Physics and Uncertainty-Aware Ensemble-Based ML-Models for Interatomic Potentials and Prediction of Reaction Networks and Cell-Level Performance

Understanding the dynamic processes at solid-liquid interfaces in electrochemical devices like batteries is key to developing more efficient and durable technologies for the green transition. Fundamental and performance-limiting interfacial processes like the formation of the Solid-Electrolyte Interphase (SEI) [1] and dendritic growth [2] span numerous time- and length scales. Despite decades of research, the fundamental understanding of structure-property relations remains elusive. Ab initio molecular dynamics (AIMD) generally provides sufficient accuracy to describe chemical reactions and the making and breaking of chemical bonds at these interfaces [3]. Still, the cost is prohibitively high to reach sufficiently long time- and length scales to ensure proper statistical sampling [4]. Machine learning (ML) potentials offer a potential solution to this challenge. Still, training ML-based potentials capable of handling activated processes in organic or aqueous electrolytes remains a fundamental challenge since the potential must capture both intra- and intermolecular interactions in the electrolyte and during chemical reactions at the interface [4]. Here, we present new approaches using phase field models [2], graph neural networks [5] and new transition state training sets [6] for chemical reaction networks, and machine/deep learning models to predict the spatio-temporal evolution of electrochemical interphases [3]. We also discuss the development of methods like symbolic regression to learn the laws of electrolyte transport [8], the predictions of energy and forces with calibrated uncertainty quantification for interatomic potentials using neural network ensemble models [8,9]. Finally, we discuss how such models trained on multi-sourced and multi-fidelity data from multiscale computer simulations, operando characterization, high-throughput synthesis, and testing, to provide uncertainty-aware and explainable ML for early prediction of, e.g., of degradation trajectories for battery cells [10].
1. Diddens, Bhowmik, et al., Adv. Mater. Inter., 2022, 2101734
2. Jeon, Ho, Vegge, Chang, ACS Appl. Mater. Inter., 2022, 14, 15275
3. Bhowmik, Castelli, Garcia-Lastra, Jørgensen, Winther, Vegge, Energy Storage Mater., 2019, 21, 446
4. Bhowmik, Vegge, et al., Adv. Energy Mater, 2021, 2102698
5. Schreiner, Bhowmik, Vegge, Winther, Mach. Learn. Sci. Tech., 2022, 3, 045022
6. Schreiner, Bhowmik, Vegge, Winther, Sci. Data, 2022, 9, 779
7. Flores, Wölke, Yan, Winter, Vegge, Cekic-Laskovic, Bhowmik, Digital Discovery, 2022, 1, 440
8. Busk, Jørgensen, Bhowmik, Schmidt, Winther, Vegge, Mach. Learn.: Sci. Technol, 2022, 3, 015012
9. Busk, Schmidt, Winther, Vegge, Jørgensen, Phys. Chem. Chem. Phys., 2023, 5, 25828
10. Rieger, Flores, Nielsen, Winther, Ayerbe, Vegge, Bhowmik, Digital Discovery, 2023, 2, 112