Machine Learning for the Design of Novel OLED Materials
Hadi Abroshan, Paul Winget, H. Shaun Kwak, Yuling An, Christopher T. Brown, Mathew D. Halls
Abstract
One of the central paradigms in materials science is that data-driven methods will decrease the time needed to develop optimal solutions. In other words, the time required to go from discovery to market is partially a product of the historical trial-and-error approach to designing novel materials. Explicit integration of a tightly-woven data feedback loop into design workflows has led to rapid ideation and additional insight into numerous material applications. High-throughput virtual screening (HTVS) for materials discovery and optimization was an early implementation of this approach. Currently, the computation of molecular properties, with reasonable accuracy, of thousands of molecules using density functional theory (DFT) is routine, enabling the development of machine learning (ML) models to reproduce calculated quantities explicitly. One particular example of data-driven methods used for materials design is the discovery of materials for organic electronics. This chapter briefly reviews some of the latest efforts to develop organic light-emitting diode (OLED) materials using ML techniques.