Error-controlled multiscale modeling approaches to surface chemistry and catalysis

COLL 10

Karsten Reuter, reuter@fhi-berlin.mpg.de, Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, 14195 Berlin, Germany
A materials science modeling that is based on understanding, predictive, and applicable to a wide range of realistic conditions requires to cover a wide range of length and time scales. Electronic structure theory deals with the finest scale and is thus the base for such a multiscale modeling of materials properties and functions. First-principles statistical mechanics and continuum mechanics techniques represent the next levels that build on this basis. Using examples from heterogeneous catalysis I will demonstrate the approach using density-functional theory, kinetic Monte Carlo and rate equation theory to treat the electronic, statistical and continuum levels, respectively. I will discuss the need for error-controlled links between the levels, e.g. using sensitivity analyses, as well as an approach to address binding interactions that challenge present-day exchange-correlation functionals.