Facilitating ab initio configurational sampling of multicomponent solids using an on-lattice neural network model and active learning.

We propose a scheme for ab initio configurational sampling in multicomponent crystalline solids using Behler-Parinello type neural network potentials (NNPs) in an unconventional way: the NNPs are trained to predict the energies of relaxed structures from the perfect lattice with configurational disorder instead of the usual way of training to predict energies as functions of continuous atom coordinates. An active learning scheme is employed to obtain a training set containing configurations of thermodynamic relevance. This enables bypassing of the structural relaxation procedure that is necessary when applying conventional NNP approaches to the lattice configuration problem. The idea is demonstrated on the calculation of the temperature dependence of the degree of A/B site inversion in three spinel oxides, MgAl2O4, ZnAl2O4, and MgGa2O4. The present scheme may serve as an alternative to cluster expansion for "difficult" systems, e.g., complex bulk or interface systems with many components and sublattices that are relevant to many technological applications today.

Oxide-ion diffusion in brownmillerite-type Ca2AlMnO5+δ from first-principles calculations.

Oxide-ion diffusion pathways in brownmillerite oxides Ca2AlMnO5 and Ca2AlMnO5.5 are systematically investigated using first-principles calculations. These structures reversibly transform into each other by oxidation and reduction. We examine oxide-ion migration in Ca2AlMnO5 and Ca2AlMnO5.5 using the nudged elastic band method. In the reduced structure (Ca2AlMnO5), oxide-ion migration through a vacancy channel is found to have the lowest migration energy barrier, at 0.58 eV. The migration energy barrier of the second-lowest energy path, perpendicular to the vacancy channel, is found to be 0.98 eV. In the oxidized structure (Ca2AlMnO5.5), oxide-ion migration within AlO6 layers has migration energy barriers of 0.55 eV and 0.56 eV in the [100] and [001] directions, respectively. Oxide-ion migration perpendicular to the AlO6 layer has a migration energy barrier of 1.33 eV, suggesting that oxide-ion diffusion in the [010] direction is difficult even at elevated temperature. These results indicate that diffusion in the reduced phase is predominantly one-dimensional whereas it is two-dimensional in the oxidized phase.