Assigning the Electronic Absorption Spectral Features of Highly Conjugated Organic Macromolecules with Multiscale Simulation and Machine Learning

Conjugated organic macromolecules are excellent sources for light-emitting diodes and can be tuned to have distinct response properties. Our chemical intuition provides rough guidelines for interpreting optical response but obtaining a more quantitative answer requires the use of at-least time-dependent density functional theory (TD-DFT) methods. TD-DFT provides a comparatively low-cost avenue for direct calculation of the electronic spectroscopy in the gas phase; however, there is a need to consider the ensemble of conformations present in the condensed phase to interpret the spectra. To definitively understand and guide future ideas in next-generation materials design, developing new methods and adapting current protocols are needed to calculate and interpret absorption spectral features for conjugated macromolecules with physically informed and intelligent searches. Here, we demonstrate our utilities in our approaches for shedding light on unique features in UV-Vis spectra of conjugated organic macromolecules, by generating conformers representative of the UV-Vis spectrum from experiment. Using multiscale simulation and Bayesian optimization methods, we decipher the key internal coordinates responsible for electronic absorption spectral features in quantifying our protocol with organic macromolecules that possess interesting optical properties, and comment on the future developments.