单体
材料科学
聚合物
共聚物
电介质
数量结构-活动关系
玻璃化转变
高分子化学
热力学
复合材料
计算机科学
机器学习
物理
光电子学
作者
Hiroaki Taniwaki,Hiromasa Kaneko
摘要
Abstract The dielectric constant (DC) and glass transition temperature (Tg) properties were used to design polymers by quantitative structure–property relationship (QSPR). Using polymer data sets that included homopolymers and copolymers, property prediction models were constructed using the polymer properties and molecular descriptors calculated from monomer structures. The best combination of regression methods, types of descriptors, and preprocessing methods for constructing regression models were determined, and thus, DC and Tg prediction models with a high prediction accuracy could be constructed. Then, theoretical monomer structures, and theoretical copolymers with different combinations of monomer structures and composition ratios were generated, and the DC and Tg of the generated structures were predicted using the constructed models. Candidate structures met the desired DC and Tg values, and the proposed method could select promising monomer structures and composition ratios. Thus, monomer structures that would produce polymers with a high or low DC and high Tg were designed.
科研通智能强力驱动
Strongly Powered by AbleSci AI