Microscopic insights into amorphous battery anodes

As the world transitions from fossil fuel power to clean electric power, the development of next- generation batteries is becoming vital to facilitate energy storage. Lithium-ion batteries (LIBs) have long been the benchmark technology for high-power, intermediate-scale energy storage, owing to their rapid charging capability, high cyclic stability, and safety. Nonetheless, the elec- trification of sectors such as long-haul trucking and aviation remains challenging, demanding batteries with much higher energy densities.
Over the decades, researchers have identified various approaches to address the current limita- tions of battery technology. This includes searching for enhanced electrode materials for LIBs, and harnessing other more abundant alkali metals, such as sodium and potassium, as alterna- tives to lithium. Regardless of the chosen direction, both strategies necessitate the development of novel electrode materials, since in the latter case, larger ion sizes of sodium and potassium de- mand different solid-state networks to properly function.
Aligning with the EPSRC's focus on advancing material science, this project seeks to identify and evaluate battery anode candidates compatible with sodium and potassium. The goal is to develop anodes that not only meet but surpass the performance of current LIB anodes, thereby enhancing the energy capacity for cutting-edge applications.
To achieve this, experimental battery materials research is often complemented by computa- tional simulations of the atomistic structures and reactivity, to understand the fundamental principles behind materials' properties. Quantum mechanical-based Density Functional The- ory (DFT) methods are known for their high accuracy in small-scale simulations; however, their application is constrained to small and relatively simple model systems due to their large computational costs. On the other hand, empirical potentials, which parameterise interatomic po- tentials based on physical models, bypass the detailed treatment of electrons. Despite being many orders of magnitude faster, these hand-crafted models often fall short of accuracy and transferability.
Recently, machine learning (ML)-fitted interatomic potentials derived from quantum mechan- ical reference data have emerged as a revolutionary approach. This approach combines the ben- efits of both traditional methods, offering a way to conduct atomistic simulations of complex disordered (amorphous) materials with unprecedented speed and accuracy, for simulation sizes and timescales that were previously beyond reach. Leveraging this breakthrough, this project employs ML-based interatomic potentials to understand the atomic structure and investigate the electrochemical performance of amorphous anodes for next-generation batteries.