Non-adiabatic Couplings in Surface Hopping with Tight Binding Density Functional Theory: The Case of Molecular Motors

Hybrid Functional DFTB Parametrizations for Modeling Organic Photovoltaic Systems

Density functional tight binding (DFTB) is a quantum chemical simulation method based on an approximate density functional theory (DFT), known for its low computational cost and comparable accuracy to DFT. For several years, the application of DFTB in organic photovoltaics (OPV) has been limited by the absence of an appropriate set of parameters that adequately account for the relevant elements and necessary corrections. Here we have developed new parametrizations using hybrid functionals, including B3LYP and CAM-B3LYP, for OPV applications within the DFTB method in order to overcome the self-interaction error present in DFT functionals lacking long-range correction. These parametrizations encompass electronic and repulsive parameters for the elements H, C, N, O, F, S, and Cl. A Bayesian optimization approach was employed to optimize the free atom eigenenergies of unoccupied shells. The effectiveness of these new parametrizations was evaluated by a data set of 12 OPV donor and acceptor molecules, showing consistent performance when compared with their corresponding DFT references. Frontier molecular orbitals and optimized geometries were examined to evaluate the performance of the new parametrizations in predicting ground-state properties. Furthermore, the excited-state properties of monomers and dimers were investigated by means of real-time time-dependent DFTB (real-time TD-DFTB). The appearance of charge-transfer (CT) excitations in the dimers was observed, and the influence of alkyl side-chains on the photoinduced CT process was explored. This work paves the way for studying ground- and excited-state properties, including band alignments and CT mechanisms at donor-acceptor interfaces, in realistic OPV systems.

Solvent Effects on Nonadiabatic Dynamics: Ab Initio Multiple Spawning Propagated on CASPT2/xTB Potentials

An approach to simulate nonadiabatic dynamics in solution is introduced, which relies on the propagation of the nuclear wavepacket with the Ab Initio Multiple Spawning (AIMS) method under the effect of potential energy calculated with a hybrid but fully quantum mechanical scheme (QM/QM'). The electronic energies of the excited states of the chromophore are calculated with multireference perturbation theory (CASPT2), and the embedding molecules are described with a tight binding Hamiltonian (GFN2-xTB). This implementation is fully open source and relies on the combination of PySpawn, OpenMolcas, and xTB. Additionally, ORCA is used to properly generate the initial conditions in solution, showing how the combination of cutting-edge implementations in several commonly used software can push the state of the art of nonadiabatic dynamics in solution toward a new high standard of accuracy. The dynamics of ethylene in vacuum, in acetone, and in chloroform is reported as a test case, with a detailed analysis of the AIMS runs that shows important geometrical and electronic effects of the solvents on the decay mechanism of the chromophore.