Fluoride-Based Anion Doping: A New Strategy for Improving the Performance of Protonic Ceramic Conductors of the Form BaZrO3

A new strategy is reported to improve the performance of BaZrO3-based protonic ceramic conductors by substitution on the anion sublattice, specifically replacing F in the O site. The F -doped material exhibited higher conductivity under both reducing and oxidizing atmospheres compared to the undoped oxide. Ab initio calculations and X-ray photoelectron spectroscopy results showed that the improved performance can be attributed to weakened chemical bonding between the cations (A or B site) and the oxygen ions as a result of F doping, resulting in improved oxygen mobility. Finally, improved performance was demonstrated by using BaZr0.8Y0.2O3-δ-σFσ (σ= 0.1) as the electrolyte in protonic ceramic fuel cells measurements. The anions doping strategy shows great promise for the development of a new generation of high-performance electrolyte materials.


Introduction
Solid oxide fuel cells (SOFCs) are promising energy generation systems which have attracted worldwide interest due to their high efficiency, fuel flexibility and low pollution. [1][2][3][4][5] However, the high operating temperature (more than 800°C) for traditional oxygen ion SOFCs based on the yttrium stabilized zirconia (YSZ) electrolyte limits practical applications. Recently, SOFCs based on protonic conductors called protonic ceramic fuel cells (PCFCs) have attracted significant interest due to the lower operation temperature (below 700°C) resulting from the lower activation energy for protonic transport as compared to oxygen ion transport in the respective materials systems. [6] The state of the art electrolyte materials for PCFCs are BaCeO 3 and BaZrO 3 -based materials. [7][8][9][10] Doped BaCeO 3 shows a higher proton conductivity under humified atmosphere. However, it exhibits low chemical stability under CO 2 and H 2 O due to the large Brønsted basicity. [11] Doped BaZrO 3 exhibits lower proton conductivity while maintaining good chemical stability under water and carbon dioxide containing atmosphere resulting from its shorter and stronger ZrÀ O bonds (in comparison with CeÀ O) and practically an ideal combination between ZrÀ O and BaÀ O bond distances. [12] A large number of elements were selected as the dopants in BaZrO 3 to improve its proton conductivity or transport characteristics such as yttrium, ytterbium, thulium, scandium, holmium and Zinc. [13][14][15][16][17] Among them, yttrium is considered as the most promising dopant. For example. Tong et al. developed a cost-effective solid-state reactive sintering method to fabricate dense BaZr 0.8 Y 0.2 O 3-δ (BZY) membranes with high proton conductivity. [18] In addition, Bi et al. successfully fabricated BZY electrolyte films by utilizing the driving force from anode substrate to obtain a dense electrolyte at temperatures near 1400°C. [19] To date, the doping strategy has focused on the doping cations in the A or B site. In this work, we report a new route for improving the conductivity of BZY via doping F À in the anion sublattice at oxygen sites. Fluoride has a higher electronegativity than oxygen, which will lead to the attraction of electrons. As a result, doping F À into the oxygen sites may reduce the valence electron density of oxygen ions, weaken the chemical bonds between cations and oxygen ions, and lead to faster oxygen mobility in the bulk oxide. The improved oxygen mobility should contribute to the improvement of protonic mobility and conductivity. This strategy has been widely studied in mixed conductors. For example, Zhu et al. reported unprecedented oxygen permeation membrane by doping F À in SrCo 0.9 Nb 0.1 O 3-δ . [20] High performance perovskite cathode for SOFCs was developed by Zhang et al. through embedding anions (F À ) in oxygen sites. [21] Also, Chen et al. found the higher hydrogen permeability and better stability hydrogen separation membranes can be achieved by doping Cl À in La 5.5 W 0.6 Mo 0.4 O 11.25 . [22] This suggests that through anions doping, the properties and performance of a wide range of proton conducting ceramics may be tailored.
However, so far, only a small number of reports have examined the strategy in protonic ceramic condutors. Tarasova et al, found that the mobility of oxygen and proton were improved by doping small amount of F À in Ba 2 In 2 O 5 . [23] Su et al. demonstrated a stable electrolyte by doping F À in barium cerate oxides. [24] Higher conductivity was observed for the Br À doped BaCe 0.8 Y 0.2 O 3-δ but with poor stability than the parent perovskite. [25] In this work, we demonstrated that anion doping can enhance the performance of BaZrO 3 -based protonic ceramic conductors. F À was selected as the dopant for O 2À site in BZY. We found an enhanced conductivity under both reducing and oxidizing atmospheres, and the improvement was contributed by the increased oxygen mobility in the lattice. The results show that anions doping is a promising way to tailor the properties of BaZrO 3 -based protonic ceramic conductors which are the compositions of interest for a wide range of energy conversion and storage devices at present. Figure 1 displays the X-ray diffraction (XRD) patterns in BaZr 0.8 Y 0.2 O 3-δ-σ F 0.1 (BZYF 0.1 ), indicating phase pure perovskite with no BaF 2 secondary phase detected. All samples exhibited a cubic structure, which suggest that F À were successfully doped into the oxide lattice for BZYF 0.1 . In addition, patterns show that doping F À induces a shift in diffraction peaks to higher 2-theta, which corresponds to a decrease in the lattice parameter with F À dopant. Figure 2 shows a representative Nyquist plot of BaZr 0.
As expected, addition of F À will reduce the oxygen vacancies in BZY. However, fluoride has a higher electronegativity (4.0) than oxygen (3.44), which allows it to attract electrons more strongly than oxygen. This would result in a decrease in the metal-oxygen bond energy, making oxygen ions easier to be displaced from neighboring cations contributing to the formation of additional oxygen vacancies. Clearly, BZYF 0.1 has higher conductivity compared to un-doped materials in the different atmospheres examined in this study. In wet nitrogen, the higher total conductivity resulting from the contribution of protons originates from the increased concentration of oxygen vacancies or increased mobility of oxygen lattice, as shown by Eq. (4) Improved oxygen ion and proton conductivity were achieved under dry and wet air after introduction of F À in the oxygen lattice. The improved proton conductivity is determined by the oxygen vacancy concentration and the mobility of ionic charge carriers in the perovskite.
Oxygen vacancy concentration is one factor that affects the proton conductivity in wet atmosphere. Thermal expansion behavior analysis was used to identify the oxygen vacancies after F À doping. According to Tor S. Bjørheim's calculation work, [26] there appears to be a correlation between the oxygen vacancy concentration with the linear thermal expansion over a certain temperature range. As shown in Figure 4a, BZYF 0.1 has a lower thermal expansion than the parent BZY sample from 400 to 1000 K, which indicates that doped F À in BZY had a decreased oxygen vacancy concentration. This observation was consistent with the calculated thermal expansion coefficient by first-principles density functional theory (DFT). As show in Figure 4b, BZYF 0.1 has a lower calculated thermal expansion coefficient. This result suggests that the introduction of F À in the lattice will decrease the oxygen vacancies according to the Kröger-Vink notation. Therefore, the observed improvement in the proton conductivity in F À doped BZY membranes is not a result of the increased oxygen vacancy concentration. Rather, the increased mobility of oxygen and oxygen defects is likely the main reason for the observed performance enhancement.
To understand the increased oxygen mobility in BZYF 0.1 membrane, XPS was used to analyze the binding energies of oxygen in BaZr 0.8 Y 0.2 O 3-δ-σ F σ (σ = 0, 0.1). The binding energy of the oxygen species absorbed on the surface is often higher than that of the oxygen lattice. [27] Figure 5 displays the XPS results indicating the O 1s binding energy of the lattice oxygen species is 526.88 and 527.46 eV for pristine BZY and doped BZYF 0.1 , respectively. The increased binding energy may result from the difference in electronegativity between F À and O 2À , which will weaken the Coulombic force between A or B-site ions and O 2À , indicating an increased activity of the lattice oxygen. Notably, the decreased binding energy for metal ions indicates a lower bond energy for metal and oxygen in the perovskite oxides, thus generating faster oxygen mobility in the bulk material. Apparently, the enhanced proton conductivity under wet nitrogen and air originated from the increased mobility of the oxygen lattice and oxygen defects.
The core level of O1s was calculated and shown on Figure 6. The binding energy increased by 0.1 eV after F À doping, which is consistent with our XPS results. The difference in the oxygen mobility in perovskite material can account for the improved proton conductivity in BZYF 0.1 . This observation is consistent with current literature regarding anion doping on enhancing the mobility of ionic charge carriers as has been observed in other solid-state material systems such as Ba 2 In 2 O 5 . [23] Chemical stability is another important factor for proton ceramic conductors. The sintered pellets were ground into powders and underwent thermogravimetric analysis (TGA) measurements in air and CO 2 , as shown on Figure 7. There was an apparent weight loss around 250°C, resulting from the absorbed water in the samples. The BZYF 0.1 sample maintained good chemical stability both at air and CO 2 at high temperature range, indicating the F À doping in the perovskite materials does not deteriorate the chemical stability of BZY.
The improved oxygen activity or proton conductivity was further confirmed by investigating the single cells' performance by using BZY and BZYF 0.1 as electrolyte. Figure 8 shows the SEM micrographs of single cells with BZYF 0.1 as the electrolyte. The representative porous electrodes with 200 nm grain size exhibited good interfacial connection with the electrolyte and revealed a fine microstructure, no apparent pores were observed in the dense electrolyte (50 μm).
We measured the power density of single cells at temperatures ranging from 550-700°C. Figure 9 shows the representative performance of single cells with BZYF 0.1 as the electrolyte at different temperatures. In the EIS plot Figure 9b, single cells with BZY and BZYF 0.1 as the electrolyte displayed a similar polarization resistance due to the similarity in fabrication process and electrode materials. The only difference observed between the cells was the ohmic resistance resulting from the improved proton conductivity in BZYF 0.1 attributed to the factors discussed in this manuscript.

Conclusions
In conclusion, we have successfully prepared dense oxyfluoride perovskite BaZr 0.8 Y 0.2 O 3-δ-σ F σ (σ = 0.1) membranes by a solid-state reaction sintering method. The targeted substitution of F À , for O 2À on the anion sublattice resulted in enhanced proton conductivity originating from the improved oxygen mobility comparing to the parent un-doped oxide. The suggested underlying mechanism is the weakened chemical bonds between cations and oxygen ions in perovskite materials after F À doping. Finally, improved performance was demonstrated by using BaZr 0.8 Y 0.2 O 3-δ-σ F σ (σ = 0.1) as the electrolyte in single cell

Cell Fabrication
40 wt.% BaZr 0.8 Y 0.2 O 3-δ-σ F σ (σ = 0, 0.1) + 60 wt.% NiO anode precursor powder with starch as the pore former were prepared by the same mixing and drying process used for the electrolyte precursor. The anode precursor powder was dry pressed under 250 MPa for 1 min in circular die with the diameter of 15 mm to produce green anode pellets. Electrolyte precursor powder was mixed with binder (5 wt% V-006 (Heraeus) dissolved in terpinol) and dispersant (20 wt% solsperse 28000 dissolved in terpinol) with a ratio of 15 : 3 : 1 to form electrolyte slurry that was deposited on each side of the green anode pellets by screen-printing, followed by co-sintering at 1500°C for 24 h. PBSCF slurry was prepared by mixing the PBSCF powder with binder and dispersant. The as-prepared cathode slurry was printed on the electrolyte side and sintered at 950°C for 4 h to obtain single cells.