Quantum-inspired genetic algorithm for designing thermally conductive polymers

Abstract

Designing polymers with high intrinsic thermal conductivity (TC) faces challenges due to the vast chemical space and substantial resource requirements of conventional methods. Here, we develop a quantum-inspired genetic algorithm (QGA) that integrates quantum computing concepts with classical evolutionary optimization to enable efficient polymer design. Using a deep neural network trained on molecular fingerprints as a surrogate model for rapid property evaluation, the QGA demonstrates superior optimization capability and convergence stability compared to classical genetic algorithms in designing ternary alternating copolymers. When applied to the design of pentameric alternating copolymers within a candidate space comprising over 1 × 107 possible structures, the method successfully identified that 10.4% of the 9975 designed candidates achieved a predicted TC > 0.40 W m−1 K−1. Molecular dynamics simulations validate the predictions, while structural analysis reveals that rigid, conjugated fragments serve as critical building blocks that facilitate thermal transport primarily through intra-chain energy transfer. This work establishes an effective strategy for inverse design of thermally conductive polymers and demonstrates the potential of quantum-inspired optimization in the development of advanced materials.