Machine Learning Aided Investigation of Reactive Chemistry at Surfaces

点击翻译按钮获取中文摘要

The project takes on tackling the challenges of creating accurate machine learning potentials, in particular creating a robust and transferable automated protocol that explores desired molecule-surface catalytic interaction, with equal accuracy in describing the reactants and the catalysts. This will involve the development of an efficient iterative training protocol. This will involve determining a suitable set of simulations to do and the amount of data from each to include in the next iteration to explore relevant surface processes. Throughout the project we will be gradually expanding the complexity of the reactions and surfaces where the potential can be applied.The work will expand on multiple fronts. Firstly, with the creation of a protocol for optimal data generation, specifically for understanding surface process. Secondly, testing and developing on the realm of prebiotic organic reaction could lead to important insights into the origins of life and the role that surface interactions and catalysis played in the process. The topic of prebiotic chemical reactions are of high interest. The research will begin with investigating amino acid synthesis and the bonding between glycine on alumina and silica. These two oxides are relevant in the final goal of reproducing these reactions on clays. This is of interest due to the potential role of clays in relation to the creation of the first biomolecules. Finally, we hope to explore the addition of dopants and the prospects of surface bound replicating of RNA.