Cation diffusion and related degradation phenomena in La2Ni04+δ for oxygen separation membranes

Abstract

Oxides with layered perovskite-like K2NiF4-type structure have attracted significant attention since the discovery of high temperature superconductivity in La2CuO4+δ based systems. Further investigation of this type of oxides has shown that their mixed ion and electron conductivity makes them interesting candidates for high temperature applications, such as cathode material for Solid Oxide Fuel Cells (SOFCs) or as membrane material for gas separation. Besides their electrical features, these oxide materials have to have other suitable characteristics, such as thermal expansion, chemical stability, etc. Furthermore, when these materials are exposed to harsh working conditions, i.e., high temperature and large oxygen potential gradient, it induces cation transport towards the side with higher partial pressure of oxygen (pO2) which has significant influence on the lifetime of the material. Cation diffusion may cause kinetic demixing and kinetic decomposition due to the different diffusivities of different cations, and morphological instability (membrane walk-out) due to the high diffusivities of the cations. The main topic of this thesis is the determination of the cation diffusion and degradation phenomena in oxides with K2NiF4-type structure (general formula A2BO4), notably lanthanum nickelate, La2NiO4+δ (LNO).

Structure and transport properties of the undoped and doped LNO have been the subject of extensive research. Although cation diffusion is one of the properties of fundamental interest for the durability of components serving under chemical potential gradients, data on cation transport and degradation processes is generally scarce, particularly for materials with K2NiF4-type structure. To date there is no data reported on cation diffusion in LNO-based systems.

This thesis comprises studies of A- and B-site cation diffusion in the LNO system, utilising inter-diffusion, chemical tracer diffusion and solid state reaction (SSR) methods. The study of the degradation of the LNO membrane under the gradient of pO2 is also included in this work. Diffusion couples between La2NiO4+δ and Nd2NiO4+δ and between La2NiO4+δ and La2CuO4+δ were used in the inter-diffusion method for the estimation of A- and B-site inter-diffusion coefficients, respectively. The chemical tracer diffusion method employed Pr, Nd and Co as impurity tracers for the study of A- and B-site cation diffusion, respectively. The self-diffusion coefficient of the fastest moving cation was determined by SSR between the binary oxides NiO and La2O3. The samples were analysed extensively using Scanning Electron Microscopy (SEM) and its additional features (Energy-dispersive X-ray spectroscopy (EDX) and Electron backscatter diffraction (EBSD)). Secondary Ion Mass Spectrometry (SIMS), Electron Probe Micro Analysis (EPMA), X-ray Diffraction (XRD) and X-ray Photoelectron Spectroscopy (XPS) were the methods of instrumental analysis used in this work.

The inter-diffusion experiment showed that the A-site cation diffuses mainly through bulk, while the B-site cation is faster due to enhanced grain boundary diffusion. The diffusion paths of the A- and B-site cations were visualised by EPMA element mapping of the crosssection of the diffusion couples. The obtained activation energies for diffusion are 275 ±12 kJ/mol for the A-site cation (La) bulk, 450 ±20 kJ/mol for the B-site cation (Ni) bulk, and 125 ±24 kJ/mol for the grain boundary.

The chemical tracer experiment revealed kinetics of diffusion of the Pr, Nd and Co chemical (impurity) tracers in LNO. By means of analyses of the SIMS data it is shown that the A-site cation (Pr and Nd) diffuses mainly through bulk, meaning that according to Harrison’s classification of diffusion kinetics it belongs to the “A”-type, and that the B-site cation (Co) has enhanced diffusion along grain boundaries and exhibits “B”-type diffusion kinetics. The average apparent activation energies for the diffusion of chemical (impurity) tracers are evaluated as 165±15 kJ/mol for Pr and Nd in bulk, 295±15 kJ/mol for Co in bulk, and 380 ±20 kJ/mol for Co in grain boundaries. The unexpectedly high activation energy of the fast Co grain boundary diffusion in comparison to bulk is also discussed.

Through the use of inert platinum markers, the SSR confirmed that the fastest moving cation in LNO is Ni. The growth of the reaction layer between NiO and La2O3 was parabolic, as given by Wagner’s parabolic rate law. The obtained self-diffusion coefficient of Ni shows Arrhenius behaviour at higher experimental temperatures (1100-1450 °C), with activation energy of 243 ±21 kJ/mol. At lower experimental temperatures (950-1100°C) the self-diffusion coefficient deviates from Arrhenius behaviour by bending downward and attaining apparently higher activation energies. Some suggestions for this rather unusual behaviour are given in this thesis. In addition, the measurement shows that the Ni self-diffusion coefficient at the higher temperatures is independent of the surrounding pO2, indicating that the thin product layer formed between the NiO and La2O3 reactants cannot equilibrate fast enough with the surrounding atmosphere at these temperatures.

When an LNO membrane was exposed to a gradient in pO2 for 800 h at 1000 °C, the high pO2 side of the membrane became decorated by precipitations of NiO. This was expected because the results from the inter-diffusion, chemical tracer and SSR experiments had shown that the B-site cation (Ni) exhibited enhanced diffusion over the A-site cation (La). This decomposition experiment also showed signs of “walk-out” toward the side with higher pO2. At the membrane’s low pO2 side porosity and etched-away structure developed, reflecting the loss of material from this side of the membrane.

The different data for cation diffusion in LNO obtained by the above mentioned methods agrees, showing similar cation diffusion within anticipated uncertainties. Anyhow, LNO exhibits rather high diffusion values for both A- and B-site cations, even when compared with the data for cation diffusion of related perovskite compounds. It is demonstrated here that long-term utilisation in a gradient of pO2 can severely degrade and perhaps eventually break the membrane due to “walk-out” and decomposition.