Abstract
Understanding the behavior of electron spins in nanomaterials is crucial for the advancement of quantum technologies. However, studying single spins presents significant challenges both experimentally and theoretically, as radicals are often unstable under experimental conditions, and sophisticated quantum mechanical approaches with extremely high computational demands, such as CASPT2 and NEVPT2, are often required. This has resulted in a limited understanding of many spin processes at the nanoscale, as emphasized by recent cutting-edge scanning probe microscopy experiments exploring conjugated hydrocarbons known as nanographenes. Some of these systems have diradical character, with two unpaired π-electrons in a non-Kekulé electronic structure that yields magnetic and electric properties not seen in classical hydrocarbons. In this work, we use a recently developed (range-separated) multiconfigurational on-top pair density functional theory (MC-ctPDFT or MC-sr-ctPDFT) method to study the magnetic coupling in both small model non-Kekulé diradicals and models of nanographenes. We show that standard density functional theory fails to correctly describe those molecules whose ground state is a singlet instead of a triplet, sometimes considered a violation of Hund's rule. Our method accounts for both static and dynamic correlations and compares well with common approaches such as NEVPT2 and CASPT2, but exhibits a much lower asymptotic computational cost. By expansion of the method to encompass more complex molecular environments, it could be further utilized to investigate surface interactions and catalytic processes.