Structure of Confined Water in SiO2 Nanopores from Machine Learning Interatomic Potential Molecular Dynamics Simulation

Understanding water-oxide interfaces at the atomistic level is important as they are responsible for many catalytic reactions in solutions. Here, we simulate the water-SiO2 interface using molecular dynamics with a machine-learning interatomic potential based on Chebyshev polynomials. We focus on understanding the effects of confinement and surface chemistry on the hydrogen-bond network and water structure and dynamics at the interface. Both stronger confinement and lower surface OH coverage makes water become more structured at the interface. We find that the surface with one OH per metal site facilitates the formation of ice in the confined region, increasing the melting temperature of water by 30-40 K. Our study demonstrates that water structure at the interface can be varied by controlling pore size and surface chemistry.