Project

Grain boundary diffusion in (La,Sr)FeO3

(La,Sr)FeO3 is a well-known mixed ionic and electronic conducting perovskite oxide, with application in Solid Oxide Fuel Cells (SOFCs) and Solid Oxide Electrolysis Cells (SOECs). The theoretical investigation of transport processes in such materials has emerged with the development of the first atomistic simulation codes. However, limited computational resources have restricted calculations to simple structural models, short molecular dynamics simulation times, and only a few temperatures.

Project Details

Project term

May 2, 2022–August 24, 2023

Affiliations

RWTH Aachen University

Institute

Institute of Physical Chemistry

Principal Investigator

Prof. Dr. Roger A. De Souza

Methods

In this project we employed molecular-dynamics (MD) simulations with empirical force fields to investigate oxygen transport in polycrystalline La1-xSrxFeO3-x/2 (x = 0.1 and 0.4). We used the Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) code. LAMMPS is based on integrating Newton’s equations of motion for interacting particles. Interactions were based on atomistic models, avoiding the high computational cost of ab initio methods. A rigid-ion model was used to describe short-range interactions via Buckingham-type potentials and long-range interactions via coulombic forces. Monocrystalline cells were created by a 15×10×15 expansion of the orthorhombic Pnma unit cell and thus contained ca. 45000 ions. For each composition (LSF10 and LSF40), two cells were created by randomly replacing 10% and 40% of lanthanum ions with strontium ions, respectively. Oxygen vacancies were introduced, i.e., oxide ions were removed, until the total cell charge was zero. Polycrystalline cells containing 2, 5, 10 or 20 grains were constructed with the ATOMSK code. In each case, the grains were created in a cubic simulation cell with cell dimensions of 80 Å × 80 Å × 80 Å, and thus with ca. 40000 ions. For ion pairs at the boundaries separated by less than 1.5 Å, one ion was removed at random to avoid instabilities during the minimisation. Since this leads to A:B≠1, cations at the grain boundary were removed until the numbers of A- and B-site ions were identical. Subsequently, strontium-containing cells were obtained by replacing at random the appropriate fraction of lanthanum ions with strontium ions. Oxide ions were removed at random to achieve a charge-neutral simulation cell. To capture some of the variability of grain-boundary structures, we used three different random seeds (= differently oriented grain centres, grain volumes, and Sr′La/vO configurations) to create cells containing 2, 5, 10 or 20 grains, resulting in a total of 12 cells for each compound (LSF10 and LSF40). Simulations were performed at temperatures 1000 ≤ T / K ≤ 3000. Oxygen tracer diffusion coefficients obtained for both orthorhombic La0.9Sr0.1FeO2.95 and cubic La0.6Sr0.4FeO2.8 are compared with experimental data, and are found to give a satisfactory description of the oxygen transport behaviour as a function of T and x. Subsequently, we examined oxygen tracer diffusion in polycrystalline cells of La1-xSrxFeO3-x/2 with differing numbers of grain boundaries. In all cells with random distributions of Sr, we find that oxygen tracer diffusion in the polycrystalline systems is lower than in the single-crystal systems. Our results thus indicate that in perovskite-oxide ceramics with homogeneous cation distributions there is no faster diffusion of oxygen along grain boundaries. However, by mimicking low temperature behaviour, where samples are oxidised (i.e., lower oxygen-vacancy concentrations), we find a reversed trend of higher tracer diffusion coefficients in polycrystalline systems than in the respective single-crystal systems.

Results

The main result of this computational study is the observation of faster grain-boundary diffusion of oxygen in polycrystalline LSF materials only when bulk diffusion of oxygen is relatively slow. This agrees very well with experimental studies of oxygen diffusion in polycrystalline LSCF6482, in which faster grain-boundary diffusion was observed at low temperatures, where bulk diffusion is substantially diminished on account of sample oxidation.

Additional Project Information

DFG classification: 302-03 Chemical Solid State and Surface Research, Theory and Modelling
Cluster: CLAIX

Publications

A. Bonkowski, John A. Kilner, Roger A. De Souza, Molecular-dynamics simulations of oxygen diffusion in
polycrystalline (La,Sr)FeO3−δ perovskite, in preparation.