Dresden 2017 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
O: Fachverband Oberflächenphysik
O 107: Heterogeneous Catalysis: Theory II
O 107.7: Talk
Friday, March 24, 2017, 12:00–12:15, TRE Phy
Is DFT accurate enough for modeling chemical kinetics? — •Sandra Döpking1, Daniel Strobusch2, Christoph Scheurer2, Craig Plaisance2, Karsten Reuter2, and Sebastian Matera1 — 1FU Berlin, Germany — 2TU München, Germany
Targeting at an understanding of interplay between bond making and bond breaking and observed reactivity, first-principles based kinetic methodologies for heterogenous catalysis employ energetic information derived from electronic structure calculations. These energies, usually obtained on the density functional theory (DFT) level, have typically a potential error of 0.2 eV-0.3 eV.
We present a systematic approach to address the propagtion of these errors to the kinetic model’s result based on adaptive sparse grids for the sampling of the error space. We demonstrate the approach on a model for the oxygen evolution on Co3O4[1]. Our results indicate that the simulated reactivities carry an uncertainty of more than five orders of magnitude, questioning the trustability of the model. However, a decomposition of this uncertainty into contributions of different input errors reveals that only a small number of DFT energies have an impact. So, it is still possible to draw conclusions about the potential driving forces behind catalytic activity, although the activity itself can not be estimated accurately.
Besides a qualified discussion of the effect of the modeling error, the approach can be employed for the reduction of the number of expensive first-principles simulations or for computational materials screening.
[1] C. Plaisance and R. A. van Santen, J. Am. Chem. Soc. 137, 14660-14672 (2015)