AI-aided De Novo Molecular Design of Phosphamide for Fire-Retardant Polylactide with Enhanced Integrated Properties

• A descriptive AI model can describe fire performances of FRs for PLA; • This model can also predict the impact of FRs on mechanical and thermal properties; • A GAI model can generate new FR molecules with better performances; • A high-efficiency phosphamide FR was synthesized for PLA. In polymer science, particularly in flame retardant (FR) development, Artificial Intelligence (AI) has primarily been used to predict flame-retardant performance metrics, such as limiting oxygen index (LOI), vertical burning (UL-94) rating, and peak heat release rate (PHRR), while its potential for discovering novel FR molecules remains largely unexplored. Meanwhile, predicting the effects of FRs on mechanical properties is also untapped. To address this, a generative AI-driven de novo molecular design strategy, GAI4FR, is introduced aimed at generating novel FR molecules for polylactic acid (PLA) with improved performance. A machine learning (ML) model is also trained to predict the fire-retardant performance, tensile strength ( σ t), and glass transition temperature ( T g ) of the AI-generated FRs, ultimately leading to the discovery of a high-efficiency molecule, EDP. EDP was synthesized, and its AI-predicted performance metrics were successfully confirmed through experimental validation. This first-of-its-kind GAI4FR framework enables the discovery of FRs and establishes a foundation for the development of other advanced functional materials.