Accurate Prediction of Vertical Emission from Excited-State Tuning of Range-Separated Density Functional Theory
Journal
Journal of Physical Chemistry C
Journal Volume
124
Journal Issue
33
Pages
17964-17970
Date Issued
2020
Author(s)
Abstract
We report for the first time an excited-state (EX) tuning approach to accurately predict vertical emissions (VEs) based on the study of 36 medium and large organic molecules (∼66 heavy atoms) used mostly in thermally activated delayed fluorescence applications. Compared to the conventional ground-state (GS) tuning of the range-separation parameter (ω) in long-range-corrected functionals, which by far provides accurate optical properties (compared to B3LYP, M062X, CAM-B3LYP, ωB97XD, etc.), the proposed EX tuning approach [LC-ωPBE/6-31G(d)] predicts more accurate VEs in the density functional theory (DFT) framework. The mean absolute errors (MAEs) of VEs reduced by 20%, from 0.25 eV with GS tuning to 0.20 eV with EX tuning, against the experimental benchmarks. We find that further improvement in accuracy is possible if accurate EX geometries are employed in the EX tuning approach. For nine smaller molecules (for which higher level optimized EX geometries are available at CCSDR(3) and CC3), MAEs of VEs reduced from 0.32 to 0.10 eV, a nearly 70% improvement (against theoretical benchmarks) employing higher level EX geometries compared to DFT EX geometries. Tuning the range-separation parameter based on the optimized EX geometry enables us to capture the delocalization patterns of EX accurately, which results in accurate prediction of VEs. These results prove that the EX-tuning approach in combination with accurate EX geometries provides the most accurate VEs comparable to CC3 benchmarks.
SDGs
Type
journal article