High-Efficiency Non-Fullerene Acceptors Developed by Machine Learning and Quantum Chemistry

Area of Science:

• Materials Science
• Organic Electronics
• Computational Chemistry

Background:

• Y6 and its derivatives have significantly advanced organic photovoltaics (OPVs) power conversion efficiency (PCE).
• Developing novel high-performance Y6 derivative acceptor materials is crucial for accelerating OPV technology progress.
• Understanding the structure-property relationships of these materials is key to designing better acceptors.

Purpose of the Study:

• To utilize machine learning (ML) and quantum chemistry to elucidate structure-property relationships in Y6 derivatives.
• To screen for new, high-performance OPV acceptor materials within a virtual chemical space.
• To provide a rational design strategy for future high-performance organic photovoltaic acceptor development.

Main Methods:

• Employed an improved one-hot encoding method for molecular representation in ML models.
• Trained a machine learning model to predict the performance of Y6 derivative acceptors.
• Utilized quantum chemistry calculations to analyze the electronic properties and surface potentials of promising candidates.

Main Results:

• The ML model demonstrated good predictive accuracy for PCE.
• Identified 22 novel Y6 derivative acceptors with predicted PCE values exceeding 17%.
• Discovered that medium-length side chains in Y6 derivatives correlate with higher performance.
• Quantum chemistry analysis indicated end acceptor units critically influence molecular orbital energies and surface electrostatic potential.

Conclusions:

• A rational design guide for high-performance OPV acceptors was established based on structure-property insights.
• The study successfully screened promising Y6 derivative candidates for OPV applications.
• The employed computational approach offers a framework for rapid materials discovery applicable to other material systems.