(159e) Accelerating the Discovery of Organic Electrode Materials for Li-Ion Batteries: An Integrated Approach Using DFT, Machine Learning, and High-Throughput Virtual Screening

Density functional theory (DFT) and machine learning (ML) are employed to investigate the structure-electrochemical performance relationships of organic electrode materials for Li-ion batteries. DFT is employed to compute the redox potential of various organic moieties, providing valuable insights into their electrochemical properties. However, due to the significant computational time required for high-efficacy DFT modeling, it is not ideal for vast material screening. To overcome this limitation, we implement ML techniques, including artificial neural networks (ANN), kernel ridge regression (KRR), and gradient-boosting regression (GBR), for accelerated discovery and assessment of structure-electrochemical relationships. To further enhance the material screening process, we develop a high-throughput virtual screening (HTVS) pipeline consisting of several surrogate learning models with increasing fidelity. The pipeline effectively prunes material candidates at each stage, ensuring only a subset with higher probability for desirable redox potential progresses to more computationally expensive models. By combining DFT, ML, and HTVS, we create an advanced scheme for accurate prediction and analysis of electrochemical activity, facilitating the discovery of novel organic electrode materials with enhanced electrochemical properties. This integrated approach enables more efficient exploration of the materials space and helps to improve the overall performance of the learning model, paving the way for the development of better energy storage devices.