Solid oxide fuel cells (SOFC) have emerged as a promising high-efficiency and low or zero-carbon power generation technology. Yttria-stabilized zirconia (YSZ), invented by Nernst in 18991, was the predominant choice of electrolyte material for SOFC for years. However, its inherent insufficient oxygen ion conductivity necessitates high operating temperatures over 700 °C2,3, which poses significant challenges, e.g., material compatibility issues and costs, hampering the commercialization of SOFC.
To overcome these challenges, alternative electrolyte materials have been explored3-5. Fluorite ceria has shown a significant ionic conductivity value of 0.01 S/cm above 600°C, but it suffers from electronic conduction with a risk of micro-crack formation1,6.
In parallel, the discovery of proton-conducting perovskite BaZrO3/BaCeO3 in the 1980s opened up a new field of research on low-temperature SOFC7. However, these perovskite electrolytes still exhibit limited ionic conductivity of 10-3-10-2 S/cm at 600°C8 and require sophisticated ultra-thin membrane fabrication technologies. The proton conductivity is still a significant challenge for electrolytes.
To advance in the next-generation SOFC, electrolyte compounds with ionic conductivity of 0.1 S/cm in the temperature range of 300-500 °C would be highly desirable. This study introduces a novel alumina-based material with a radically new fluorite phase as a potential solution to this quest. Alumina is a versatile material with various structural phases (including alpha, beta, and gamma) for broad applications because of its characteristics of high melting point, strong mechanical strength, and chemical stability9. However, the close-packed lattice structure in the pure phase enables negligible ionic conductivity, triggering rare interest as an electrolyte component in SOFC10,11. An exception is the well-known beta-alumina, a unique layer structure material formed with sodium for wide use as Na+ conducting electrolyte in sodium-sulfur battery12.
Herein, we successfully synthesized fluorite alumina-based material with high ionic conductivity beyond 0.1 S/cm at 500°C and explored it for advanced low-temperature SOFC at 300-500°C. This fluorite alumina native structure, without doping, possesses significantly higher oxygen vacancy concentrations, which are 5-10 times that in fluorite YSZ and Gd or Sm-doped ceria, resulting in inherently high ionic conductivity. We conducted fuel cell electrochemical characterization to analyze its ionic transport properties and assess its potential and suitability for next-generation low-temperature SOFC.
Through a nitrate-critic acid approach, the synthesized alumina samples unveil a distinct fluorite structure, though pure Alumina exhibits a strong amorphous background (Fig. 1a). The XRD peaks of fluorite alumina show minor deviations from the reference structure, such as CeO2; the introduction of high-valency Ce4+ doping into Alumina remarkably eliminates the amorphous background, resulting in a well-defined fluorite crystalline. In typical aluminum oxide phases, coordination numbers 4 and 6 with oxygen constructs tetrahedral and octahedral arrangements, as observed in alpha and gamma Al2O3, underpins stable configurations within the crystal lattice. Intriguingly, a recent study has also reported the presence of 5-coordination in γ-Al2O313. These reports demonstrated that Al-coordination numbers are flexible in Alumina, allowing the fluorite structure of Alumina, which has never been reported before.
It is quite puzzling how Al2O3 (with an Al:O ratio of 1:1.5) can adopt the fluorite structure to form the typical apparent chemical AlO2 (with an Al:O ratio of 1:2), where an 8-fold coordination to oxygen atoms should be taken in the fluorite structure. To gain insight into such transformation, we draw upon the reference of the CeO2 fluorite structure. The fluorite structure is formed with cation Ce4+ and anion O2- in a 1:2 ratio. In this approach, if we use four Al3+ cations to replace Ce 4+ ions within the structural unit (Fig. 1b), then to maintain the charge neutrality, two oxygen anions must be removed from the unit cell to maintain the fluorite structure due to four positive charges loss by Al3+ substitution. This creates two oxygen vacancies (Fig. 1b). Hence, these vacancies correspond to 25% of all oxygen ions as each unit cell contains 8 oxygen ions, one-quarter less to form fluorite alumina.
Consequently, we can represent fluorite alumina with a general chemical formula AlO2-d, where d = 0.5 as a standard notation, but in practice, non-stoichiometry’s takes place. This uniqueness of fluorite alumina is extended to Ce4+-doped fluorite alumina (Fig. 1b–d) as being validated by the neutron scattering (NS) experiments (Extended data Fig. 1a). The NS validated the fluorite structure and determined the oxygen occupation of 73% (Extended data Fig. 1b), well in agreement with a theoretical value of 75% of fluorite Alumina.
HR-TEM micrographs of fluorite alumina and Ce-doped Alumina are depicted in Fig. 1c and e. The interplanar d-spacing of the fluorite alumina (labeled as AlO2-d) at 0.2857 nm and Ce-doped Alumina leads to an enlarged interplanar spacing of 0.307 nm, corresponding to the [111] lattice plane with space group Fm3m. Some more details are included in Extended data Fig. 2. XPS analyses unveil oxygen vacancies in the fluorite alumina and Ce-doped Alumina (Fig. 1f-g). For Alumina in the fluorite structure, a 22% oxygen vacancy is disclosed by XPS to be close to the theoretical value of 25%, as discussed above. In Ce-doped Alumina, oxygen vacancy concentration reduces (14%), corresponding to high valance Ce4+ incorporation by recovering oxygen in the lattice to maintain electrical neutrality (Fig. 1g). EPR also supports significant oxygen vacancy presence in alumina fluorite, distinct from other aluminum oxides such as alpha and gamma-Al2O3 (Fig. 1h).
Raman analysis results show that a distinctive F2g Raman mode at 464 cm-1 emerges prominently (Fig. 1I, and Extended data Fig. 3a-b), echoing the typical CeO2 Raman spectrum14. This F2g mode distinctly represents the specific vibration of oxygen atoms within the fluorite structure of AlO2-d, providing strong evidence for adopting a fluorite structure in Alumina.
This oxygen non-stoichiometric alumina discovery is remarkable because the commonly used doped ceria, such as 10GDC/20SDC, with optimized ionic conductivity, have only 0.2 and 0.4 oxygen vacancies per unit cell created by foreign atom doping in their fluorite structure and are similar in the YSZ case. While two native oxygen vacancies of the unit cell (Fig. 1b) in the fluorite alumina could significantly improve the ionic conductivity. It is a common approach to create a high content of oxygen vacancies through ion doping in the traditional YSZ or doped ceria fluorite structures. It is a fundamental requirement for high ionic conductivity and high-performance SOFC6, but it is an inherent material nature for fluorite alumina without doping.
However, such a high oxygen vacancy concentration and minimized lattice oxygen can collapse local structures, leading to a conspicuous amorphous background (Fig. 1a) to indicate a less-defined crystalline structure. We further investigated the process of AlO2-d structural formation, in which a significant structural relaxation emerged as a pivotal factor. Extended data Fig. 4 illustrates repeated syntheses of AlO2-d, revealing some cases of an amorphous alumina structure appearance, while in most cases, the distinct peaks of the fluorite structure appear superimposed on a robust amorphous backdrop.
These findings offer compelling evidence of a structural transition process during the evolution from amorphous to crystalline Alumina. Notably, this structural relaxation can be adeptly manipulated by introducing higher-valency Ce4+ ions, which are homogeneously distributed (Extended data Fig. 2g-j), to substitute Al3+. Fig. 1a also shows a transformation in that all Ce4+-doped alumina samples feature well-defined crystallized fluorite structures, and the amorphous background gradually disappears with increased Ce4+ doping content. This Ce4+ cation doping modulates the structure of the anion lattice by reintegrating oxygen into the lattice, thus mending collapsed regions and stabilizing the fluorite structure. Therefore, incorporating higher-valency Ce4+ ions into the AlO2-d structure can restore a finely crystallized fluorite structure and structure durability as demonstrated by the temperature dependence of XRD up to 800 oC (Extended data Fig. 5 for Ce-doped Alumina).
Fig. 2 comprehensively compares fluorite electrolyte generations across the SOFC research field. Distinct characteristics emerge among three types of fluorite electrolytes: YSZ (1st Gen), GDC/SDC (2nd Gen), and AlO2-d (3rd Gen). Theoretical calculations reveal YSZ and doped ceria with the density of electronic states (DOS) as wide bandgaps, exceeding 4.30 eV for YSZ and 2.57 eV for GDC, respectively, standing out with their insulating traits. In contrast, the Fermi level aligns at the valance band's top or conduction band's onset for AlO2-d and Ce-doped fluorite alumina, implying a metallic- state of their characteristics, Since the theoretical calculations are based on atom level, while practical particles' structural formation takes place in especially inherent disorder or defects within the crystal structure. This disruption ushers in localized states in the electronic density of states, leading to very different situations for the obtained fluorite alumina materials with the emergence of insulating oxide phases. Indeed, oxygen-driven or disorder effects (Anderson type) metal–insulator transition are commonly observed in the literature17,18. Anyway, our theoretical calculating result gives us a meaningful vision and hints that distinctly different electronic states and band structures exist in the alumina fluorite structure. While we further calculate that Ce4+ doping partially tunes local ordering, it simultaneously shifts the material away from its more metallic traits toward regulated fluorite characteristics. However, as illustrated in Fig. 2d, owing to the distinct differences in DOS and bandgap, the fluorite AlO2-d demonstrates a notably improved ionic conductivity (0.186 S/cm at 500 °C) calculated using I-V curve and reduced activation energy of 0.33 eV for ionic transportation, which is quite different from that of YSZ, GDC, and SDC15,16. An impressive ionic conduction activation energy, 0.33 eV, of fluorite AlO2-d is achieved, which is remarkably lower than those of the doped ceria and stabilized zirconia. Extended data Fig. 1b-c offer additional insights into the oxygen vacancy per unit cell and its variance, particularly about the fluorite electrolyte by commonly lower-valency Sm3+-doping to create oxygen vacancy and the unusually higher-valency Ce4+-doped fluorite alumina to reduce the oxygen vacancy and regulate the alumina fluorite structure.
Due to superior structural characteristics, fluorite alumina offers excellent electrochemical performance as the electrolyte for low-temperature (300-500) fuel cells (Fig. 3). All these alumina-based fluorite electrolyte fuel cells exhibit exceptional power densities within the range of 800-1057 mWcm-2 at 500 °C (Fig. 3a). Notably, there is a trend of increased power output with higher Ce dopant concentrations. In the case of the 30% Ce-doped alumina (30CDA), we achieved the best power output of 1057 mW cm-2 at 500 oC (Fig. 3a and Extended data Fig. 6) which illustrates further detailed electrical and electrochemical performances for an optimized fluorite sample. Even at 370 oC, the SOFCs showed a power density of 259 mW cm-2.
Subsequent investigations show that the fluorite alumina electrolyte presents co-H+/O2- transportation under a typical fuel cell atmosphere (Fig. 3). With fuel cell setup under controlled small gas flow (from an air/air to H2/air environment), in-situ activation of material electrical properties for precise measurements through electrochemical impedance spectroscopy (EIS) at the regular 3-minute interval was observed (Fig. 3b, c) and Fig. S6. Initially, the conductivity was approximately 1.3*10-3 S/cm in air. As H2 replaced one side of the air supply, the resistance rapidly decreased over time, indicating that continuous hydrogen exposure is crucial in augmenting total ionic conduction, indicating an electrochemical proton injection process19,20. This transition signifies the shift from inherent O2- to co-O2-/H+ transport, substantially increasing total ionic conductivity. In fact, after 94 minutes, the conductivity remarkably escalated to 0.15 S/cm (Fig. 3d). The resulting ionic conductivity at this lower temperature (500oC) is orders of magnitude higher than that of YSZ/GDC/SDC. These materials usually exhibit ionic conductivities of 10-3-10-2 S/cm at the same temperature15,16. Proton transport property and evidence have been further proved by i) determining H+ conducting fuel cell property by using a proton conductor, BZCY, as the ionic filter. The result shows a 80% fuel cell performance compared to the filter free one (Extended data Fig. 7a), and Extended data Fig. 7b provides further evidence on proton conduction in the material based on isotopic effect measurements. The presence of a clear 1H signal in Solid-state NMR after proton injection (Extended data Fig. 8) and the significant effect on Alumina by NMR 27Al signal (Extended data Fig. 9), suggesting significant changes and modifications in the alumina lattice induced by the deep proton injection.
Extended data Fig. 10 illustrates our proton injection process through an electrochemical fuel cell device, where the time-dependent distribution of relaxation times (DRT) analysis on EIS for the 30Ce-doped Alumina is presented in Fig. 3e-f. In the air atmosphere, P1 and P2 at 9135.10 and 1392.67 Hz manifest the charge transfer in the electrode and the oxygen-ion transportation (Fig. 3e)21. After 50 min proton injection (Fig. 3f), P1 is convoluted, causing the extra electrons in the reduction atmosphere to make a more excellent electronic conduction. At the same time, Pnew1 emerged at 12651.08 Hz, which represents proton transportation. There is not much change in P2. Pnew3-Pnew5 emerge at 189.45, 22.33, and 0.80 Hz, representing the related process electrode22-24. It indicates the coexistence of proton conduction coupled with intrinsic mobile oxide ions. Proton conduction is more pronounced than oxygen ion one because of the higher frequency and smaller peak area of Pnew1, as previously elucidated. After 94 min proton injection, the frequency where Pnew1 is located continues to get larger, from 12651.08 to 15697.05 Hz, indicating the ion conduction process is still accelerating. This corresponds to a rising conductivity, as shown in Fig. 3d. These variations might be attributed to adjustments in proton concentration and pathways across the sample's bulk and surface. At the same time, the area of Pnew1 and P2 are very similar. This is most likely because proton conduction has a positive effect on oxygen ion conduction during the operation of the cell, and eventually, the two achieve equilibrium. Extended Fig.10 illustrates proton injection through an in-situ fuel cell electrochemical process and the proton-coupled oxygen vacancy diffusion mechanism, representing a synergistic effect on H+/O2- transport within fluorite alumina material25,26. We provided further evidence of alumina-based fluorite acting as electrolyte materials for LT-SOFC operated below 450 oC for practical applications (Extended data Fig. 6-7 and a video attached).
This study introduces a cutting-edge finding for SOFCs operated at 300-500 oC based on the successful synthesis of fluorite alumina with inherent high oxygen vacancy (Ov) concentration and subsequently exceptional ionic conductivity, offering distinct advantages over traditional fluorite electrolytes such as YSZ and doped ceria. Unlike traditional fluorite electrolytes using lower valency cation doping, high-valency Ce4+ doping can intensify the structure, ionic transport properties, and exceptional power outputs at low temperatures. In addition, the proton associated with the mobile oxygen ion transport mechanism is interesting, resulting in outstanding ionic conductivity and fuel cell performances and markedly differing from the transport behaviors found in classic fluorite electrolytes based solely on O2- conduction.
The findings of this study set a new pathway for advancing functional materials development and next-generation fuel cell technology.