Agent-Based Modeling: Partitioned Density Analysis of Unidirectional Atomistic Diffusion in Li-Ion Batteries

Abstract Atomic-scale simulations, such as molecular dynamics (MD) and Monte Carlo (MC), are powerful tools for uncovering microscopic diffusion mechanisms in lithium-ion batteries (LIBs). However, these methods are computationally demanding, requiring substantial time and resources, especially for large systems that generate extensive trajectory data. These challenges highlight the need for alternative, more efficient modeling approaches. In this study, we identify unidirectional diffusion in LIBs as analogous to disordered traffic systems. Building on this analogy, we propose the use of agent-based modeling (ABM)—a well-established framework for simulating heterogeneous and disordered traffic—to model atomistic diffusion processes. We assess the feasibility of this approach and outline a systematic framework for the effective design of such models. Our results show that heterogeneous systems exhibit distinct density distributions across different atomic species. Lighter atoms experience larger displacements from their initial positions compared to heavier atoms, leading to higher diffusivity, ionic conductivity, and diffusion lengths. Analysis of the density spectrum indicates that diffusion in carbon (C) is more pronounced than in copper (Cu), a finding further supported by their respective velocity profiles. The proposed analogy between single-lane, heterogeneous, disordered traffic, and unidirectional diffusion in LIBs offers a novel conceptual framework for developing ABM-based diffusion models. These findings lay the groundwork for future research, emphasizing the need for further refinement and interdisciplinary collaboration to establish robust rules and parameters for ABM in this context.