In the quest for sustainable, high-performance energy storage, the development of advanced batteries is crucial. Addressing the complexities of battery materials and processes requires novel, data-driven research approaches. The Amsterdam Modeling Suite (AMS) offers a robust framework for simulating battery materials at multiple levels of theory, integrating atomistic engines (DFT, DFTB, ReaxFF, ML potentials) with a central driver for exploring the potential energy surfaces (PES) via molecular dynamics and Grand Canonical Monte Carlo simulations. The combination of the AMS driver with its dedicated Python interface enables the automatic screening of materials to optimize critical battery properties such as intercalation potentials, diffusion constants, activation energies, redox potentials, mechanical properties, etc.
A key innovation of AMS2024 is its platform for machine learning interatomic potentials, enhancing simulation accuracy while conserving low computational resources. With ParAMS, you can train a single or a committee of machine learning potentials, from scratch, fine-tune universal models, or actively learn the PES based on target molecular dynamics simulation. This facilitates rapid prediction of material properties and reaction mechanisms, accelerating the design and testing of new battery technologies.
Our data-driven approach empowers the battery R&D community to overcome traditional development barriers, enabling precise material performance predictions and the discovery of new materials and mechanisms towards realizing high-energy, stable, and safe energy storage systems.