Determining and Understanding Long-Range Proton Conduction Pathways and Patterns in Perovskite Systems



Journal Title

Journal ISSN

Volume Title



Fuel cells with solid oxide electrolytes (SOFCs) have attracted increasing interest as efficient alternatives to combustion-based energy production. Doped perovskites have displayed mechanical stability and high proton conductivity in the key 500-700 ̊C temperature range, making investigation of proton conduction in perovskites an important step in the fuel cell development process. This work details the characterization of a perovskite grain boundary expected to display unique conduction behavior, the fine-tuning and testing of a new multistep Kinetic Monte Carlo (KMC) algorithm for generating long-range conduction pathways, and the application of a graph theoretical centrality measure to help explain system-specific conduction patterns. Although proton conduction in bulk perovskites is well understood, many questions remain about proton conduction in grain boundaries, which represent major barriers to conduction. Theoretical study a grain boundary system begins with full energetic characterization of conduction in the system. As a first step in this process, optimum energies for Y/BaZrO3 (310) tilt grain boundaries doped at different sites have been calculated using density functional theory. Results from different exchange correlation functionals have been compared. Because proton conduction in perovskite systems involves a series of rare proton movements, it can be modeled with KMC simulations. The standard KMC algorithm involves moving a proton between binding sites, using random numbers and rate constants to pick moves in proportion with their probability. We have modified this algorithm to generate n-step pathways from each site. At each iteration, a pathway rather than a single move is selected. Results for the Y/BaZrO3, Al/BaZrO3, Y/SrZrO3, and Al/SrZrO3 systems have been in good agreement with standard KMC results, showing that probable pathways run through different regions in systems with different dopants. A valid method of calculating rate constants for multi-step moves has been developed, and the multi- step algorithm has been shown to increase computational efficiency. Graph theoretical methods have been applied to provide additional insight into the relationships between binding sites. Average hitting time between sites has been used to rank binding sites based on accessibility. The contrasting patterns of rankings in each system are consistent with contrasting trends in conduction pathways and barriers, suggesting that graph theoretical ranking may prove a quick, powerful tool for predicting conduction trends in perovskite systems. Thus, this work refines an existing technique for generating proton conduction pathways, applies a graph theoretical centrality measure to proton conductors for the first time, and begins the characterization of a grain boundary system, the first step toward full analysis of conduction trends.



kinetic monte carlo, perovskite, solid oxide fuel cell, centrality measures, proton conduction, density functional theory, grain boundary, markov centrality