Development of low-density and high-modulus thermosetting resins using machine learning

Abstract

We developed a thermosetting resin with low density and high modulus using machine learning to meet the demand for lightweight materials in the mobility field. Conventionally, reducing resin density while increasing modulus has been a trade-off, making molecular design challenging. In this study, machine learning was applied to predict density, modulus, and specific modulus, enabling a workflow to propose promising molecular structures from a virtual molecule library. As a result, it was suggested that alicyclic and aniline moieties are effective for achieving high specific modulus. Based on these findings, we designed and synthesized a novel curing agent, ADF, which exhibited low density, high modulus, and a 23% improvement in specific modulus compared to conventional systems. Finally, we succeeded in developing a resin with lightweight, stiff, and easy to handle, making it suitable for composite materials in mobility applications. Furthermore, this information-science-driven development process has demonstrated the possibility of further improving the efficiency of materials development.