A molecular dynamics study of oxygen ion diffusion in A-site ordered perovskite PrBaCo2O5.5: data mining the oxygen trajectories

Abstract

Molecular dynamics (MD) simulations have been widely used to study oxygen ion diffusion in crystals. In the data analysis, one typically calculates the mean squared displacements to obtain the self-diffusion coefficients. Further information extraction for each individual atom poses significant challenges due to the lack of general methods. In this work, oxygen ion diffusion in A-site ordered perovskite PrBaCo₂O_5.5 is studied using MD simulations and the oxygen migration is analyzed by k-means clustering, a machine learning algorithm. The clustering analysis allows the tracking of each individual oxygen jump along with its corresponding location, i.e., oxygen site in BaO, PrO_0.5 and CoO₂ layers. Therefore it increases the understanding of the factors influencing oxygen diffusion. For example, it is found that the oxygen occupation fraction in the PrO_0.5 layers increases with temperature, while in the CoO₂ layers it decreases with temperature. Additionally, the activation enthalpies of oxygen jumps from CoO₂ to CoO₂, CoO₂ to PrO_0.5 and PrO_0.5 to CoO₂ are 0.22 eV, 0.54 eV and 0.34 eV, respectively, exhibiting anisotropic characteristics. Furthermore, the dwell times of oxygen atoms suggest that they are highly mobile in PrO_0.5 layers. Combining the analysis of activation enthalpies and dwell times, it is suggested that the oxygen transport is fast within the CoO₂ layers while the PrO_0.5 layers work as oxygen vacancy reservoirs.