Abstract

Zeolites are an important and well-studied class of porous inorganic solids, with widespread and diverse industrial applications. Their effective utilization and optimization requires an understanding of dynamic solid-liquid processes occurring both at the surface and inside the pore system. In this presentation I will discuss recent developments within our group, in which we aim to tackle the complexity and timescale problems in modelling zeolites, moving from first principles molecular dynamics to the development of reactive, tuneable neural network potentials with (up to) hybrid DFT quality. I will illustrate the importance of i) accelerating zeolite simulations and ii) unbiased exploration of configuration space, with examples of cation (micro-)solvation, spectroscopic characterization, chemistry at the water-silicate interface and global optimization of heteroatom distributions for screening of potential new zeolites.