Neue Materialien errechnen : HPC in der Vorhersage neuartiger elektronischer und magnetischer Eigenschaften auf der Nanoskala mittels atomistischer quantenmechanischer Simulationen
Der vorliegende Artikel illustriert den Beitrag der theoretischen Materialphysik im Bereich Hochleistungsrechnen mit dem Fokus auf Materialien für elektronische Bauelemente sowie für die Energiegewinnung (Thermoelektrika), Energiespeicherung (Wasserspaltung) und Wärmetransport (magnetische Kühlung).
State-of-the-art supercomputers have meanwhile become an indispensable tool in the quest for new materials addressing the technological challenges of our society. One important strategy is a knowledgebased search relying on the understanding of materials properties on the electronic scale with parameterfree, material-specific calculations in the framework of density functional theory. This method yields an accurate, quantum-mechanical access to materials properties, which is independent from experiment but in turn also very computationally expensive, in particular, when the size of investigated systems becomes large. Further challenges concern the treatment of electron-electron interactions. Within this contribution, we illustrate how high performance computing contributes to this task using three paradigmatic examples related to interface-induced electronic and magnetic phases, materials for energy harvesting and magnetic refrigeration. Transition metal oxide heterostructures not only allow us to combine functional properties in one composite material, but, most importantly, novel electronic states can be stabilized at their interfaces that are not available in the bulk parent compounds. In this contribution we show how the functionality at the interface can be tailored exploring systematically a set of parameters such as polar discontinuities that can give rise to 2D conductivity and/or magnetic phases between insulators. Moreover, strain and confinement can tune the electronic behavior between a metal and an insulator. Last but not least, buckled honeycomb patterns in oxide superlattices can lead to exotic topologically nontrivial states that bear analogies to graphene with additional functionality due to the correlated nature of d and f-electrons. Furthermore, we demonstrate how strain can be used to enhance the thermoelectric performance of highly anisotropic cobalt delafossites. In future, one may thus expect to exploit spin- and orbital degrees of freedom in the design of electronic devices or the development of highly efficient thermoelectric generators. The third example aims at the intimate coupling between magnetic, lattice and electronic degrees of freedom in modern magnetocaloric materials. Their understanding is of primary importance for the development of energy-saving refrigeration and air-conditioning concepts.