Abstract
Water’s ability to autoionize into hydronium (H3O+) and hydroxide (OH−) ions dictates the acidity or basicity of aqueous solutions, influencing the reaction pathways of many chemical and biochemical processes. In this study, we determine the molecular mechanisms of the autoionization process by leveraging both the computational efficiency of a deep neural network potential trained on highly accurate data calculated within density-corrected density functional theory and the ability of enhanced sampling techniques to ensure a comprehensive exploration of the underlying multidimensional free- energy landscape. By properly accounting for nuclear quantum effects, our simulations provide an accurate estimate of autoionization constant of liquid water (Kw = 1.23 × 10−14 ), offering the first realistic molecular-level picture of the autoionization process and emphasizing its quantum-mechanical nature. Importantly, our simulations highlight the central role played by the Grotthuss mechanism in stabilizing solvent-separated ion pair configurations, revealing its profound impact on acid/base equilibria in aqueous environments.
Supplementary materials
Title
Supplementary Information
Description
Details about the development and validation of the DNN@DC- r2SCAN potentials, along with details about all MD simulations.
Actions