Desorption dynamics of interstellar molecule on amorphous solid water investigated by machine learning potential-based PaCS-MD simulation

Abstract

Molecular adsorption, diffusion, reaction, and desorption on interstellar amorphous solid water (ASW) dominate the chemical processes in interstellar molecular clouds. Elucidating the chemical evolution in molecular clouds requires understanding the interactions between molecules and the ASW surface. Investigating the dynamics on ASW surfaces requires highly accurate and efficient molecular dynamics (MD) simulations. This study proposes a computational scheme that combines a rare-event sampling method, parallel cascade selection MD (PaCS-MD), with machine learning potential-based MD (MLP-MD), referred to as PaCS-MLP-MD. Because PaCS-MLP-MD explores rare-event dynamics without applying external forces or increasing the temperature, it provides a suitable framework for investigating astrochemically relevant molecular processes. Using PaCS-MLP-MD simulations, we simulated desorption of aminoacetonitrile (AAN) from an ASW surface model. Two potential desorption processes were identified: one in which the amino group desorbs first, followed by the nitrile group, and another in which the nitrile group desorbs first, followed by the amino group. Additional unbiased MLP-MD simulations, initiated from representative PaCS-MLP-MD snapshots, were performed to reveal desorption and re-adsorption behaviors of AAN on the ASW surface. These new findings from the PaCS-MLP-MD simulations may contribute to a more detailed understanding of processes occurring on ASW surfaces in molecular cloud environments.