Exploring the oxygen vacancies and electrical properties in pyrochlore-type Ca_1.46Ti_1.38Nb_1.11O₇

Material synthesis

Ca_1.46Ti_1.38Nb_1.11O₇ was synthesized using a solid-state reaction method. High-purity CaCO₃, TiO₂ and Nb₂O₅ powders (≥ 99.9%, Aladdin Biochemical Technology Co., Ltd., China) were used as starting materials, which were dried in an oven (DHG-9075A, Yiheng Scientific Instrument Co., Ltd., China) to remove adsorbed water prior to weighing. The starting materials were ground in anhydrous ethanol (AR, ≥ 99.7%, Sinopharm Chemical Reagent Co., Ltd., China) for approximately 30 min. The mixtures were pressed into pellets and pre-sintered at 1,000 °C for 12 h, and then sintered at 1,350 °C for 8 h. The sintered sample was ground, pressed into pellets, and finally sintered at 1,350 °C for 8 h for densification. All sintering processes were carried out in Ar gas (≥ 99.999%, Dalian Special Gases Co., Ltd., China) using a horizontal tube furnace (GYS-1700C, Hefei Weiyuans Laboratory Equipment Co., Ltd., China).

Material characterization

The X-ray diffraction (XRD) pattern was obtained using a powder diffractometer (X'Pert3 Powder, PANalytical, The Netherlands) with Cu Kα radiation. Rietveld refinement and Maximum Entropy Method (MEM) analysis were carried out using the Z-Rietveld software. BVSE calculations were performed using the softBV program to calculate the oxide-ion migration barriers. Visualization of the results was achieved using VESTA software. The elemental composition of the sample was determined by X-ray fluorescence spectrometry (XRF) using a wavelength-dispersive spectrometer (S8 Tiger, Bruker, Germany). Field emission scanning electron microscopy (SEM) (JSM-7900F, JEOL Ltd., Japan) equipped with energy-dispersive X-ray spectroscopy (EDS) was employed to investigate the morphology of the pellet samples and the elemental distribution of the powder samples. The particle size distribution of the sample was determined using a laser particle size analyzer (BT-9300H, Dandong Bettersize Instruments Co., Ltd., China). For electrochemical impedance spectroscopy measurements, pellets (approximately 10 mm in diameter and 2 mm in thickness) were used. Platinum paste electrodes (55H-1800, Shenzhen Saiya Electronic Paste Co., Ltd., China) were coated on its opposite faces. The sample was then heated at 1,000 °C for 2 h in a horizontal tube furnace (GYS-1700C, Hefei Weiyuans Laboratory Equipment Co., Ltd., China) to form dense platinum electrodes. Electrochemical measurements were performed on an electrochemical workstation (SP-300, Bio-Logic, France; or DH7002A, Donghua Analytical Instrument Co., Ltd., China) using the two-electrode method during the heating process. The sample was heated to temperatures between 500 and 900 °C in either a tubular furnace (JS-G5012-S, Tianjin Jiusuo Technology Development Co., Ltd., China) or a solid oxide fuel cell test station (HBSOC-5B, Zhejiang H₂-Bank Energy Technology Co., Ltd., China), followed by electrochemical measurements at each temperature. The process consisted of purging the system with various gases, including pure Ar, pure air, an Ar-H₂ mixture, an H₂-N₂ mixture, and an N₂-O₂ mixture. All gases (including the component gases in the mixtures) were of high purity (≥ 99.999%) and supplied by Dalian Special Gases Co., Ltd., China. The impedance was measured over a frequency range of 7 MHz to 0.1 Hz with an applied voltage of 100 mV. The obtained impedance data were fitted using Zview 3.1 software. The thermogravimetry (TG) curve was recorded on a thermogravimetric analyzer (STA 449 F3, NETZSCH-Gerätebau GmbH, Germany) over a temperature range of 80 to 1,000 °C. The gas flow rate was set at 20 mL/min, and the heating rate was set at 10 °C/min. High-purity Ar or air gas (≥ 99.999%, Dalian Special Gases Co., Ltd., China) was used during the measurement. X-Ray Photoelectron Spectroscopy (XPS) characterization was performed using an X-ray photoelectron spectrometer (Axis Supra+, Shimadzu, China) to analyze the chemical state of elements in the samples. The Al/Ag double anode X-ray source was used for excitation. The main peak of C1s at 284.8 eV was used as the reference for charge correction, and the spectra were analyzed using Avantage software. Electron paramagnetic resonance (EPR) spectra were recorded at room temperature using a spectrometer (E500, Bruker Corporation, Germany) operating in the X-band at a frequency of 9.42 GHz.

Structural and morphological characterization

After the first sintering at 1,350 °C, the room-temperature XRD pattern of Ca_1.46Ti_1.38Nb_1.11O₇ confirmed the formation of the target phase, along with a small amount of TiO₂ as an impurity. This suggests that sintering temperatures below 1,350 °C may result in higher impurity levels due to incomplete reaction. To eliminate the TiO₂ impurity and achieve densification, the material was subjected to a second sintering at 1,350 °C for 8 h. Figure 1 shows the refined XRD pattern obtained at 700 °C of Ca_1.46Ti_1.38Nb_1.11O₇ after the second sintering at 1,350 °C. All diffraction peaks can be indexed to the pyrochlore structure, indicating the formation of a pure phase. Table 1 lists the refined structural parameters obtained from in situ XRD data. Because the data were collected at 700 °C, they provide a more realistic representation of the crystal structure at high temperatures, which is essential for accurately predicting the ionic conduction mechanism under operating conditions. Ca_1.46Ti_1.38Nb_1.11O₇ is a typical pyrochlore structure with a space group of Fd-3m (No. 227, origin choice 2). There are two different kinds of oxygen sites, O1 (48f) and O2 (32e). The O2 site contains a high concentration of oxygen vacancies, while the O1 site exhibits a relatively low vacancy concentration. These parameters were then used to predict the influence of structural changes on the material's properties. The occupancies of Ca, Ti, and Nb were determined based on the stoichiometry of starting materials. The R_F is below 7%, and R_wp and R_p are below 4%, indicating that the refined crystal structure is in excellent agreement with the experimental data^[16]. XRF analysis was conducted to verify the elemental composition of the synthesized material. Quantification was performed using a standardless fundamental parameters (FP) method. The measured oxide contents are 24.5 wt% CaO, 32.3 wt% TiO₂, and 43.2 wt% Nb₂O₅. Based on these values, the calculated molar ratios of Ca, Ti, Nb and O are 1.46:1.35:1.09:6.88, which are very close to the nominal stoichiometry. As shown in Figure 2, the EDS mapping diagrams of Ca, Ti, Nb, O confirm that the elements distribute homogeneously without any apparent clustering or segregation^[31]. Combined with the XRD results, this confirms the formation of a single-phase Ca_1.46Ti_1.38Nb_1.11O₇. Figure 3A shows the cross-section SEM image of the Ca_1.46Ti_1.38Nb_1.11O₇ pellet. A relatively dense microstructure can be observed, with particle sizes visually estimated to be in the range of several hundred nm. The particles are needle-like in shape. The relative density of the sample reached 92.6% as measured by the Archimedes method, which is close to the upper limit typically achieved by conventional solid-state sintering (manual grinding, uniaxial pressing). This indicates that a sintering temperature of 1,350 °C is sufficient for Ca_1.46Ti_1.38Nb_1.11O₇ and that higher temperatures are unnecessary. However, the particle size distribution appears inhomogeneous in the SEM image. This observation is consistent with the particle size distribution measured by the laser diffraction particle size analyzer, which gives an average particle size of approximately 412.8 nm with a PDI of 0.371 [Figure 3B], confirming a broad size distribution. The inhomogeneity in particle size can be attributed to non-uniform temperature distribution within the pellet during sintering.

Figure 1. Refined XRD pattern of Ca_1.46Ti_1.38Nb_1.11O₇ obtained at 700 °C .

Figure 2. SEM image and corresponding elemental distribution maps of Ca_1.46Ti_1.38Nb_1.11O₇ powder.

Figure 3. (A) Cross-sectional SEM image of the Ca_1.46Ti_1.38Nb_1.11O₇ pellet; (B) Particle size distribution of Ca_1.46Ti_1.38Nb_1.11O₇ measured by laser diffraction.

Electrical properties

Conductivity measurements were first performed in Ar to both prevent oxidation of the material and to assess its electrical response following sintering in Ar. Electrochemical impedance spectroscopy was used to characterize the electrical properties of Ca_1.46Ti_1.38Nb_1.11O₇. The resistance in the high frequency range represents the grain resistance, the resistance in the mid-frequency range corresponds to the grain-boundary resistance, and the resistance in the low-frequency range can be due to the polarization effect of the electrodes. The impedance plots [Figure 4A-C] exhibit a single, nearly symmetric semicircular arc, which arises from overlapping grain and grain-boundary responses and can be modeled using an equivalent circuit with a single RC (resistance-capacitance) element. The total conductivity (σ_b+gb, representing the sum of bulk and grain-boundary contributions) of Ca_1.46Ti_1.38Nb_1.11O₇ is presented in the Arrhenius plot in Figure 4D. These values were derived from the impedance spectra measured in Ar, 5% H₂/Ar, and air atmospheres over the temperature range of 500-900 °C. The conductivity (σ) was calculated from the resistance using Equation (1)^[32],

where R, l, and A represent the resistance, the spacing between the probes, and the cross-sectional area of the sample, respectively.

Figure 4. (A) The complex impedance plot obtained at 600 °C in Ar of Ca_1.46Ti_1.38Nb_1.11O₇; (B) The complex impedance plot obtained at 600 °C in air of Ca_1.46Ti_1.38Nb_1.11O₇; (C) The complex impedance plot obtained at 600 °C in 5% H₂/Ar of Ca_1.46Ti_1.38Nb_1.11O₇; (D) The total conductivity σ_b+gb of Ca_1.46Ti_1.38Nb_1.11O₇ in the range of 500-900 °C in 5% H₂/Ar, Ar, and air (data from impedance curve fitting).

σ_b+gb of Ca_1.46Ti_1.38Nb_1.11O₇ increases gradually with increasing temperature from 500 to 900 °C, indicating a thermally activated ionic diffusion process. The conductivity of the sample varies approximately linearly with temperature in air. The temperature dependence of the conductivity in air can be described by Equation (2)^[33],

where σ, T, σ₀, E, and k_B are the conductivity, absolute temperature, pre-exponential factor, activation energy, and Boltzmann constant, respectively.

Across the entire temperature range, the conductivity follows the order: 5% H₂/Ar > Ar > air, indicating that the material exhibits n-type conduction. The conductivity of Ca_1.46Ti_1.38Nb_1.11O₇ was 4.20 × 10^-2 S cm^-1 in 5% H₂/Ar, 9.22 × 10^-3 S cm^-1 in Ar and 2.92 × 10^-5 S cm^-1 in air at 900 °C. Figure 5 demonstrates that the change in conductivity upon gas switching is reversible. This sensitive and reversible conductivity response to different atmospheres suggests that the material is a promising candidate for gas sensor applications. The DC conductivity σ_DC of Ca_1.46Ti_1.38Nb_1.11O₇ as a function of oxygen partial pressure (pO₂) is shown in Figure 6A. At 900 °C, σ_DC decreases with increasing pO₂, which is consistent with n-type conduction. σ_DC changes little with the oxygen partial pressure in the range of 1.0 × 10^-5 to 1.0 atm at 500 °C, indicating that the conductivity is primarily ionic at this temperature. Negligible proton conduction is observed because there is little difference in σ_DC of Ca_1.46Ti_1.38Nb_1.11O₇ under wet and dry air flows [Figure 6B]. Electromotive force (EMF) measurements were conducted to determine the ion transport number (t_ion) of the material under different atmospheres [Figure 7]. Using an air|N₂ gas concentration cell, the t_ion value is 0.995 at 500 °C, indicating that the material is an almost pure oxide-ion conductor at this temperature. The t_ion values in the entire temperature range are above 0.8, suggesting that oxide-ion conduction is dominant in oxidizing (air, O₂) and inert (Ar, N₂) atmospheres. In contrast, EMF measurements using an air|5% H₂/N₂ gas concentration cell revealed a drastically different behavior. The t_ion values are below 0.25 in the whole temperature range, indicating that the material exhibits predominantly electronic conduction under a reducing atmosphere (e.g., 5% H₂/N_2, 5% H₂/Ar). Figure 8 presents a comparison of the conductivity of Ca_1.46Ti_1.38Nb_1.11O₇ with that of recently reported pyrochlores. Compared with these pyrochlores, the conductivity of Ca_1.46Ti_1.38Nb_1.11O₇ is lower in air, comparable in Ar, and significantly higher in 5% H₂/Ar. σ_DC remains stable over 27 h [Figure 9A], demonstrating the excellent long-term stability of the material. No significant degradation in σ_b+gb is observed over 7 thermal cycles between 100 and 900 °C [Figure 9B], demonstrating the high structural stability of the material under repeated thermal cycling. These results indicate that Ca_1.46Ti_1.38Nb_1.11O₇ is suitable for high-temperature applications.

Figure 5. Reversibility of the conductivity of Ca_1.46Ti_1.38Nb_1.11O₇ upon switching between 5% H₂/Ar, air and Ar at 900 °C.

Figure 6. (A) Oxygen partial pressure pO₂ dependence of the DC conductivity σ_DC of Ca_1.46Ti_1.38Nb_1.11O₇ at 500 and 900 °C; (B) Arrhenius plots of σ_DC of Ca_1.46Ti_1.38Nb_1.11O₇ in wet (water vapor pressure of 0.042 atm) and dry air.

Figure 7. Ion transport number t_ion of Ca_1.46Ti_1.38Nb_1.11O₇ from EMF measurements using air|N₂ and air|5% H₂/N₂ gas concentration cells.

Figure 8. Comparison of the conductivity of Ca_1.46Ti_1.38Nb_1.11O₇ with that of recently reported pyrochlores^{[11,14,16,34,35]}.

Figure 9. (A) Time dependence of the DC conductivity σ_DC of Ca_1.46Ti_1.38Nb_1.11O₇ at 800 °C in air; (B) The total conductivity σ_b+gb in air of Ca_1.46Ti_1.38Nb_1.11O₇ from 600 to 900 °C during the thermal cycling process between 100 and 900 °C.

Conduction mechanism

Thermogravimetry

The TG curve of Ca_1.46Ti_1.38Nb_1.11O₇ in air [Figure 10] showed an overall mass gain with increasing temperature, reaching 101.13% of the initial mass up to 1,000 °C. The mass gain is attributed to the addition of oxygen atoms in the air into the lattice. The incorporation of oxygen atoms decreases the oxygen vacancy concentration, thereby reducing the conductivity. Ca_1.46Ti_1.38Nb_1.11O₇ exhibited negligible mass change upon heating in Ar [Figure 10], owing to the extremely low oxygen partial pressure. It can be inferred from the TG curve in Ar that no significant change in oxygen vacancy concentration occurred during the conductivity measurements from 500 to 900 °C in Ar. Therefore, the oxygen vacancy concentration in the material is significantly higher during conductivity measurement in Ar than in air, leading to a notably higher conductivity in Ar.

Figure 10. The TG curve of Ca_1.46Ti_1.38Nb_1.11O₇ in air and Ar.

X-ray photoelectron spectroscopy

To directly compare the oxygen vacancy concentrations in Ca_1.46Ti_1.38Nb_1.11O₇ under different atmospheres, the Ar-synthesized sample was subjected to annealing under contrasting atmospheres (in air and H₂). The samples before and after annealing treatment were then characterized by XPS for comparison. The wide-survey XPS spectra [Figure 11] show all characteristic peaks of Ca_1.46Ti_1.38Nb_1.11O₇. The core-level peaks observed at 530.0, 458.1, 346.9, and 207.1 eV correspond to O 1s, Ti 2p, Ca 2p, and Nb 3d, respectively, confirming the presence of these elements. No other elements were observed in the wide-survey spectra, indicating that no impurities were incorporated during the preparation process. The Ca 2p spectra [Figure 12A] show prominent peaks at 346.7 and 350.2 eV, corresponding to the Ca 2p3/2 and Ca 2p1/2 peaks, respectively^[36]. The spin-orbit splitting is 3.5 eV, and the chemical valence is +2. The Ti 2p spectra of both the pristine and the air-annealed samples [Figure 12B] exhibited prominent peaks at 458.4 and 464.3 eV, corresponding to the Ti 2p3/2 and Ti 2p1/2 peaks, respectively^[37]. A spin-orbit splitting of 5.9 eV was observed, common to both samples, corresponding to Ti⁴⁺. In the Ti 2p spectrum of the H₂-annealed sample [Figure 12B], the peaks observed at 456.6 and 462.8 eV are assigned to the Ti³⁺ 2p3/2 and Ti³⁺ 2p1/2, respectively^[38-40]. During H₂-annealing of Ca_1.46Ti_1.38Nb_1.11O₇, electron injection into the conduction band results in the partial reduction of Ti⁴⁺ to Ti³⁺, which contributes to the material's high conductivity in 5% H₂/Ar^[41]. This is consistent with EMF measurements under an air|5% H₂/N₂ gas concentration cell, which revealed that the t_ion values are below 0.25 across the entire temperature range, confirming predominantly electronic conduction under a reducing atmosphere [Figure 7]. The Nb 3d spectra [Figure 12C] show prominent peaks at 206.8 and 209.7 eV, corresponding to the Nb 3d5/2 and Nb 3d3/2 peaks, respectively^[42]. The spin-orbit splitting is 2.9 eV, and the chemical valence is +5. In the O 1s spectra [Figure 12D], the peaks near 529.7 eV represent metal-oxygen bonds (Ca-O, Ti-O, Nb-O), while the peaks near 531.6 eV represent oxygen vacancies^[43-48]. The spectral asymmetry suggests variations in metal-oxygen bonding and the presence of oxygen vacancies in the lattice. The corresponding computed percentage of oxygen vacancies is 45.2% before air-annealing and 31.7% after air-annealing^[11]. The higher percentage of oxygen vacancies leads to the higher conductivity, while oxygen vacancies provide better channels for oxide-ion conduction^[49]. After the Ar-synthesized sample was air-annealed, it exhibited a reduction in oxygen vacancy concentration and corresponding low conductivity. Following H₂-annealing, the lattice oxygen content in Ca_1.46Ti_1.38Nb_1.11O₇ was further reduced to 34.48%, as H₂ removed lattice oxygen in the form of water vapor. This led to the formation of oxygen vacancies, which are largely occupied by adsorbed oxygen species. The reduction process was facilitated by electron injection into the conduction band, weakening metal-oxygen bonds and promoting oxygen loss^[41]. The reversibility of the conductivity change upon atmosphere switching suggests that the changes of oxygen vacancies are also reversible [Figure 5].

Figure 11. Wide XPS survey spectra of Ca_1.46Ti_1.38Nb_1.11O₇.

Figure 12. XPS spectra of Ca_1.46Ti_1.38Nb_1.11O₇. (A) Ca 2p spectra; (B) Ti 2p spectra; (C) Nb 3d spectra; (D) O 1s spectra.

Electron paramagnetic resonance

Although sintering in Ar generated a large number of oxygen vacancies in Ca_1.46Ti_1.38Nb_1.11O₇, the valence states of the metal elements remained unchanged. This phenomenon, where oxygen vacancies form without reduction of the metal cations, was also observed in R-WO₃ by Zhang et al.^[43]. While the charge-compensating electrons from oxygen vacancies can be either trapped by metal ions or localized at the vacancy sites, Ca_1.46Ti_1.38Nb_1.11O₇ is an example of the latter^[41]. The EPR spectrum [Figure 13] shows a symmetric signal at g = 2.0014, which is consistent with electrons being trapped at oxygen vacancies^[43,50-56]. The weak intensity of the EPR signal indicates a very small number of unpaired electrons, suggesting a relatively minor contribution of electronic conduction in Ca_1.46Ti_1.38Nb_1.11O₇. This is supported by preceding EMF measurements under an air|N₂ gas concentration cell, which revealed that the t_ion values are above 0.8 across the entire temperature range [Figure 7]. When heating Ca_1.46Ti_1.38Nb_1.11O₇ in Ar, the generation of oxygen vacancies can be described by Equation (3)^[41],

(3) $$ \begin{equation} \begin{aligned} O_{O}^{x} \rightleftharpoons V_{\ddot{O} }+2 e^{-}+\frac{1}{2} O_{2} \end{aligned} \end{equation} $$

where $$O_{O}^{x}$$ represents a neutral lattice oxygen atom, $$V_{\ddot{O} }$$ represents an oxygen vacancy in the +2 charge state, and e^- represents electrons.

Figure 13. EPR spectrum at room-temperature of Ca_1.46Ti_1.38Nb_1.11O₇.

Accordingly, the equilibrium $$V_{\ddot{O} }$$ concentration, [$$V_{\ddot{O} }$$], can be described by Equation (4)^[41],

(4) $$ \begin{equation} \begin{aligned} K=\left[V_{\ddot{O} }\right] \cdot {pO}_{2}^{1 / 2} /\left[O_{o}^{x}\right]=\exp \left(\frac{-\Delta G_{f}}{k_{B}T}\right) \end{aligned} \end{equation} $$

where K is the equilibrium constant, k_B is the Boltzmann constant, pO₂ is the O₂ partial pressure, ΔG_f is the formation free energy of $$V_{\ddot{O} }$$, and [$$O_{O}^{x}$$] is the lattice oxygen concentration.

Based on Equation (4), a decrease in pO₂ leads to an increase in [$$V_{\ddot{O} }$$]. This provides a rationale for creating oxygen vacancies in a low-oxygen atmosphere or under vacuum.

Bond valence site energy calculations and maximum entropy method

Studying the oxide-ion migration pathways is crucial for understanding the electrochemical behavior of the material. BVSE calculations, based on Pauling's principle of local charge neutrality, were performed by placing the mobile ion at all points of a three-dimensional grid that spans the unit cell^[57]. This grid represents a fixed, rigid structural framework. The BVSE method has been extensively used to identify crystalline frameworks with infinite networks that facilitate the transport of ions. It also serves to rank materials based on the accessible volume within the crystal structure where the valence mismatch is low. The migration pathways are simulated through a two-step process: first, transition positions between equilibrium sites are identified by calculating the ion's valence state; then, the complete pathways are reconstructed by connecting these positions. The oxide-ion migration pathways in Ca_1.46Ti_1.38Nb_1.11O₇ were simulated by BVSE calculations. At an iso-surface value (oxide-ion migration barrier) of 0.55 eV, the O1-site atoms are interconnected [Figure 14A]. Simultaneously, O1 and O2 sites are connected. This indicates that oxide ion migration occurs through O1-O1 and O2-O1-O2 pathways within the lattice, confirming three-dimensional oxide-ion diffusion. The results highlight the importance of O1-site vacancies for oxide-ion migration, as O2-site atoms must pass through the O1 site during migration. Thus, creating oxygen vacancies at the O1 site will enhance the conductivity of O2-site atoms. The MEM electron density distribution of oxide ions along the migration pathway exhibits good agreement with the BVSE calculations [Figure 14B], thereby further confirming the reliability of the BVSE results. Notably, the oxide-ion migration barrier calculated in this work (0.55 eV) is lower than that derived from the crystal structure reported by Roth et al.^[27] (1.14 eV). This difference can be attributed primarily to the different conditions under which the XRD data were obtained. The structure reported by Roth et al. was determined from room-temperature XRD data, whereas our calculations are based on crystal structure parameters refined from high-temperature XRD data. Oxide-ion conduction is a thermally activated process. Therefore, the high-temperature structure provides a more accurate representation of the conduction behavior. This structure more accurately captures the migration pathways and local coordination environments under operating conditions, thereby providing a more reliable basis for the calculated migration barrier. The reliability of our refined result is further supported by the high quality of the Rietveld refinement, with R_F < 7% and both R_wp and R_p < 4%. These low residual factors ensure the accuracy of the refined crystal structure and further guarantee the credibility of the subsequent BVSE calculations. Therefore, the calculation results in this work represent the high-temperature conduction behavior of Ca_1.46Ti_1.38Nb_1.11O₇ more realistically.

Figure 14. BVSE and MEM calculations for Ca_1.46Ti_1.38Nb_1.11O₇. (A) The calculated BVSE map of oxide ions with an iso-surface at 0.55 eV, where the iso-surface is marked by yellow areas; (B) MEM electron density distribution for oxide ions, where the electron density is marked by pink areas.