Ternary cathode materials are pivotal in high-performance battery technologies, with grain size influencing their electrochemical performance. However, the absence of real-time grain size inspection during material preparation poses challenges in maintaining the consistent quality of ternary cathode materials. To address this, this paper proposes a novel method employing cell automata to model the topological evolution of grain growth in these materials, integrated with deep neural networks (DNN). The grain growth process is divided into two stages: heating and constant temperature. In the heating stage, varying heating rates and plateau temperatures serve as DNN inputs, yielding the primary grain distribution for the cell automata model in the constant temperature stage. Based on cell automata, the grain growth model links the grain growth rate to the grain size distribution in the constant temperature stage. A surface energy constraint rule, based on local curvature and grain boundary surface tension, governs growth rates. The grain boundary growth ratio is also used to create a grain ID transition variable, dictating grain ID conversion in the model. This approach accurately simulates the dynamic evolution of polycrystalline grain size and morphology. Simulation results show that this method effectively models grain growth in ternary cathode materials, offering insights for optimising the sintering process and improving material quality.
Read full abstract