Abstract
It is highly desirable to enhance the heat conduction of polymeric heat-dissipating materials for applications in power electronics, automotive components, light-emitting diodes, and telecommunication equipment. However, the thermal conductivity of polymeric materials is one to three orders of magnitude lower than that of metals and ceramics due to phonon scattering in amorphous regions. Several attempts have been made to overcome this limitation by increasing heat flow along the orientation axis of the self-assembly of liquid crystalline polymers, but the design of liquid crystalline polymers remains largely empirical. In this study, we developed a machine learning model that can predict with over 96% accuracy whether a polymer will form a liquid crystalline state based on its compositional or structural characteristics. By exploring the inverse mapping of this model, we identified a comprehensive set of chemical structures for liquid crystalline polyimides. De novo monomer synthesis and polymerization reactions of six selected polymers were performed, and all synthesized polymers were experimentally confirmed to form liquid crystalline structures with thermal conductivities ranging from 0.722 to 1.26 W m−1 K−1. These are the first liquid crystalline polymers predicted and discovered via machine learning.
Supplementary materials
Title
Supporting Information
Description
Descriptions of machine learning methodologies, synthesis procedures, and results from measurements including NMR, FT-IR, density, specific heat, μ-TWA, FSC, and X-ray structural analysis.
Actions