Electrolytomics: A unified big data approach for electrolyte design and discovery

Chibueze Amanchukwu,Ritesh Kumar,Minh Canh Vu,Peiyuan Ma

DOI: https://doi.org/10.26434/chemrxiv-2024-vqtc7

2024-04-24

Abstract:Electrolyte discovery is the bottleneck for developing next generation batteries. For example, lithium metal batteries (LMBs) promise to double the energy density of current Li-ion batteries (LIBs) while next generation LIBs are desired for operations at extreme temperature conditions and with high voltage cathodes. However, there are no suitable electrolytes to support these battery chemistries. Electrolyte requirements are complex (conductivity, stability, safety) and the chemical design space (salts, solvents, additives, composition) is practically infinite; hence discovery is primarily guided through trial-and-error which slows the deployment of new battery chemistries. Inspired by artificial intelligence (AI)-enabled drug discovery, we usher in a new paradigm for electrolyte discovery. We assemble the largest small molecule experimental liquid electrolyte ionic conductivity dataset and build highly accurate machine learning (ML) and deep learning models to predict ionic conductivity across a wide range of electrolyte classes. The developed models outperform molecular dynamic (MD) simulations and are interpretable without explicit encoding of ionic solvation. While most ML-based approaches target a single property, we build additional models of oxidative stability and Coulombic efficiency and develop a new metric called the electrolyte score (eScore) to unify the predicted disparate electrolyte properties. Deploying these models on large unlabeled datasets, we discover new electrolyte solvents, experimentally validate that the electrolyte is conductive (> 1 mS cm-1), stable up to 6V, supports efficient anode-free LMB, and even LIB cycling at extreme temperatures. Our work heralds a new age in electrolyte design and battery materials discovery.

Chemistry

What problem does this paper attempt to address?

This paper focuses on the design and discovery of battery electrolytes, which is the bottleneck for the development of next-generation battery technologies. The current electrolytes cannot support next-generation lithium metal batteries (LMBs) with high energy density and lithium-ion batteries (LIBs) operated at extreme temperatures. Due to the complex requirements of electrolytes, such as conductivity, stability, and safety, the design space is almost infinite, thus traditional trial-and-error methods hinder the application of new battery chemistry.

The paper proposes an AI-driven new paradigm to accelerate the discovery of electrolytes. The researchers have established the largest experimental liquid electrolyte ion conductivity database to date and constructed highly accurate machine learning (ML) and deep learning models to predict the ion conductivity of different electrolyte categories. These models outperform molecular dynamics simulations and do not require explicit encoding of ion-solvent structures. In addition, they have developed additional models to predict oxidative stability and coulombic efficiency, and introduced a new metric called "electrolyte score" (eScore) to unify the predictions of multiple electrolyte properties.

By applying these models on a large unlabeled dataset, the researchers have discovered new electrolyte solvents and demonstrated that these electrolytes have high conductivity (>1 mS cm-1), oxidative stability up to 6V, can effectively support anode-free LMBs, and can cycle LIBs at extreme temperatures. This work marks a new era in electrolyte design and battery material discovery, and is of great significance for advancing the development of new energy batteries.