Perspective: Driving Electric Vehicle Battery Design With Atomistic Modeling

Abstract Electric vehicles (EVs) continue to increase their share of the automotive market. To spur this growth, original equipment manufacturers (OEMs) and battery cell manufacturers have invested in atomistic modeling approaches based on fundamental science. This effort has been successful in improving vehicle performance through modifications to battery cell chemistry. However, it is our view that atomistic modeling can go a step further to affect vehicle battery design. In this perspective, we demonstrate a multiscale modeling approach to link atomic-scale phenomena with full cell predictions relevant for battery design engineers. Recent multiscale modeling approaches undertaken at General Motors are discussed. We show that variation in the predicted diffusivity of lithium ions in the electrolyte leads to variation in final cell temperatures of 3 °C in small-format cells, and that reversible volume change for common cathode materials can be as large as 10% for the full state of lithiation window. Additionally, machine learning (ML) will be presented as another growing area in the literature to drive cell-level battery design. Linking ML approaches with datasets from atomistic modeling represents a key direction of growth for vehicle design.