# Probing Glass Formation in Perylene Derivatives via Atomic-Scale Simulations and Bayesian Regression

## TL;DR

This paper uses simulations and statistical methods to study how perylene derivatives form glassy materials, linking molecular behavior to macroscopic properties.

## Contribution

A novel workflow combining atomic-scale simulations and Bayesian regression to study glass formation in chromophores.

## Key findings

- Predicted glass transition temperatures and kinetic fragilities align semiquantitatively with experiments.
- β and α-relaxation processes are linked to caged dynamics and cooperative rotations.
- Force field overestimation of intermolecular cohesion may explain remaining prediction errors.

## Abstract

While the structural dynamics of chromophores are of
interest for
a range of applications, it is experimentally very challenging to
resolve the underlying microscopic mechanisms. At the same time, glassy
dynamics are also challenging for atomistic simulations due to the
underlying dramatic slowdown over many orders of magnitude. Here,
we address this issue by combining atomic scale simulations with autocorrelation
function analysis and Bayesian regression, and apply this approach
to a set of perylene derivatives as prototypical chromophores. The
predicted glass transition temperatures and kinetic fragilities are
in semiquantitative agreement with experimental data. We suggest that
the remaining error could be caused by an overestimation of the intermolecular
cohesion by the force field used in this work. By analyzing the underlying
dynamics via the normal vector autocorrelation function, we are able
to connect the β and α-relaxation processes in these materials
to caged (or librational) dynamics and cooperative rotations of the
molecules, respectively. The workflow presented in this work serves
as a stepping stone toward understanding glassy dynamics in many-component
mixtures of perylene derivatives and is readily extendable to other
systems of chromophores.