Using quantum chemical methods, we can calculate the potential energy landscape of almost any given molecule. Then, molecular dynamics methods can be used to travel on this landscape, effectively mimicking a virtual experiment, and calculate any property of interest.
Unfortunately, for many interesting systems (such as proteins, heterogeneous solids and complex liquids) and under many environmental conditions, such a theoretical treatment remains out of reach due to the associated prohibitive computational costs.
How can we describe a large-scale, complex system with over a million reacting atoms?
In my research, I combine theoretical, software engineering and modeling efforts to address this challenge. Specifically, I develop ReaxFF reactive force fields, fitted to costly quantum chemical calculations, for subsequent sampling techniques. Reactive molecular dynamics is used to sample the developed PES and follow the natural flow of events on the atomic scale. Energy Landscape techniques are employed to circumvent high barriers and accurately characterize specific chemical transformations with no need for arbitrary projections to low-dimensional spaces.
Prominent applications include elucidation of prebiotic mechanisms of peptide formation on early Earth, mechanochemistry of high-energy materials and hydrogen-bonded networks dynamics in the solid state.