ÖPG-CMD Joint Meeting 2026

Elucidating the Complex Chemistry and Importance of Quantum Effects in High-Performing Electrolyte Solutions using SE(3)-Equivariant Transformer Network Potentials

Sep 23, 2026, 12:15 PM

30m

HS 12.11 (University of Graz)

4) Invited talk M21 - Recent Developments in Machine Learned Interatomic Potentials Mini-Colloquium

Speaker

Mark E. Tuckerman (1Department of Chemistry, New York University, New York, NY 10003 USA 2Courant Institute of Mathematical Sciences, New York University, New York, NY 10014 USA 3NYU-ECNU Center for Computational Chemistry at NYU Shanghai, Shanghai, China 200062 4Simons Center for Computational Physical Chemistry at New York University, New York, NY 10003 USA)

Description

Candidate systems for next-generation battery electrolyte materials, such as deep eutectic solvents and ionic liquids, often suffer from the limitation of an empirical inverse relation between viscosity and conductivity, known as Walden’s rule, which suppresses rates of charge transport and limits their electrochemical performance characteristics. An alternative to these ionic systems involves a class of systems known as concentrated hydrogen-bonded electrolytes (CoHBEs), which are structured, electrochemically stable and less volatile. CoHBEs can also be designed such that charge transport kinetics and solvent dynamics are largely decoupled in such a way as to break the viscosity-conductivity tradeoff implied by Walden’s rule. The basic strategy for achieving this breakthrough performance is to leverage the Grotthuss transport mechanism by choosing organic molecular solvent species, such as imidazole, capable of supporting proton hops through a dynamic, amphoteric hydrogen-bond network along with redox-active molecules capable of reversibly exchanging protons with the solvent species and undergoing proton-coupled electron transfer (PCET) reactions with each other. Accurate modeling of the charge transfer reactions and proton transport properties that give rise to high charge conductivities in these electrolytes proves computationally challenging because of the need to perform lengthy condensed phase simulations, treating both the electronic and nuclear degrees of freedom quantum mechanically. I will demonstrate that such a modeling task can be efficiently achieved with the use of DFT-trained SE(3)-equivariant transformer network potentials (MLP) to accelerate path integral molecular dynamics (PIMD) simulations. We highlight the practical utility of this approach by using it to benchmark how well PIMD simulations employing different DFT exchange-correlation functionals reproduce the composition-dependent densities, diffusion coefficients, and electrical conductivities of mixtures consisting of imidazole and either levulinic or acetic acid. Even with the speedup afforded by our MLPs, PIMD simulations remain quite expensive. In order to render PIMD more computationally tractable, we introduce and benchmark the accuracy of a ring polymer contraction approach that leverages a computationally efficient short-range MLP to accelerate our PIMD simulations by an additional factor of four.