Authers
Yoshihiro Hayashi, Junichiro Shiomi, Junko Morikawa, Ryo Yoshida
Abstract
The spread of data-driven materials research has increased the need for systematically designed materials property databases. However, the development of polymer databases has lagged far behind other material systems. We present RadonPy, an open-source library that can automate the complete process of all-atom classical molecular dynamics (MD) simulations applicable to a wide variety of polymeric materials. Herein, 15 different properties were calculated for more than 1000 amorphous polymers. The MD-calculated properties were systematically compared with experimental data to validate the calculation conditions; the bias and variance in the MD-calculated properties were successfully calibrated by a machine learning technique. During the high-throughput data production, we identified eight amorphous polymers with extremely high thermal conductivity (>0.4 W ∙ m–1 ∙ K–1) and their underlying mechanisms. Similar to the advancement of materials informatics since the advent of computational property databases for inorganic crystals, database construction using RadonPy will promote the development of polymer informatics.
npj Computational Materials: https://www.nature.com/articles/s41524-022-00906-4