Active discovery of organic semiconductors

Area of Science:

• Materials Science
• Computational Chemistry
• Organic Electronics

Background:

• Organic molecules offer vast design possibilities for organic semiconductors (OSCs).
• Efficient search strategies are crucial for discovering new OSCs within this large design space.
• Current computational methods can be limited in exploring the full potential of OSCs.

Purpose of the Study:

• To develop and evaluate an active machine learning (AML) approach for exploring the design space of organic semiconductors.
• To identify novel OSC candidates with superior charge conduction properties.
• To demonstrate the efficiency of AML compared to traditional computational methods.

Main Methods:

• Utilized an active machine learning (AML) strategy combined with molecular morphing operations.
• Employed charge injection and mobility descriptors to evaluate OSC candidates.
• Integrated first-principles calculations to refine a surrogate model.
• Visualized the chemical space exploration as a network for methodological insight.

Main Results:

• The AML approach significantly outperformed conventional computational funnel methods.
• Successfully identified known and novel molecular OSC candidates with enhanced charge conduction properties.
• Demonstrated continuous discovery of high-efficiency OSC candidates within an unlimited search space.

Conclusions:

• Active machine learning provides an efficient and powerful strategy for discovering high-performance organic semiconductors.
• The presented AML approach enables rapid exploration of vast chemical spaces for materials discovery.
• This methodology holds significant promise for advancing organic electronics through novel material identification.