Skip to main content

Accelerating Theoretical Spectroscopy through Machine-learning

Supervisor: Dr Nicholas Hine (Physics)

Interactions between light and matter are fundamental to both technological (photovoltaics, sensors and displays) and natural systems (photosynthetic pigment-protein complexes, vision receptors). Quantum mechanical modelling can predict optical properties even of large, complex systems rather accurately, but the required calculations are too demanding to be used for dynamically-evolving systems such as a dye molecule in solvent. In this project, we will combine state-of-the-art QM modelling with machine-learning techniques to dramatically accelerate prediction of optical properties over long length- and time-scales, leading to quantifiably accurate predictions of properties such as colour, excited state lifetimes and photostability.

bchl_in_fmo

Figure: Excited state of bacteriochloropyhll in the Fenna-Matthews-Olsen Pigment Protein Complex