Impedance spectroscopy studies show that the K NiF -type La NiO is a promising cathode material for intermediate-temperature solid oxide fuel cells (IT-SOFCs) with a lower overpotential than that reported for La Sr MnO in the temperature range of 500–800 C . With its total conductivity of 80 S/cm (at 400 C) and intermediate thermal expansion coefficient (TEC) values, La NiO exhibited similar oxygen diffusion compared with that for La Sr Ga Mg O and La Sr Co Fe O . However, Skinner et al. reported that the activation energy of surface exchange in La NiO and La Sr NiO was approximately twice higher compared to the diffusion process . The surface exchange limitations in La NiO were evidenced by oxygen permeation studies : the maximal fluxes were obtained for the tape-casted 3-layer La NiO membranes with porous layers on both sides.
The A- and B-site doping of La NiO was shown to affect its properties, including oxygen diffusion and oxygen surface exchange [2, 4–8]. In general, acceptor-type A-site doping leads to a decrease in oxygen over-stoichiometry and an increase in electron hole transport due to partial oxidation of Ni to Ni . The A-site substitution in the range of x = 0–0.4 in La A NiO (A = Ca, Sr) drastically decreases oxygen-ion conductivity and slightly improves the surface exchange kinetics  resulting in inferior electrochemical properties of La A NiO compared to the undoped La NiO . The B-site substitution with donor-type dopants, such as Fe or Co, qualitatively produces the opposite effect on the properties of La NiO compared with that for the Ca/Sr doping . Introduction of these dopants does not have a significant effect on the oxygen-ion conductivity, although the surface exchange kinetics of oxygen is noticeably improved [7,8].
As shown for La Sr Ni Fe O , the simultaneous doping with strontium and iron eliminates the surface oxygen exchange limitations at x = 0.5, y = 0.4  improving its performance as a cathode material for SOFCs . Taking into account the recent results reported in , where an increase in strontium content from x = 0.2 to x = 0.8 in La Sr NiO decreased the area-specific resistance (ASR) of a symmetrical cell based on the Ce Sm O substrate almost fourfold (from 0.41 to 0.12 Ω/cm at 800 C), it is interesting to study bulk oxygen diffusion and surface oxygen exchange limitations in La Sr Ni Fe O and compare it with the studied earlier oxygen-excessive La Sr Ni Fe O (y = 0.1, 0.3, and 0.4) oxides .
The La Sr Ni Fe O complex oxide was synthesized by a citrate-nitrate technique using the following starting materials: La(NO ) 6H O ( 99%), SrCO ( 99.9%), FeC O 2H O ( 99.9%), Ni(CH COO) 4H O (99%), C H O H O (99%), and HNO (65%, 99.9%). The synthesis procedure can be found in . The obtained powders were uniaxially pressed into pellets (15 mm in diameter) and sintered at 1320 C for 20 h in air yielding relative density of 95%. The X-ray powder diffraction performed at room temperature (using an Equinox 3000 instrument, Cu–K radiation) revealed phase purity of the La Sr Ni Fe O membranes. The sintered pellets were mechanically polished to obtain the necessary thicknesses of 0.8 and 1.2 mm.
The preparation of cell for oxygen permeation studies and the experimental details were described in . The data were collected at 800, 850, 900, 950, and 1000 C in the oxygen partial pressure (p ) gradient interval of log(0.21/p ) = 0.2–1.8. The experiment with the 0.8 mm thick membrane was carried out twice to evaluate the reproducibility of the obtained results.
The steady-state oxygen permeation fluxes (j) through the dense La Sr Ni Fe O membranes with 0.8 and 1.2 mm thickness are shown in Figure 1(a) as a function of p gradient. As can be seen from Figure 1(a), the results of two parallel experiments with the 0.8 mm thick membrane illustrate good reproducibility. Figure 1(b) shows the specific oxygen permeability (J) that was calculated as follows:
(1) where d is a membrane thickness.
According to Eq. (1) the specific oxygen permeability, which is proportional to the j d product, will not depend on membrane thickness, if the oxygen permeation flux through a dense membrane is limited by bulk oxygen diffusion. As can be seen from Figure 1(b), the obtained J values were independent of the membrane thickness within the experimental error at 800, 850, and 900 C. However, at 950 and 1000 C a notable gain in permeability with increase in membrane thickness can be observed indicating significant surface exchange limitations .
In order to assess the role of bulk diffusion and surface exchange in the oxygen permeation flux, the critical membrane thickness (d ) can be calculated. The d value corresponds to the change of rate-limiting step from surface exchange to bulk diffusion (with increase in a membrane thickness) at fixed temperature and p gradient. It can be evaluated by the modified Wagner equation, which can be presented as follows :
(2) where p (= 0.21 atm), and p (= p ), denote the oxygen partial pressure and surface exchange coefficient at the feed- and permeate-side of a membrane, respectively; σ stands for ambipolar conductivity.
The critical membrane thickness, ambipolar conductivity, and the sum as functions of p gradient were calculated using Eq. (2) as described in . Figure 2(a) shows the obtained d = f(log(0.21/p )) dependencies. The critical membrane thickness was lower than 0.8 mm in the temperature range of 800–900 C, but it noticeably increased at 950 and 1000 C reaching values higher than 1.2 mm. Such behavior of d clearly indicated the change of rate-limiting step from bulk diffusion to surface exchange in the temperature range of 900–950 C. One should note that the change of rate-limiting step correlated with the beginning of active oxygen release from the La Sr Ni Fe O sample (at T 900 C) followed by a reduction of Ni to Ni and formation of oxygen vacancies . The latter observation hints that Ni /Ni ratio may affect the surface exchange in La Sr Ni Fe O .
The calculated values of ambipolar conductivity and the sum of reverse exchange constants are shown in Figures 2(b) and 2(c), respectively. As can be seen from Figure 2(b), the ambipolar conductivity possessed a relatively weak dependency versus p gradient showing a minor decrease in the range of 0.3 log(0.21/p ) 0.6. The latter fact allowed defining the σ values in air by extrapolating the corresponding dependencies to zero p gradient (Figure 2(b)). One should note that the σ values are approximately equal to the bulk oxygen-ion conductivity in La Sr Ni Fe O according to our previous studies [10,14]. Furthermore, the oxygen diffusion coefficients can be calculated using the obtained values of oxygen-ion conductivity and the data on oxygen non-stoichiometry from  by the Nernst–Einstein equation as it was performed in .
The reciprocal temperature dependencies of oxygen-ion conductivity (σ ) and oxygen diffusion coefficients (D) for La Sr Ni Fe O in air are plotted in Figures 3(a) and 3(b), respectively, together with the data for the oxygen-excessive La Sr Ni Fe O oxides reported earlier in . As can be seen from Figures 3(a) and 3(b), La Sr Ni Fe O and La Sr Ni Fe O showed comparable values of oxygen diffusion coefficients at all studied temperatures, while oxygen-ion conductivity in the former was lower in the 800–900 C temperature interval. Both, σ and D in oxygen-deficient La Sr Ni Fe O showed higher activation energies compared to that for oxygen-excessive La Sr Ni Fe O .
The surface oxygen exchange coefficients in air can be estimated if one assumes that at low p gradients and are approximately equal and, hence, the k value can be easily found from the corresponding sum . The latter can be determined as a fitting parameter in the model function, Eq. (2), presented in the form of logj = f(ln(p /p )), where σ σ is the extrapolated value at a given temperature. In this case, the sum should be constant or vary insignificantly in the given range of p gradients. For La Sr Ni Fe O , this is achieved only at relatively low p gradients in the temperature range of 900–1000 C (Figure 2(c)). Therefore, the model function was fitted to the experimental oxygen permeation fluxes in the range of log(0.21/p ) = 0.2–0.6 at 900, 950, and 1000 C. The results are shown in Figure 1(a) by the solid green lines indicating good agreement with the experimental data. Similarly, the k values were estimated for La Sr Ni Fe O using the data from . As can be seen from Figure 4, the activation energy for surface exchange in La Sr Ni Fe O was almost twice lower compared to that for the oxygen-excessive La Sr Ni Fe O oxides. The obtained results also indicated that the surface oxygen exchange kinetics for La Sr Ni Fe O should be similar to La Sr Ni Fe O in the intermediate temperature range (see the dashed lines in Figure 4).
The studies of oxygen permeation flux through dense La Sr Ni Fe O membranes with different thicknesses (0.8 and 1.2 mm) allowed calculating the critical membrane thickness. The obtained values indicated significant surface limitations in the oxide at 950 and 1000 C. The latter induced the change of rate-limiting step from bulk diffusion (800–900 C) to surface exchange (900–1000 C) upon temperature increase. The values of oxygen diffusion coefficients for La Sr Ni Fe O were comparable with that obtained earlier for oxygen-excessive La Sr Ni Fe O at all studied temperatures while the oxygen-ion conductivity was lower by almost one order in magnitude at 800 and 850 C. The calculated values of surface exchange coefficients showed that increase in strontium content from x = 0.5 to 0.8 reduced the activation energy of surface exchange almost twice leading to a similar surface exchange kinetics in La Sr Ni Fe O and La Sr Ni Fe O in the intermediate temperature range. One can conclude that La Sr Ni Fe O should be considered as a superior cathode material compared to the strontium-rich La Sr Ni Fe O : although both exhibited comparable surface exchange kinetics the latter showed significantly lower oxygen-ion conductivity values in the intermediate temperature range.
The authors would like to acknowledge that the equipment of the Ural Center for Shared Use `Modern nanotechnology' SNSM UrFU was used in this research.