. .

Computational Materials Science

The lab Computational Materials Science headed by Erik Koch is now fully integrated into the Forschungszentrum Jülich.

Research
Materials science is arguably the field where physics has the greatest impact on our everyday lives. Striking examples are the development of key technologies like the transistor, solid-state lasers, or giant magneto-resistance devices. Up to now these developments have relied on the simple band-picture of solids, that is the basis for the classification of materials into metals, semiconductors, and band insulators.

speedup   spin-charge separation
The method for describing materials within this picture, density-functional theory, has been developed about fifty years ago and is now a mature and highly successful technique. Using such ab-inito methods it is possible to understand the electronic structure of a large class of solids and even to describe the dynamics of chemical reactions via molecular dynamics simulations. Despite these immense successes, there are wide classes of materials where density-functional methods fail not only quantitatively but qualitatively. Typical situations are materials with localized orbitals, e.g. of transition metals or rare earths. Such strongly correlated systems have recently moved into the center of attention because of their unusual properties that make them prime candidates for future technologies. Examples are the high-temperature superconducting cuprates or the colossal magneto-resistance manganites. A second important class is comprised of the correlated organic crystals like the superconducting Fullerides or low-dimensional organic charge-transfer salts. Moreover, many important proteins, a typical example being the oxygen-carrier hemoglobin, rely on correlated transition metal complexes for their biological functionality. The basis for the unusual properties of these materials is the correlation between their electrons, which means that we cannot attribute properties to individual electrons but have to confront the full quantum mechanical many-body problem. So the very basis of what makes correlated materials interesting also makes their simulation extremely challenging.

Realistic simulations of correlated materials have only recently become possible by the convergence of two developments: the formulation of a practical method, dynamical mean-field theory, for describing correlated systems and the availability of massively parallel computers that make the non-perturbative solution of the resulting models possible. Our group addresses all the relevant issues of these recent developments in materials science: We work on applications for transition-metal oxides and correlated organic materials as well as on the methodological development of dynamical mean-field theory and on massively parallel implementations of non-perturbative many-body solvers. Our work is embedded in the recently (July 2010) formed DFG Forschergruppe FOR 1346: Dynamical Mean-Field Approach with Predictive Power for Strongly Correlated Materials. In Jülich we have a particularly fruitful collaboration with the group of Prof. Pavarini at the IAS/IFF at Forschungszentrum Jülich. Our algorithmic developments for the IBM Blue Gene and for the Cell Chip are frequently featured together with the Jülich Supercomputer Center.

Teaching

Teleteaching

Advanced Solid State Physics (Mark Jarrell)