Thermal, spin, and electric transport from first principles for engineering applications
Title: Thermal, spin, and electric transport from first principles for engineering applications
SNIC Project: SNIC 2013/26-12
Project Type: SNAC Large
Principal Investigator: Anna Delin <>
Affiliation: KTH Royal Institute of Technology
Duration: 2014-01-01 – 2015-01-01
Classification: 10304 10407


In this project, transport properties of magnetic and nonmagnetic thermoelectric materials as well as graphene systems will be calculated using computational methods based on quantum mechanics, notably density functional theory (DFT) based methods and build-ons on such methods, e.g. spin-dynamic calculations. Specifically, we will continue to investigate the effects that nanostructuring and the spin degree of freedom may have on the thermoelectric properties. Polaron transport in polymers will be investigated. The electrical conductivity of graphene systems, where different types of atoms are chemisorbed or physisorbed on top of the graphene layer, seems to behave in counterintuitive ways. We are investigating these effects in order to explain the physics behind the observed phenomena. Further, a very intriguing prospect is to generalize the thermoelectric concept to spin voltages and spin currents, leading to spin caloritronics. Transport properties are often rather complex and difficult to calculate from first principles. For instance, only recently, HPC resources have become powerful enough to allow advanced calculations, founded in quantum mechanics, of thermoelectric effects and conductivity in systems where disorder play an important role. Recent method development has also helped to make this subject come into focus for computations. Thermoelectric devices promise to play very important roles in future global sustainable energy solutions, since they enable the recovery of energy otherwise just wasted as heat. The VASP, SIESTA/SMEAGOL, CPMD, and Quantum Espresso calculations are envisaged to be mainly concentrated to Triolith and Abisko, whereas the UppASD calculations will be, for the most part, performed on Lindgren. In the latter calculations, the number of atoms in a calculation can be counted in billions, and the code runs efficiently on over 10 000 cores in parallel. Our average total usage during the period Jan 2013-Oct 2013 was 685 kcore hours per month - well above our allocation of 650 kcore hours/month.