Abstract
Molecular monolayers on a supporting substrate can order in different configurations, i.e., polymorphs. These different polymorphs can also exhibit distinctly different optical properties owing to variations of the intermolecular coupling between the transition dipole moments, which in turn may affect transition energies and oscillator strengths. In this study, we computationally investigate the impact of polymorphism of organic molecular monolayers on optical absorption spectra using a combination of machine-learning-assisted structure search and density functional theory. Specifically, we systematically detail various influencing factors, including the geometric distortions upon adsorption, interactions between transition dipole moments, and changes in selection rules due to variations in symmetry of different polymorphs. As an example, we predict the polymorphism of the dipolar organic molecule 2-nitro-pyrene-7-amine on NaCl(100) using a machine learning-based structure search and calculate its optical properties based on density functional theory and the random phase approximation. Specifically, we find two different polymorphs where the lowest excitation energy significantly differs by approximately 0.2 eV.