IBM Research Award 2021 – Pablo Miguel Piaggi
Entropy as a tool for crystal discovery
EPFL thesis n°9148
Thesis directors: Prof. N. Marzari, Prof. M. Parrinello
For the development of advanced simulation methods that have allowed the phase diagrams of complex materials to be computed.
Many substances can solidify into different crystal structures - a phenomenon called polymorphism. One of the most common examples of this behavior is carbon that can crystallize both as diamond and graphite. Recently, the computational prediction of crystal structures has emerged as an useful alternative to expensive experiments. In this thesis, we proposed an approach to the prediction of polymorphism based on reproducing the crystallization process on the computer. The main hurdle faced by such an approach is that crystallization usually takes place in timescales much longer than those that can be afforded with standard molecular simulations. In order to circumvent this difficulty we developed a method to promote crystallization. This approach can only have true predictive power if it does not include information about any particular crystal structure. Therefore, we took inspiration from thermodynamics and proposed to use an entropy surrogate as an order parameter to drive crystallization. In this way we were able to explore polymorphism in many different substances and discover a possible new polymorph of urea stabilized by entropic effects. We also exploited the entropy surrogate for the characterization of atomic environments and the classification of the polymorphs that crystallize during the simulations.