Deep Learning-Based Chemical Similarity for Accelerated Organic Light-Emitting Diode Materials Discovery

Area of Science:

• Materials Science
• Organic Electronics
• Computational Chemistry

Background:

• Thermally activated delayed fluorescence (TADF) materials are crucial for efficient, metal-free organic light-emitting diodes (OLEDs).
• Current computer-aided design strategies for novel TADF materials are limited by computationally tractable properties.
• Accelerating the discovery of new TADF emitters is essential for advancing OLED technology.

Purpose of the Study:

• To introduce TADF-likeness, a novel quantitative score for evaluating the TADF potential of molecules.
• To leverage a data-driven approach based on chemical similarity to existing TADF materials.
• To provide a cost-effective and efficient method for accelerating the discovery of new TADF materials.

Main Methods:

• Utilized a deep autoencoder to identify common features among existing TADF molecules using chemical descriptors.
• Developed a quantitative score, TADF-likeness, based on the characterized features and chemical similarity.
• Applied the TADF-likeness score in large-scale virtual screening of millions of molecules.

Main Results:

• The TADF-likeness score demonstrated a high correlation with four essential electronic properties of TADF molecules.
• The score achieved a high success rate in identifying promising TADF candidates during virtual screening.
• The method proved to be highly feasible and cost-effective for accelerating TADF material discovery.

Conclusions:

• TADF-likeness offers a powerful, data-driven tool for predicting and discovering novel TADF materials.
• This approach significantly accelerates the identification of efficient, metal-free OLED emitters.
• The concept of TADF-likeness is adaptable for materials discovery in other scientific fields.