Lecture
Van Marum Colloquium: How can machine learning facilitate computational electrochemistry
- Dr. Jia-Xin Zhu (Forschungszentrum Jülich, Germany)
- Date
- Thursday 19 February 2026
- Time
- Location
-
Gorlaeus Building
Einsteinweg 55
2333 CC Leiden - Room
- CM.3.23
Abstract
Electrochemistry is foundational to modern sustainable energy technologies, yet its computational modeling has long been hindered by the inherent trade-off between efficiency and accuracy. Recently, emerging machine learning (ML) techniques have enabled a more robust compromise. Notably, the integration of ML in this field is increasingly driven by physical intuition, incorporating essential elements from traditional non-ML computational methods. In this talk, I will first provide an overview of foundational methodologies to contextualize the unique physical requirements of electrical double layers. I will then trace the evolution of machine learning potentials, from early short-range local descriptors to advanced architectures capable of capturing long-range electrostatic interactions. A critical analysis will follow on a central challenge: accurately modeling the disparate dielectric responses of metallic conductors versus ionic insulators. I will discuss the emergence of hybrid frameworks as a promising solution to this complexity. Finally, I will offer an outlook on the future of the field, highlighting the necessity of synergistically integrating machine learning with multiscale modeling to address mesoscopic electrochemical systems.