Water Auto-ionization Pathways at Graphene Interfaces via Enhanced Sampling and Machine Learning

The auto-ionization of water into hydronium and hydroxide ions is a rare but fundamental process that governs pH and proton transport in aqueous environments. While the mechanism in bulk water has been widely investigated, much less is known about how confinement and interfaces reshape the free energy landscape. Using machine-learned interatomic potentials and enhanced sampling techniques, we investigate water auto-ionization between graphene sheets across slit-pore sizes from nanometer to angstrom scale. We find that multilayer-confined water closely preserves bulk-like free energy profiles, whereas monolayer-confined water exhibits significantly elevated barriers due to geometric constraints that hinder ion-pair separation. This contrast arises despite both systems sharing the same stepwise mechanism as the bulk, featuring a water dimer intermediate state and two transition states associated with the formation of 3- and 4-members water wires. Notably, in the multilayer system, both ionization and subsequent proton transfer are localized within the high-density interfacial layers. These findings establish how confinement modulates proton-transfer pathways and tunes the free-energy cost of auto-ionization, offering fundamental insights into confined aqueous chemistry.