AI-Accelerated Design of Electrolytes for Robust Inorganic-Rich SEI in Li/Na/Mg Batteries

Descripción del proyecto

Modelo de inteligencia artificial de disolventes para baterías metálicas

La tecnología de las baterías es clave para la descarbonización, pero el desarrollo de baterías de litio-metal de alta energía se ha retrasado por problemas de seguridad relacionados con la interfase electrolito sólido (SEI). La SEI rica en componentes inorgánicos ofrece mayor seguridad y estabilidad a largo plazo, y los electrolitos débilmente solvatantes (WSE) muestran un gran potencial en este ámbito. Con el apoyo de las acciones Marie Skłodowska-Curie, el proyecto conjunto AI4Bat está desarrollando un modelo de inteligencia artificial capaz de predecir los disolventes óptimos para los WSE en baterías metálicas de litio, sodio y magnesio. El proyecto utilizará simulaciones a nivel atómico para establecer vínculos entre las propiedades de los disolventes y los procesos de reducción de aniones en las baterías de litio. A continuación, los disolventes prometedores se formularán en WSE y se probarán en células de bolsa para evaluar su rendimiento.

Objetivo

Battery technology is critical to achieving global decarbonization, but the commercialization of high-energy Li-metal batteries has been stalled for decades due to safety risks and limited lifespan. These issues stem from the failure of the solid electrolyte interphase (SEI) on the Li-metal anode. Conventional carbonate-based electrolytes lead to an organic-rich SEI that is porous, ionically insulating, prone to dendrites formation, dissolution, and electron tunneling, resulting in continuous battery degradation. In contrast, an inorganic-rich SEI, formed by anion decomposition, is dense, insoluble, electron-insulating, mechanically robust, and dendrite-free—offering a path toward safer and more durable batteries. Weakly solvating electrolytes (WSE) have emerged as a promising solution for creating such SEIs, but their development remains slow, relying on inefficient trial-and-error methods. The AI4Bat project aims to revolutionize this process by developing an AI-model capable of predicting optimal solvents to form WSEs for Li, Na, and Mg-metal batteries. Starting with Li-metal batteries, the AI-model will link solvent properties to anion reduction in the electric double layer using atomic simulations. This model will then screen new solvents from extensive databases, which will be formulated into WSEs and tested in pouch cells. The performance of these WSEs will be rigorously evaluated, focusing on the inorganic, ion-conductive, electron-insulating, mechanical properties and insolubility of the SEI, and correlated with overall battery performance. Upon successfully validating the AI-model in Li batteries, I will transfer it to Na/Mg-metal batteries to enhance their electrochemical performance through the development of robust inorganic-rich SEIs. This project will significantly accelerate the design of safer, high-performance batteries, paving the way for next-generation energy storage and conversion technologies.