Metadynamics for automatic sampling of quantum property manifolds: exploration of molecular biradicality landscapes

We present a general extension of the metadynamics allowing for an automatic sampling of quantum property manifolds (ASQPM) giving rise to functional landscapes that are analogous to the potential energy surfaces in the frame of the Born-Oppenheimer approximation. For this purpose, we employ generalized electronic collective variables to carry out biased molecular dynamics simulations in the framework of quantum chemical methods that explore the desired property manifold. We illustrate our method on the example of the "biradicality landscapes", which we explore by introducing the natural orbital occupation numbers (NOONs) as the electronic collective variable driving the dynamics. We demonstrate the applicability of the method on the simulation of p-xylylene and [8]annulene allowing to automatically extract the biradical geometries. In the case of [8]annulene the ASQPM metadynamics leads to the prediction of biradical scaffolds that can be stabilized by a suitable chemical substitution, leading to the design of novel functional molecules exhibiting biradical functionality.