Identifying the Quantitative Relationship Between the Molecular Structure and the Horizontal Transition Dipole Orientation of TADF Emitters

Abstract

Realizing the horizontal orientation of molecular transition dipole moment (TDM) can greatly improve the out-coupling efficiency and the resultant external quantum efficiency (EQE) of organic light-emitting diodes (OLEDs). Herein, key parameters governing the horizontal TDM have been continuously explored. However, quantitatively identifying the key parameters from the molecular structure viewpoint is rather challenging due to the complexity of the influencing parameters. Here, by training the machine learning (ML) models using the experimental results, the quantitative relationship between the molecular structure and the horizontal TDM ratio (ϴ) of thermally activated delayed fluorescent (TADF) emitters in the host-guest films is identified. The molecular structure is represented by either quantum chemistry-calculated structural descriptors or topological/physical/chemical molecular descriptors. Key descriptors are ranked and can be used for guiding molecular structure design. Moreover, the accuracy of ML models is double-verified by comparing the predicted results with experimental ϴ values and the trend of experimental EQE based on a group of materials. Using compressed sensing technology, the low-dimension material space is also visually constructed based on key descriptors, and the results are consistent with those of the ML models.