Speaker
Description
The vibrational dynamics of molecules in non-contact junctions depend critically on the geometric structure and electronic interactions between molecule and substrate. Vibrations excited by external stimuli dissipate energy into substrate electrons and phonons, affecting spectral linewidths, vibrational lifetimes, and the coupling between molecular and substrate phonons. We present a first-principles approach to disentangle the dissipation pathways by combining density functional theory, machine learning interatomic potentials (MLIPs), and non-adiabatic molecular dynamics. Using CO-functionalised Cu surfaces—prototypical systems for scanning probe and energy dissipation experiments—we train an MLIP that accurately captures molecule–substrate interactions. Electron-phonon coupling is incorporated via molecular dynamics with electronic friction. We reveal strong vibrational mode specificity sensitive to the tip–substrate geometry. Using equilibrium correlation function analysis, we extract phonon spectral functions and identify a weak non-additive effect where electron-phonon coupling enhances phonon-phonon relaxation. Our predicted vibrational lifetimes show good agreement with infrared and helium scattering experiments.