Project 8

Non-adiabatic dynamics in LH complexes with the use of Machine Learning algorithms 

Karlsruhe Institute of Technology, Germany

Principal Investigator
Prof. Dr. Marcus Elstner

Doctoral Candidate
Ebru Akkus

Project Description:

The major goal of this project is to elucidate the excitation energy transfer (EET) mechanisms in light-harvesting complexes (LHCs), specifically light-harvesting complex II (LHC II) and fucoxanthin chlorophyll a/c-binding protein (FCP). EET is driven by factors such as site energies, excitonic couplings, and structural rearrangements, requiring accurate modeling to capture exciton dynamics. To achieve this, we employ non-adiabatic molecular dynamics (NAMD) to simulate the combined electronic-nuclear dynamics using the long-range corrected time-dependent density functional tight binding (TD-LC-DFTB) method within quantum mechanical/molecular mechanical (QM/MM) framework, which provides an efficient modeling quantum and nuclear interactions during EET processes. However, these simulations remain computationally demanding for large systems. To address this challenge, we will utilize neural networks (NNs) that trained on TD-LC-DFTB data to accurately predict site energies, excitonic couplings, and forces at a significantly reduced computational cost. By integrating NNs, we aim to accelerate the investigation of EET processes and deepen our understanding of the structural and functional interactions that drive energy transfer in light-harvesting complexes.

Publications:

  1.  Monja Sokolov, David S Hoffmann, Philipp M Dohmen, Mila Krämer, Sebastian Höfener, Ulrich Kleinekathöfer, and Marcus Elstner. Nonadiabatic molecular dynamics simulations provide new insights into the exciton transfer in the fenna–matthews–olson complex. Physical Chemistry Chemical Physics, 2024.
  2. Farhad Ghalami, Philipp M Dohmen, Mila Krämer, Marcus Elstner, and Weiwei Xie. Nonadiabatic simulation of exciton dynamics in organic semiconductors using neural network-based frenkel hamiltonian and gradients. Journal of Chemical Theory and Computation, 2024.
  3. B. M. Bold, M. Sokolov, S. Maity, M. Wanko, P. M. Dohmen, J. J. Kranz, U. Kleinekathöfer, S. Höfener, and M. Elstner. Correction: Benchmark and performance of long-range corrected time-dependent density functional tight binding (LC-TD-DFTB) on rhodopsins and light-harvesting complexes. Phys. Chem. Chem. Phys., 25:22535–22537, 2023.