Abstract:Artificial intelligence (AI) has emerged as a tool for discovering and optimizing novel battery materials. However, the adoption of AI in battery cathode representation and discovery is still limited due to the complexity of optimizing multiple performance properties and the scarcity of high-fidelity data. In this study, we present a machine-learning model (DRXNet) for battery informatics and demonstrate the application in the discovery and optimization of disordered rocksalt (DRX) cathode materials. We have compiled the electrochemistry data of DRX cathodes over the past five years, resulting in a dataset of more than 19,000 discharge voltage profiles on diverse chemistries spanning 14 different metal species. Learning from this extensive dataset, our DRXNet model can automatically capture critical features in the cycling curves of DRX cathodes under various conditions. Illustratively, the model gives rational predictions of the discharge capacity for diverse compositions in the Li--Mn--O--F chemical space as well as for high-entropy systems. As a universal model trained on diverse chemistries, our approach offers a data-driven solution to facilitate the rapid identification of novel cathode materials, accelerating the development of next-generation batteries for carbon neutralization.

What problem does this paper attempt to address?

### Problems the Paper Attempts to Solve This paper aims to address the challenges in representing and discovering battery cathode materials (specifically disordered rock salt (DRX) cathode materials) in multi-dimensional chemical space. Specifically, the paper focuses on the following aspects: 1. **Complexity of Multi-Property Optimization**: Current machine learning methods in battery research mainly focus on predicting battery life within a single chemical system or a limited chemical space. However, optimizing multiple electrochemical properties (such as rate capability, cycling stability, and different testing voltage windows) simultaneously remains challenging. 2. **Scarcity of High-Fidelity Data**: The lack of high-quality experimental data further hinders the application of artificial intelligence in the battery field. 3. **Rapid Identification and Optimization of Novel Cathode Materials**: By developing a general machine learning model (DRXNet), which can be trained on various chemical systems, the paper aims to accelerate the discovery and optimization of novel cathode materials, promoting the development of next-generation batteries to achieve carbon neutrality goals. ### Specific Problems - **How to automatically capture key features in the cycling curves of DRX cathode materials under different conditions using the machine learning model (DRXNet)?** - **How to train the model using large-scale experimental data (over 19,000 discharge voltage curves) to accurately predict the discharge capacity of DRX cathode materials with different compositions?** - **How to quickly identify novel cathode materials with improved energy storage capabilities through a data-driven approach?** ### Solutions - **Constructing the DRXNet Model**: This model uses composition, test current density, operating voltage window, and cycle number as inputs to predict the entire discharge voltage curve. - **Compiling a Large-Scale Experimental Dataset**: Collecting electrochemical data on DRX cathode materials from the past 5 years, forming a dataset containing over 19,000 discharge voltage curves. - **Validating Model Performance**: Verifying the accuracy and generalization ability of the DRXNet model through prediction results under different chemical compositions and test conditions, especially in high-entropy systems. Through these methods, the paper demonstrates the potential of the DRXNet model in exploring and optimizing novel cathode materials in multi-dimensional chemical space, providing a data-driven solution for the rapid discovery and optimization of battery materials.

Deep learning of experimental electrochemistry for battery cathodes across diverse compositions

Chemometric and Machine Learning Analysis of Lithium Concentration and Solvation Behavior in Li-Ion Battery Electrolytes

Navigating Materials Chemical Space to Discover New Battery Electrodes Using Machine Learning

Application of deep learning for informatics aided design of electrode materials in metal-ion batteries

Transfer Learning Enabled Deep Learning Model for the Prediction of Battery Performance from Electrolyte Formulations

Ab Initio Design of Ni‐Rich Cathode Material with Assistance of Machine Learning for High Energy Lithium‐Ion Batteries

Artificial intelligence for the understanding of electrolyte chemistry and electrode interface in lithium battery

Artificial Intelligence Driven Design of Cathode Materials for Sodium-Ion Batteries Using Graph Deep Learning Method

Unlocking New Insights for Electrocatalyst Design: A Unique Data Science Workflow Leveraging Internet-Sourced Big Data

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

Generating Comprehensive Lithium Battery Charging Data with Generative AI

Autonomous discovery of battery electrolytes with robotic experimentation and machine-learning

Artificial Intelligence-Driven Real-World Battery Diagnostics

Machine Learning Enabled Prediction of Cathode Materials for Zn ion Batteries

Solutions for Lithium Battery Materials Data Issues in Machine Learning: Overview and Future Outlook

AI-Driven Accelerated Discovery of Intercalation-type Cathode Materials for Magnesium Batteries

A Pre-trained Deep Potential Model for Sulfide Solid Electrolytes with Broad Coverage and High Accuracy

Machine Learning-Accelerated Discovery and Design of Electrode Materials and Electrolytes for Lithium Ion Batteries

Improving Electrolyte Performance for Target Cathode Loading Using Interpretable Data-Driven Approach

Machine Learning: An Advanced Platform for Materials Development and State Prediction in Lithium‐Ion Batteries

Accelerated Design of Electrodes for Liquid Metal Battery by Machine Learning