Exception search in databases for polymers with practically contradictory properties of heat resistance and transparency

Abstract

Although the common data-driven studies for material development use property values present in the database, the effectiveness of analysing and exploring the areas without property values in the database has hardly been clarified. The necessity to analyse such areas is evident from the fact that only 44% of polymer chemical structure entries with property values are used for studies, while the remaining 56% are not in the analysis of glass transition temperatures using the PoLyInfo database. In this study, a method for discovering polymer materials with undiscovered properties was demonstrated and experimentally verified. The study utilised a comprehensive database of polymer materials and properties with an exceptional search procedure that defines unexplored areas in the database. In addition, a filter based on the physicochemical mechanism of polymers was used to remove the common molecular structures to reveal polymers with contradictory properties. Similarly, a mechanism-based filter was also used to narrow down the candidates efficiently. We investigated a practically challenging heat-resistant transparent polymer material with this procedure and experimentally verified the screening candidate obtained by molecular dynamic simulations and machine-learning predictions. Consequently, potential polymer materials were discovered with thermal degradation temperature, T_deg, higher than 300 °C without glass transition in the entire temperature range, and high transparency (more than 80% transmittance) in the visible and UV regions.