Machine Learning-Assisted Efficient Discovery and Rational Design of Thermally Conductive Polymers.

Area of Science:

• Materials Science
• Polymer Science
• Computational Chemistry

Background:

• Advancing electronics require materials with enhanced thermal management.
• Polymers typically exhibit poor thermal conductivity, hindering their use in high-power devices.
• High thermal conductivity (TC) in polymers is crucial for flexible electronics and optoelectronics.

Purpose of the Study:

• To develop a machine learning-assisted framework for rapid screening and rational design of polymers with high TC (> 0.40 W m^-1 K^-1).
• To accelerate the discovery and design of novel polymers for advanced thermal management applications.

Main Methods:

• Utilized a deep neural network to correlate polymer microstructures with their thermal conductivity.
• Employed high-throughput screening to identify promising polymer candidates.
• Integrated Monte Carlo tree search and molecular generation rules for rational design of new structures.
• Analyzed the impact of chain stiffness, conformation, and bond strength distribution on TC in amorphous systems.

Main Results:

• Successfully screened and designed polymers exhibiting high thermal conductivity.
• Identified chain stiffness as a key factor influencing polymer TC.
• Established a link between chain conformation, bond strength distribution, and thermal transport in amorphous polymers.

Conclusions:

• The proposed machine learning framework offers an efficient approach for discovering and designing high TC polymers.
• This strategy accelerates the development of advanced polymer materials for thermal management in electronics.
• Demonstrated the importance of molecular design principles for optimizing polymer thermal properties.