A Reactive Molecular Model for Aluminosilicate Chemistry to Study Zeolite Formation

01 October 2022 → 30 September 2025
Research Foundation - Flanders (FWO)
Research disciplines
  • Natural sciences
    • Inorganic chemistry not elsewhere classified
    • Chemistry of clusters, colloids and nanomaterials
    • Solution chemistry
    • Statistical mechanics in chemistry
    • Theoretical and computational chemistry not elsewhere classified
Zeolite Formation Hydrated Silicate Ionic Liquids (HSILs) Atomic Neural Network Potentials
Project description

Despite their large commercial importance, zeolite formation is poorly understood due to the complex, heterogeneous nature of traditional synthesis. COK-KAT (KU Leuven) recently reported a novel synthesis path via hydrated silicate ionic liquids (HSILs), completely homogeneous, inorganic liquids which yield zeolites at moderate conditions. HSILs are severely subhydrated, room temperature alkali-silicate melts consisting of small oligomers. Water is not present as bulk, but as a ligand to the ionic species. HSILs are very stable, until addition of aluminate triggers nucleation and zeolite growth even at room temperature (~6 months) or within minutes at 180°C. The unique properties of HSILs allow for development of a reactive molecular model for aluminosilicate chemistry at the Center for Molecular Modeling (CMM , Ghent University), which can be carefully tested against detailed experimental results obtained at COK-KAT. As zeolite formation involves successive condensation reactions, reactive neural network potentials will be trained on high-level DFT-D data to be used in large-scale molecular dynamics simulations. To minimize the amount of expensive DFT-D calculations, an active learning scheme will be employed. Enhanced sampling methods will be used to efficiently explore the free energy surface. This, in combination with detailed experimental insight, will lead to a better understanding of the relationship between HSIL composition and the experimentally observed topology.