Modeling materials with optimized transport properties

Abstract

Following demands for materials with peculiar transport properties, e.g. in magnetoelectronics or thermoelectrics, there is a need for materials modeling at the quantum-mechanical level. We combine density-functional with various scale-bridging tools to establish correlations between the macroscopic properties and the atomic structure of materials. For examples, magnetic memory devices exploiting the tunneling magneto-resistance (TMR) effect depend crucially on the spin polarization of the electrodes. Heusler alloys, e.g. Co2MnSi, if perfectly ordered, are ferromagnetic half-metals with (ideally) 100% spin polarization. Their performance as electrodes in TMR devices is limited by atomic disorder and deviations from perfect stoichiometry, but also by interface states at the tunneling barrier. We use ab initio thermodynamics in conjunction with the cluster expansion technique to show that excess manganese in the alloy and at the interface helps to preserve the desired half-metallic property. As another example, nanostructured materials with a reduced thermal conductivity but good electrical conductivity are sought for applications in thermoelectrics. Semiconductor heterostructures with a regular arrangement of nanoscale inclusions ('quantum dot superlattices') hold the promise of a high thermoelectric figure of merit. Our theoretical analysis reveals that an increased figure of merit is to be expected if the quantum dot size, the superlattice period and the doping level are all suitably fine-tuned. Such a superlattice thus constitutes a material whose transport properties are controlled by geometrical features at the nanoscale.