General Theoretical Chemistry B02

The project develops a bottom-up multiscale framework for quantitatively modeling charge and exciton transport in organic semiconductors. Charge carriers are propagated explicitly using mixed quantum–classical non-adiabatic molecular dynamics, coupling a reduced tight-binding Hamiltonian to classical nuclear motion within a QM/MM embedding scheme. Electronic site energies, transfer integrals, and electron–phonon couplings are computed using semi-empirical DFTB and range-separated TD-DFTB methods. To overcome computational limitations, neural-network models are trained to predict Hamiltonian matrix elements and their gradients along molecular dynamics trajectories, enabling orders-of-magnitude acceleration. The methodology captures dynamic disorder, transient localization, and reorganization effects, allowing accurate prediction of intrinsic mobilities and exciton diffusion constants in crystals, polymorphs, grain boundaries, and thin-film morphologies relevant for organic electronic devices.