Dr. Marcus Elstner
Theoretical Physics
University of Paderborn
Germany

Tuesday, April 13, 2004
3:00 pm (CT)
3269 Beckman Institute

Abstract

We have developed an approximate Density Functional Theory (DFT) method, which can be derived from DFT by a second order expansion of the total energy expression with respect to the charge density. Since DFT lacks the description of dispersion forces, which is inherited by the approximate DFT method, we have included them empirically because of their crucial role for the stability of protein and DNA structures. The method has been further implemented into combined quantum mechanical molecular mechnical (QM/MM) and linear scaling algorithms. These developements allow investigations of the stability and dynamics of polypeptides, proteins, and DNA on long time scales, as exemplified by an MD simulation of crambin (650 atoms treated with QM) in solution for over 0.35 ns (a similar simulation has been performed for a DNA dodecamer in water solution). Excited state energies are calculated in the framework of time dependent density functional linear response theory. The resulting computational efficiency of this method allows molecules to fully relax in their excited states even for large systems or to perform extended MD simulations. It further enables investigating all the features of potential energy surfaces as minimum energy paths, transition states and conical intersections needed for an adiabatic analysis of photochemical problems. Applications to small organic molecules and aromatic systems demonstrate the quality of our approach, reproducing TDDFT results. However, also the shortcomings of TDDTF (GGA functionals) are inherited. We discuss some illustrative examples, such as the retinal chromophore and linear polyenes, where the investigation of potential energy surfaces exhibits a qualitatively wrong description of the respective photochemistry.

Tea and coffee will be served in R3151 Beckman Institute at 2:15pm.