Theoretical study of exciton transport in natural and synthetic light-harvesting systems

Abstract

In the first part of this dissertation, we investigate on the presence of quantum effects in the exciton dynamics of the Fenna-Matthews-Olson photosynthetic complex of green sulfur bacteria using an atomistic Quantum Mechanics / Molecular Mechanics (QM/MM) model combined with open quantum systems methods. Subsequently, we explore the theoretical connection between the atomistic QM/MM approach and the open quantum system methods and propose the correct theoretical expressions to maintain consistency when using both approaches contemporarily. In particular we show that when using the correct prefactor to extract the spectral density - the strength of coupling between excitation and other degrees of freedom - the atomistic results are in good agreement with experimental predictions. We then describe a first atomistic study of the full light-harvesting complex of green sulfur bacteria. The various units are treated atomistically and the full system's exciton dynamics is obtained using a Markovian open quantum system master equation. To conclude the first part, we describe a Machine Learning algorithm which we developed and implemented to learn time-dependent density functional theory energies by using trained neural networks and supplying these with coulomb matrices extracted from molecular dynamics simulations. This approach provides a much more rapid solution to obtaining a QM/MM Hamiltonian and subsequently extracting dynamics. It is particularly useful when multiple identical molecules are found in similar environments as one can train the network on a single molecule and predict all others. We applied this method to the Fenna-Matthews-Olson complex.

In the second part of this dissertation we focus on model systems and synthetic aggregates. In particular, we investigate the exciton dynamics in thin-film J-aggregates using a Markovian stochastic Schrödinger equation approach. We derive expressions to obtain diffusion constants from the dynamics and compare a series of different thin-film J-aggregates. The parameters of the model are obtained atomistically. From this model we obtain information on the parameters which lead to optimal exciton diffusion. This can guide the design of new exciton transfer materials.