TD-DFT benchmark for UV-visible spectra of fused-ring electron acceptors using global and range-separated hybrids

Non-fullerene acceptors, especially acceptor-donor-acceptor structured fused-ring electron acceptors (FREAs), have attracted widespread attention in organic solar cells because of their versatile molecular design in fine-tuning light absorption and energy levels. We report the accuracy of Time-Dependent Density Functional Theory (TD-DFT) for FREAs by comparing their theoretically predicted vertical absorption wavelength (lambda(ver-abso)) with the experimental maximum absorption (lambda(max)). The lambda(ver-abso) values of 50 molecules obtained from major types of FREAs have been investigated using TD-DFT by considering the solvent effects. The values of lambda(ver-abso) predicted with a pure density functional (PBE), global hybrids (B3LYP and PBE0) and range-separated schemes (CAM-B3LYP and LC-omega PBE) follow the exact exchange percentage included at an intermediate inter-electronic distance. Global hybrids outperform all other schemes. The mean absolute error provided is 22 nm by PBE0 and 38 nm by B3LYP for the whole set of molecules. The maximum deviation of 92 nm provided by B3LYP and 69 nm provided by PBE0 confirms that PBE0 is more appropriate for predicting the absorption wavelengths when designing new FREAs. By applying linear regression analysis to obtain the calibration curve, we found that the range-separated methods provide an equal or even more consistent description of FREA excited states. For the whole set of molecules, linearly corrected data yield an average error of 25 and 27 nm for CAM-B3LYP and LC-omega PBE, respectively. Consequently, when a statistical analysis technique is applicable for a certain series of FREAs, a theoretical method permits a chemically comprehensive and empirically good explanation of UV/Vis spectra for newly-designed FREAs.