additionally, the binding energies of the oxygen species in 10F15ESB (530.75 eV for O_ads and 529.20 eV for O_lat) shift toward lower values compared to those in 20ESB (531.00 eV for O_ads and 529.40 eV for O_lat), indicating increased electron participation in bonding with Er and Bi cations, thereby strengthening metal‐oxygen interactions. The XPS spectra for Bi 4f are shown in Figure 1g, and the corresponding binding energies are summarized in Table S4 (Supporting Information). Upon F⁻ doping, the binding energies of Bi species increased, further supporting the hypothesis that F⁻ doping enhances bonding interactions between Bi and surrounding anions.

Figure 1.

Physicochemical and crystallographic properties of 10F15ESB. a) XRD patterns of 10FxESB (x = 10, 15, and 20 mol%) powders with different Er³⁺ dopant concentrations. b) XRD patterns of yF15ESB (y = 2, 6, 10, and 14 mol%) powders with different F⁻ dopant concentrations. c) Refined XRD pattern of 10F15ESB, with the inset displaying the corresponding lattice fringe pattern. d) Elemental distribution of Er, Bi, and F in TEM‐EDX mapping images of 10F15ESB. e) TOF‐SIMS depth profiles of Er, Bi, and F for the 10F15ESB sample. XPS spectra of f) O 1s and g) Bi 4f orbitals for 10F15ESB and 20ESB.

2.2. Conduction Properties of F⁻‐Doped ESB

Figure 2a,b presents the time‐dependent Nyquist plots for 20ESB and 10F15ESB, respectively. Figure S7 (Supporting Information) further displays the initial electrochemical impedance spectroscopy (EIS) spectra of 10F15ESB at 600 °C over a wider frequency range of 10⁶ to 10⁻¹ Hz, indicating a fully closed arc under the measurement condition. The impedance spectra of 20ESB exhibited a significant rightward shift over time, indicating an increase in ohmic resistance due to conductivity degradation. In contrast, the impedance spectra of 10F15ESB showed little change over 100 h. For both compositions, no additional arc growth was observed, suggesting that the effect of F⁻ doping on grain boundary resistance is not significant. Figure 2c displays the time‐dependent conductivity evolution of Bi₂O₃ with varying Er³⁺ and F⁻ doping concentrations. The 20ESB sample exhibited a sharp decrease in conductivity within the first 50 h, whereas F⁻‐doped ESB samples demonstrated significantly improved durability. The degradation rates were quantified as 95.31% for 20ESB, 68.78% for 2F15ESB, 47.45% for 6F15ESB, and only 5.94% for 10F15ESB, highlighting the stabilizing effect of F⁻ doping on long‐term conductivity retention. In a prolonged long‐term stability test for 2F15ESB for 300 h, the ionic conductivity of 2F15ESB was approaching that of 20ESB, implying that F⁻ doping kinetically slows down the phase change (Figure S8, Supporting Information). In previous studies on the suppression of the cubic‐to‐rhombohedral phase transition of δ‐Bi₂O₃‐based ionic conductors, cation diffusion was identified as a key factor governing the kinetics of phase change.^{[ 35} , ³⁶ , ⁴⁶ , ^{47 ]} It has been reported that reduced cation diffusion leads to slower cubic‐rhombohedral phase transition kinetics. In this context, the higher electronegativity of fluorine may result in a stronger cation–fluorine bond compared to the cation–oxygen bond, thereby slowing cation diffusion and suppressing the phase transition. A systematic examination revealing the kinetic and thermodynamic contributions of F⁻ doping to stabilization would be a valuable direction for future study. Figure 2d compares the XRD patterns of 20ESB and 10F15ESB after prolonged operation at 600 °C. The 20ESB sample underwent a characteristic phase transition to a rhombohedral structure, whereas 10F15ESB retained its cubic phase, demonstrating its enhanced structural stability. As shown in Figure S9 (Supporting Information), both 2F15ESB and 6F15ESB also demonstrated the suppression of the phase transition. To further investigate the contributions of electronic and ionic conductivity in 10F15ESB, impedance spectra were measured for Au|10F15ESB|Au symmetrical cells at 650 °C under varying oxygen partial pressure (pO₂) ranging from 0.1 to 1.0 atm (Figure 2e). To analyze these spectra, we employed the equivalent circuit model (see inset of Figure 2e), proposed by Lai and Haile.^{[ 48 ]} With the fitting approach proposed in their previous work, the deconvolution of the ionic and electronic conductivities of the specimen was performed, which are plotted as a function of pO₂ in Figure 2f. The results demonstrate that the ionic contribution to total conductivity exceeds the electronic contribution by approximately two orders of magnitude across the examined pO₂ range. The total conductivity of 10F15ESB was compared with conventional ionic conductors through Arrhenius analysis (Figure 2g). This comparison reveals the superior ionic transport characteristics of 10F15ESB. At 600 °C 10F15ESB exhibited a conductivity of 0.199 S cm⁻¹, surpassing that of gadolinium‐doped ceria (0.018 S cm⁻¹), YSZ (0.003 S cm⁻¹), and even 20ESB (0.147 S cm⁻¹).^{[ 49 ]} Notably, 10F15ESB also exhibited a reduced oxygen ion migration enthalpy (0.68 ± 0.05 eV) compared to ESB (0.87 ± 0.04 eV), underscoring its enhanced ionic transport properties. As shown in Figure S10 (Supporting Information), 10F15ESB exhibited higher ionic conductivity than both 10F20ESB and 20ESB. Thus, we attribute the improvement in ionic conductivity to the lowered Er³⁺ concentration, which more than compensates for the reduction in oxygen vacancy concentration. These results demonstrate the effectiveness of F⁻ doping as a strategy for stabilizing the cubic δ‐phase of ESB while maintaining its superionic conductivity at 600 °C.

Figure 2.

Ionic conduction properties of the bismuth oxides with different compositions. Nyquist plots of a) 20ESB and b) 10F15ESB during long‐term operation at 600 °C. c) Time‐dependent behavior of conductivity for yF15ESB (y = 2, 6, and 10 mol%) and 20ESB at 600 °C. d) XRD patterns of 10F15ESB and 20ESB after long‐term stability test at 600 °C. e) A series of impedance spectra of Au|10F15ESB|Au symmetrical cells at 650 °C with different pO₂ values. f) Ionic and electronic conductivities of 10F15ESB as a function of pO₂. g) Comparison of electrical conductivities of various oxygen ion conductors at different temperatures.

2.3. Density Functional Theory Calculations

To elucidate the role of F⁻ doping in enhancing phase stabilization, density functional theory (DFT) calculations were performed. Figure 3a illustrates the crystal structure of δ‐Bi₂O₃, highlighting potential defect sites. For charge neutrality, the substitution of one O²⁻ ion by two F⁻ ions was considered. The optimal position of an additional F⁻ ion was initially determined within the fluorite symmetry, specifically interstitial sites (blue dotted square) and oxygen vacancy sites (red dotted circle). Figure 3b presents the calculated energy changes associated with F⁻ incorporation at these sites. The energy for incorporation at an oxygen vacancy site was found to be −4.1 eV, which is lower than that for the interstitial site (−2.9 eV). This finding indicates that F⁻ preferentially occupies unoccupied oxygen sites, effectively reinforcing the anion sublattice and stabilizing the cubic matrix. At interstitial sites, the repulsive force from surrounding oxygen ions likely destabilizes F⁻ incorporation, despite the spatial availability. Figure 3c depicts the calculated energy cost associated with the transition from cubic to rhombohedral symmetry. Notably, F⁻ doping effectively increased the phase transition energy barrier from 0.81 to 1.91 eV, demonstrating its suppressive effect on phase transformation. To understand the origin of this increased energy barrier, chemical bonding and interactions were examined. Figure 3d,e shows the crystal orbital Hamiltonian population (COHP) analysis for ESB and F⁻‐doped ESB (F‐ESB), respectively. The positive COHP area (yellow) represents anti‐bonding characteristics, whereas the negative COHP area (navy) signifies bonding characteristics. The integrated COHP (ICOHP) values provide a quantitative measure of chemical bonding strength. In ESB, the ICOHP value for cation–O²⁻ bonding was −0.79 (Figure 3d), whereas in F‐ESB, the cation‐O²⁻ ICOHP value became more negative (−0.81) (Figure 3e). Additionally, the newly formed cation–F⁻ bonds in F‐ESB exhibited a negative ICOHP value of −0.34, further contributing to the overall increase in bonding strength. To assess the structural impact of F⁻ doping, Figure 3f presents the effective coordination number (ECN) of Bi³⁺ and Er³⁺ cations, where higher ECN values correspond to a more symmetric crystal matrix. Upon F⁻ doping, the ECNs of Bi³⁺ and Er³⁺ increased, suggesting that F⁻ incorporation reinforces the isotropic nature of bonding within the cubic fluorite structure. In previous studies on perovskite oxide‐based oxygen catalysts, F⁻ doping has been reported to increase oxygen vacancy concentration, modulate electronic structure, induce lattice distortion, and activate surface exchange or bulk diffusion of oxygen.^{[ 39} , ⁵⁰ , ⁵¹ , ^{52 ]} In contrast, this study provides a new insight that F⁻ doping in highly defective fluorite structures mitigates structural frustration in anion sublattice, thereby improving phase stability (Figure 3g). Furthermore, as shown in Figure 2, it modulates oxygen vacancy concentration without compromising oxygen ion conductivity, owing to the inherently high oxygen vacancy concentration in δ‐Bi₂O₃.

Figure 3.

a) Schematics for the cubic fluorite structure of δ‐Bi₂O₃ and possible defect sites. b) Energetics of F⁻ incorporation. c) Energetics of the phase transition from the cubic to rhombohedral phase. COHP analysis for d) ESB and e) F‐ESB. f) ECN of cation species in ESB and F‐ESB. g) An Illustration summarizing the effects of F⁻ doping on the stability of δ‐Bi₂O₃.

2.4. Catalytic Activity of 10F15ESB on Oxygen Reduction Reaction

The LSM‐10F15ESB and LSM‐20ESB composite electrodes were applied onto ScSZ electrolyte to fabricate symmetrical cells, enabling an evaluation of their catalytic activity in oxygen reduction reaction (ORR). Figure S11 (Supporting Information) presents the XRD patterns of 10F15ESB combined with LSM and ScSZ after annealing for 10 h at temperatures ranging 600–800 °C, confirming their chemical compatibility. Figure 4a compares the Nyquist plots of both samples at 600 °C under ambient air conditions, following the subtraction of high‐frequency ohmic resistance. The LSM‐10F15ESB electrode exhibited an area‐specific resistance (ASR) value of 0.45 Ω cm², which was 58.9% lower than that of LSM‐20ESB (1.10 Ω cm²). Figure 4b presents the Arrhenius plots of ASRs for LSM‐10F15ESB and LSM‐20ESB electrodes as a function of inverse temperature. Although both electrodes exhibited similar activation energies (1.38 eV for LSM‐10F15ESB and 1.34 eV for LSM‐20ESB), the ASRs of LSM‐10F15ESB remained by ≈60% lower than that of LSM‐20ESB in the 600–750 °C range. To further suggest the influence of the 10F15ESB on ORR kinetics, a DRT analysis was conducted. Figure 4c displays the impedance spectra of LSM‐10F15ESB measured at 600 °C under varying pO₂ from 0.208 to 0.047 atm, while the corresponding DRT plots are shown in Figure 4d. Three distinct ORR sub‐processes were identified across different frequency ranges: high‐ (P1, 10⁴–10⁶ Hz), middle‐ (P2, 10¹–10⁴ Hz), and low‐ (P3, 10⁻²–10¹ Hz) frequencies. Among these, P1 and P2 were the dominant contributors to ORR activity, whereas P3 had a negligible effect on total polarization resistance. Figure 4e presents the ASR fitting results for P1 (ASR_P1) and P2 (ASR_P2) as a function of pO₂. Notably, the resistance of P2 varied with pO₂, whereas P1 resistance remained unchanged. Based on previous studies, each separated ORR sub‐process in the DRT curve can be identified using an equivalent circuit model (inset of Figure 4e) and fitted using the following equation:

$$\mathrm{ASR}=\mathrm{AS}{\mathrm{R}}_0{\left(p{\mathrm{O}}_2\right)}^{-n}$$ (1)

where ASR₀ is a constant and n represents the reaction order. Table S5 (Supporting Information) summarizes the n values corresponding to elementary reactions reported in prior studies. The estimated n values for LSM‐10F15ESB were −0.01 for ASR_P1 and 0.25 for ASR_P2, indicating that P1 corresponds to oxygen ion incorporation into the electrolyte $({\mathrm{O}}_{\mathrm{TPB}}^{2-}+{\mathrm{V}}_{\mathrm{O},\mathrm{electrolyte}}^{··}\leftrightarrow {\mathrm{O}}_{\mathrm{O},\mathrm{electrolyte}}^{\times })$, and P2 is related to the surface diffusion of oxygen ions $({\mathrm{O}}_{\mathrm{ad}}^{-}\leftrightarrow {\mathrm{O}}_{\mathrm{TPB}}^{-})$.^{[ 53} , ⁵⁴ , ^{55 ]} Figure 4f presents a comparison of DRT curves under ambient air at 600 °C, demonstrating a notable reduction in resistance for both P1 and P2 in LSM‐10F15ESB. Figure 4g,h show temperature‐dependent ASR_P1 and ASR_P2, respectively. Notably, at 600 °C, the ASR_P1 and ASR_P2 of LSM‐10F15ESB decreased by 25% and 58%, respectively, compared to LSM‐20ESB. These improvement would be attributed to the higher ionic conductivity of 10F15ESB compared to that of 20ESB.

Figure 4.

Role of 10F15ESB in composite oxygen electrode. a) Nyquist plots of symmetrical cells with LSM‐10F15ESB and LSM‐20ESB electrodes at 600 °C in ambient air. b) Arrhenius plots of the ASRs for LSM‐10F15ESB and LSM‐20ESB electrodes. c) pO₂‐dependent EIS spectra of LSM‐10F15ESB electrode at 600 °C for different pO₂ values. d) A series of DRT curves of LSM‐10F15ESB electrode at 600 °C. e) Deconvoluted ASRs of LSM‐10F15ESB electrode as a function of pO₂ at 600 °C. f) DRT curves of LSM‐10F15ESB and LSM‐20ESB electrodes at 600 °C in ambient air. Comparison of Arrhenius plots of LSM‐10F15ESB and LSM‐20ESB electrodes for the deconvoluted ASRs of g) P1 sub‐process and h) P2 sub‐process.

2.5. Electrochemical Performance of SOCs

We fabricated a SOC with a configuration of Ni‐YSZ fuel electrode support|Ni‐ScSZ fuel electrode functional layer|ScSZ electrolyte|LSM‐10F15ESB air electrode (referred to as LSM‐10F15ESB cell). Figure 5a presents a cross‐sectional scanning electron microscopy (SEM) image of the LSM‐10F15ESB cell, revealing a well‐integrated interface between the LSM‐10F15ESB oxygen electrode and the ScSZ electrolyte. For comparison, a reference SOC was prepared with a conventional LSM‐20ESB air electrode (LSM‐20ESB cell), exhibiting a similar microstructure with a ScSZ electrolyte thickness of ≈6.5 µm (Figure S12, Supporting Information). Figure 5b displays the I‐V‐P characteristics of both cells in FC mode at 600 °C. The open circuit voltage (OCV) of both cells were measured at ≈1.1 V, which is close to the theoretical value, confirming the gas‐tight nature of the dense ScSZ electrolyte. The LSM‐20ESB cell exhibited a maximum power density (MPD) of 0.63 W cm⁻², while the LSM‐10F15ESB cell achieved 1.55 times higher value (0.98 W cm⁻²) at 600 °C. Figure 5c depicts Nyquist plots of the two cells at 600 °C. Both exhibited comparable ohmic resistances (R _ohm) of ≈0.12 Ω cm² at the OCV conditions. The electrode resistance (R_elec) was determined to be 1.26 Ω cm² for the LSM‐10F15ESB and 1.44 Ω cm² for the LSM‐20ESB cell (inset of Figure 5c). Notably, under MPD condition (≈0.5 V), the LSM‐10F15ESB cell showed a 46% reduction in R_elec (0.11 Ω cm²) compared to the LSM‐20ESB cell (0.21 Ω cm²), accounting for the enhanced MPD. Temperature‐dependent I‐V‐P curves for LSM‐10F15ESB and LSM‐20ESB cells are displayed in Figure 5d and Figure S13 (Supporting Information), respectively. Figure 5e compares the MPDs of the LSM‐10F15ESB and LSM‐20ESB cells with those of other recently reported SOCs using various oxygen electrodes at 600 °C, demonstrating the outstanding performance of the LSM‐10F15ESB cell. Detailed cell configurations and MPD values are listed in Table S6 (Supporting Information). Long‐term durability tests were conducted on the LSM‐10F15ESB cell in FC mode at 600 °C for 360 h under an applied current density of 250 mA cm⁻² (Figure 5f). The output voltage remained stable with no abrupt degradation. Figure S14 (Supporting Information) presents the cross‐sectional SEM image of the LSM‐10F15ESB cell after prolonged operation, revealing no significant degradation at the LSM‐10F15ESB/ScSZ interface. Furthermore, elemental mapping via SEM‐EDS (Figure S15, Supporting Information) confirmed the uniform distribution of constituent elements, with no evidence of volatilization or interfacial diffusion, indicating the compositional stability of 10F15ESB. A few studies have reported that the degradation rate of composite electrodes incorporating δ‐Bi₂O₃‐based ionic conductors is not as severe as suggested by long‐term stability tests conducted on the δ‐Bi₂O₃‐based ionic conductors alone.^{[ 56} , ⁵⁷ , ⁵⁸ , ⁵⁹ , ⁶⁰ , ^{61 ]} Therefore, the contribution of ionic conductivity stability to the overall stability of SOCs may be an interesting topic for further investigation.

Figure 5.

Electrochemical evaluation for the LSM‐10F15ESB oxygen electrode‐based SOCs. a) Cross‐sectional SEM image of the LSM‐10F15ESB cell. b) I‐V‐P curves for LSM‐10F15ESB and LSM‐20ESB cells at 600 °C. c) Nyquist plots of LSM‐10F15ESB and LSM‐20ESB cells under the OCV condition at 600 °C (Inset is Nyquist plots at 0.5 V). d) I‐V‐P characteristics of the LSM‐10F15ESB cell at various temperatures. e) Comparison of MPDs of various SOCs in this work and previous works at 600 °C. f) Long‐term durability test of the LSM‐10F15ESB cell at 600 °C for 360 h. g) I–V curves of LSM‐10F15ESB and LSM‐20ESB cells in the EC mode at 600 °C. h) I–V curves of the LSM‐10F15ESB cell in the EC mode at various temperatures. i) Comparison of current densities of various SOCs in this work and previous works at different temperatures and 1.3 V.

Figure 5g depicts the I–V curves of both cells in EC mode at 600 °C, where a 50:50 H₂O/H₂ gas mixture was supplied to the fuel electrode and dry air was introduced to the oxygen electrode. The LSM‐10F15ESB cell achieved a current density of 0.63 A cm⁻² at 1.3 V (near thermo‐neutral voltage for steam electrolysis), which was 2.2 times higher than that of the LSM‐20ESB cell (0.28 A cm⁻²). I–V curves of both cells at different temperatures are shown in Figure 5h (LSM‐10F15ESB) and Figure S16 (Supporting Information) (LSM‐20ESB). At 1.3 V, the current densities of the 10F15ESB cell were 1.48, 1.10, and 0.63 A cm^{− 2} at 700, 650, and 600 °C, respectively, corresponding to hydrogen production rates of 11.2, 8.3, and 4.8 mL min⁻¹ cm², as calculated using Faraday's law. Figure 5i compares the current densities of the LSM‐10F15ESB cell with those of other SOCs utilizing zirconia‐based electrolytes at 600–700 °C. The detailed current density values are summarized in Table S7 (Supporting Information), further demonstrating the superior electrolysis performance of the LSM‐10F15ESB cell.

3. Conclusion

In this study, we developed a highly durable and conductive Bi₂O₃‐based electrolyte material by employing an F⁻ doping strategy. The 10F15ESB ionic conductor exhibited a 70‐fold enhancement in ionic conductivity compared to the conventional YSZ ionic conductor. Unlike conventional 20ESB, which undergoes conductivity degradation, 10F15ESB demonstrated exceptional long‐term stability for 100 h at 600 °C, confirming that F⁻ doping effectively suppresses the phase transformation of ESB from a high‐symmetry cubic fluorite structure to a low‐symmetry rhombohedral structure. DFT calculations provided fundamental insights into the role of F⁻ incorporation, revealing that F⁻ preferentially occupies unoccupied oxygen sites rather than interstitial sites, thereby reinforcing the anion sublattice while minimizing structural frustration. This doping mechanism strengthened both cation‐O²⁻ bonding and cation‐F⁻ bonding, further stabilizing the fluorite atomic array. Additionally, F⁻ doping enhanced the crystal symmetry of the cubic phase, as evidenced by an increase in the ECN. Mechanistic analysis of the LSM–10F15ESB composite oxygen electrode demonstrated that the addition of 10F15ESB significantly promotes the kinetics of surface diffusion and oxygen ion incorporation reactions, leading to a substantial reduction in polarization overpotential. Consequently, SOCs incorporating the LSM–10F15ESB electrode exhibited outstanding electrochemical performance, achieving an MPD of 0.98 W cm⁻² in FC mode and a current density of 0.63 A cm⁻² at 1.3 V in EC mode at 600 °C, alongside remarkable operational stability over 360 h. This work provides valuable insights into the design of stable electrolyte materials through halogen ion doping, offering a promising strategy for the development of high‐performance energy conversion devices with enhanced durability and broader practical applications.

4. Experimental Section

Materials Synthesis

10F15ESB and 20ESB powders were synthesized through the solid‐state reaction method. Appropriate stoichiometric amounts of Bi₂O₃ (99.99% pure, Alfa Aesar), Er₂O₃ (99.99% pure, Alfa Aesar), and BiF₃ (99.99% pure, Sigma–Aldrich) were mixed for 24 h via a ball‐milling process with zirconia balls in a high‐density polyethylene bottle. The ball‐milled powders were then calcined at 800 °C for 16 h. After calcination, the powders were finely ground with a mortar and pestle and then sieved using a 45 µm mesh. Lastly, the prepared powders were uniaxially pressed at 50 MPa to form disk‐shaped pellets with a diameter of 10 mm and then sintered at 850 °C for 16 h.

Symmetrical Cells Preparation

For the preparation of symmetrical cells, an oxygen electrode paste was prepared by combining 10F15ESB and LSM powders (FuelCellMaterials) in a volume ratio of 50:50 using a texanol‐based binder (441 ESL Electroscience). For reference, LSM‐20ESB composite was prepared using the same method. Each oxygen electrode paste was applied with a brush onto both sides of ScSZ electrolytes (Daiichi Kigenso Kagaku Kogyo) and subsequently annealed at 750 °C for 2 h.

Fuel Electrode‐Supported Cell Fabrication

Fuel electrode‐supported cells, consisting of a Ni–YSZ support layer, a Ni–ScSZ functional layer, and a ScSZ electrolyte, were fabricated using the tape casting, lamination, and co‐firing processes as described in the previous study.^{[ 62 ]} The LSM‐10F15ESB oxygen electrode ink was then screen‐printed onto the ScSZ electrolyte and sintered at 750 °C for 2 h. For comparison, fuel electrode‐supported cells with a single ScSZ electrolyte and an LSM‐20ESB oxygen electrode were prepared using the same method.

Characterization

The crystallographic structure of the prepared samples was examined using X‐ray diffraction (XRD) analysis with a SmartLab X‐ray diffractometer (Rigaku, KARA) equipped with Cu‐Kα radiation (λ = 1.5418 Å). The microstructural features were observed using field emission scanning electron microscopy (SEM) (S‐8230, Hitachi, KARA), while high‐resolution transmission electron microscopy (HR‐TEM) (Talos F200X, FEI, KARA) was utilized to analyze the morphology of the materials. Elemental distribution was investigated through energy‐dispersive X‐ray spectroscopy (EDX, Bruker, KARA). X‐ray photoelectron spectroscopy (XPS) (K‐alpha, Thermo VG Scientific, KARA) with a monochromatic Al Kα source was employed to analyze the chemical bonding states and atomic concentrations of the samples under a high vacuum of 7 × 10⁻⁸ Torr. Secondary Ion Mass Spectrometry (SIMS) was performed to investigate the elemental depth profiles and distributions within the samples. High‐resolution surface and depth profiling was conducted using a TOF‐SIMS V instrument (IONTOF, KARA) equipped with a Bi³⁺ primary ion source.

The electrical conductivity and electrode polarization resistance of the symmetrical cells were assessed using a potentiostat (VMP‐300, BioLogic). The impedance of each sample was measured across a frequency range of 1 MHz–1 Hz and an amplitude of 50–100 mV. The EC‐Lab software was utilized for fitting an equivalent circuit, and distribution of relaxation time (DRT) analysis was performed using DRT Tools, which were developed by the Ciucci group.^{[ 63} , ^{64 ]} The oxygen partial pressure (pO₂) was modulated using a controlled Ar/O₂ mixture gas through a mass flow controller while maintaining a total flow rate of 200 sccm.

The electrochemical performance of the single cells was assessed using a testing apparatus where the edge between the cell and the alumina reactor was made gas‐tight by sealing it with Ceramabond 517 (Aremco). The I‐V characteristics of the single cells were measured using a potentiostat (VMP‐300, Bio‐Logic). The humidified hydrogen gas supply for the FC and EC performance measurements was controlled by varying the temperature of the water bubbler and maintained at 3% and 50% H₂O (200 sccm), respectively.

Computational Details

DFT calculations were performed using VASP.^{[ 65} , ^{66 ]} The generalized gradient approximation (GGA‐PBE) was chosen for the exchange‐correlation functionals. Projector‐augmented‐wave (PAW) potentials with valence configurations of 4f ²5d ¹⁰6s ²6p ³, 2s ²2p ⁴, and 2s ²2p ⁵ were applied to describe Bi, O, and F, respectively. Plane waves with an energy cutoff of 450 eV were used to expand the electronic wave functions. A 4 × 4 × 4 k‐meshes within the Monkhorst–Pack scheme ^{[ 67 ]} was used, optimized by energy convergence tests based on 1 × 1 × 1 Bi₂O₃ unit cell. The convergence threshold for the electronic self‐consistent iterations was set to 10⁻⁶ eV/cell. The atomic positions and cell parameters were relaxed until the residual force reached 10⁻³ eV/Å. COHP analyses were performed using LOBSTER.^{[ 49 ]} The effective coordination number of Bi ions was calculated using as below:

$$\mathrm{Effective}\mathrm{coordination}\mathrm{number}=\sum _{\mathrm{i}}\exp \left(1-{\left(\frac{l_i}{l_{\min }}\right)}^6\right)$$ (2)

where l_i denotes to the bond lengths between the central Bi ion and nearby oxygen ion i while l _min denotes to the minimum bond lengths among l_i .

Conflict of Interest

The authors declare no conflicts of interest.

Author Contributions

D.L., H.K., S.J.J., and H.Y. contributed equally to this work. D.L., H.K., S.J.J., and H.Y. carried out experiments and analyzed the data. I.J. conducted the DFT calculations and supported part of the experiments. D.L., H.K., H.Y., I.J., W.J., and K.T.L. contributed to the writing of the paper.

Supporting information

Supporting Information

Lee D., Kim H., Jeong S. J., et al. “Anion Sublattice Engineering via Fluorine Doping to Enhance δ‐Bi₂O₃ Stability for Low‐Temperature Solid Oxide Electrochemical Cells.” Small 21, no. 41 (2025): 2503922. 10.1002/smll.202503922

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.