Non-Markovian Memory Effects in Molecular Dynamics with Electronic Friction

What is this about?

In contrast to the conventional chemistry in labs, my project is a fully programming-based and computational theories demanding research. It actually has more physics, mathematics and CS than chemistry.

Imagine we fire a particle onto a surface, this particle will interact with the surface atoms and change the electronic configuration of the whole system. This type of configurational perturbation kicks in by energy transfering from the particle to the surface (i.e. the virbational energy of particle). This amount of energy causes some of the electrons hop onto higher energy states which is so called electronic excitiation. In some cases, partcle might bounce back after the collision eventually where it is how scienists define it as scattering. There can be an another case called desorption where the particle is absorbed by the surface where its chemical bonds might be broken as a result of the collision. In this way, we fundamentally explain why some metal calayst could accelerate the chemical reactions. A better theroetical understanding of these processes can help us to design more efficient medium for releasing or storing energy such as battery, fuel cell, etc. What's more, if we replaced the metal with seminconductor and particle with photon, our knowledge would be applied into advancing the photolithography on the seminconductor surface where helps those integrated circuit factories like TSMC, Qualcomm and Intel to produce smaller and faster chips.

Instead of spending time and money on the experiments, an explainable and easy-to-use computational model will accelerate the designs and testings of those material and particle. This is where my project and most of the projects in my group come in.

Why am I doing this?

Back to the electronic excitiation brought by the energy transfering (energy coupling), the electrons fail to adjust themselves with the nuclear motion instantaneously showed by many experimental results. This phenomenon implies the break down of Born-Oppenheimer approximation. Frankly speaking, the nuclear motion is no longer simply determined by the a single adiabatic potential energy surface. Instead, the nuclear motion is affected by non-adiabatic coupling/superposition between those adiabatic PESs. Electronic friction is derived for accounting this non-adiabatic effect while we keep using the adiabatic PESs as a driving force to the nuclei.

Most of the electronic friction nowadays are simplied into a Markovian form where non-adiabatic effect is determined by the instantaneous nuclear velocity with it friction at that moment. This is a very coarse-grained approximation where assuming the particle's spectrum function only simpling only one frequency from the bath's spectrum(density of states). It leads us to a memoryless friction where the particle's past history is not taken into account. However, the complete form of the electronic friction should involve the memory effects where particle should experience multiple frequencies from the bath's spectrum in the reality.