Ultrafast Linear Scaling Quantum Chemical Methods: Methodology and Applications to Problems in Materials Sciences

Understanding chemical structure and reactivity at physical interfaces is fundamentally important for modern society. Interfacial structure and reactivity are involved in myriad everyday processes: catalysis, tribology, electrochemistry, industrial-scale chemical processing and the production of advanced materials. They are also of crucial interest in the context of nuclear fusion where a better understanding of plasma-wall interactions (PWI) requires accurate atomic-scale modeling. Unfortunately, despite leaps in modern computer technology, quantum chemical simulations remain incapable of probing interfacial structure and reactivity on experimentally relevant scales. Consequently, small-scale theoretical models of physical interfaces remain largely disconnected from their real-world counterparts.  In order to bridge this gap, our group has recently developed an ultrafast linear-scaling quantum chemical method by the combination of an approximate density functional theory method (DFTB) [1] and the fragment molecular orbital (FMO) [2] approach we call “FMO-DFTB” [3] for molecular systems such as polymers and liquid/solid interfaces. In collaboration with the Nakai group we also developed a divide-and-conquer-based DFTB method (DC-DFTB) [4] more suitable for extended systems. Both methods can be employed on massively parallel computer architectures. In the presentation I will briefly introduce these methods and report recent DFTB-based simulations of PWI and graphene growth on transition metal surfaces as representative examples.