Advancing Density Functional Theory for Chemically Accurate Reactive and Non-Reactive Condensed Phase Simulations with Machine Learning
ORAL
Abstract
Accurately representing the electronic structure and dynamics of condensed phases is essential for understanding their properties and reactivity. Density Functional Theory (DFT), one of the most widely employed quantum chemistry methods for modeling condensed-phase systems, is often limited by its computational cost, which restricts the use of more accurate modern functionals. To overcome these limitations, we introduce data-driven many-body potential energy functions trained with arbitrary DFT functionals (MB-DFT). These functions retain the accuracy of the parent DFT functional across both gas and condensed phases. However, despite recent advances, no existing density functional achieves the required accuracy to predict the properties of hydrated systems across different phases. To address this challenge, we present the Density-Corrected SCAN (DC-SCAN) method, which elevates the SCAN functional's accuracy to that of coupled-cluster theory—widely regarded as the gold standard for chemical accuracy. This breakthrough enables the accurate modeling of aqueous systems from gas to condensed phases within our data-driven many-body DFT framework, providing quantitative agreement with the experiments. Nevertheless, like most polarizable force fields, the data-driven MB-DFT potentials are not designed to handle chemical reactions. To accurately model chemical reactions in condensed phases, we have integrated the DC-SCAN accuracy with the flexibility of neural network potentials. This approach has led to the development of reactive machine-learned potentials capable of studying complex phenomena, such as water autoionization and proton transport in confined environments, using advanced enhanced-sampling techniques. Our results not only provide an accurate estimation of the autoionization constant but also highlight the crucial role of the Grotthuss mechanism in acid/base equilibria in bulk and nano-confined water
–
Publication: [1] Lambros, E.; Dasgupta, S.; Palos, E.; Swee, S.; Hu, J.; Paesani, F. General Many-Body Framework for Data-Driven Potentials With Arbitrary Quantum Mechanical Accuracy: Water as a Case Study. J. Chem. Theory. Comput. 2021, 17, 5635–5650.<br>[2] Dasgupta, S.; Lambros, E.; Perdew, J. P.; Paesani, F. Elevating Density Functional Theory to Chemical Accuracy for Water Simulations Through a Density-Corrected Many-Body Formalism. Nat. Commun. 2021, 12, 6359.<br>[3] Dasgupta, S.; Shahi, C.; Bhetwal, P.; Perdew, J. P.; Paesani, F. How Good Is the Density-Corrected Scan Functional for Neutral and Ionic Aqueous Systems and What Is So Right About the Hartree–Fock Density? J. Chem. Theory. Comput. 2022, 18, 4745–4761.<br>[4] Dasgupta, S.; Cassone, G.; Paesani, F. Nuclear Quantum Effects and the Grotthuss Mechanism Dictate the pH of Liquid Water. ChemRxiv 2024
