Interfaces between different material phases or materials of different chemical composition often define or allow tuning the properties of a system of interest. An atomistic level of understanding such interfaces provides an invaluable insight into the electronic properties and microscopic mechanisms that lead to interfacial reactions, atomic restructuring, and emergent electronic properties. In the first part of this seminar, I will introduce several fundamental concepts of the methodology our group uses in its research, such as the Born—Oppenheimer approximation for the dynamics of nuclei and the notion of the first-principles potential energy surface, the efficient and rigorous treatment of nuclear quantum effects via the imaginary-time path-integral formulation, and the representation of first-principles electronic structure by machine learning models. I will then discuss in more detail how one can connect to experiments by using such simulations, in particular regarding the simulation of Raman scattering signals. I will discuss our developments of first-principles simulations of tip-enhanced Raman spectroscopy (TERS) images of molecular adsorbates on metal substrates. The results demonstrate that accurate simulations are capable of reproducing experimental measurements and serve as a powerful interpretative tool that allows us to shed light on the role of the metal substrate in shaping TERS images and discuss the underlying physics. To reach beyond the harmonic approximation, we demonstrate that the methods of molecular dynamics and machine learning can be seamlessly integrated into the TERS simulations and enable efficient, large-scale, quantitative predictions of nuclear quantum and finite-temperature effects on such spectra.
[more]
