A research team from East China University of Science and Technology has developed an AI-assisted materials-genome approach that enables rapid design of high-performance thermosetting polyimides, addressing long-standing challenges in balancing competing mechanical properties. Their study published in the Chinese Journal of Polymer Science introduces a machine-learning model capable of predicting three key mechanical parameters—Young's modulus, tensile strength, and elongation at break—across thousands of candidate structures.
This breakthrough is significant because polyimide films are essential components in aerospace, flexible electronics, and micro-display technologies, where thermal stability and insulation properties are critical. However, mechanical optimization has remained elusive as improving one property typically compromises others. Traditional trial-and-error synthesis methods are slow, costly, and limited in exploring complex molecular spaces, creating bottlenecks in materials development for advanced technologies.
The research team constructed Gaussian process regression models trained on over 120 experimental datasets of polyimide films. By treating polymer structural fragments—dianhydride, diamine, and end-capping units—as molecular "genes," they defined a vast chemical space of 1,720 phenylethynyl-terminated polyimides. The models achieved high predictive accuracy for all three mechanical metrics and successfully identified a new formulation, PPI-TB, whose performance surpassed well-known benchmark polyimides.
Molecular dynamics simulations validated the screening process, showing that PPI-TB exhibited superior modulus, toughness, and strength indicators compared with established systems. Subsequent experiments on representative polyimides confirmed strong consistency between predicted and measured data. The detailed findings are available in the study published at https://doi.org/10.1007/s10118-025-3403-x.
Further analysis revealed key design principles: conjugated aromatic structures enhance stiffness, heteroatoms and heterocycles strengthen molecular interactions, and flexible units improve elongation. These insights demonstrate how integrating AI predictions with molecular interpretation can uncover structure-property relationships and accelerate polymer innovation. The approach successfully identified formulation PPI-TB with simultaneously high Young's modulus, tensile strength, and elongation at break—properties that typically compete against each other in material design.
The implications of this research extend across multiple industries. For aerospace applications, the ability to rapidly develop lightweight, durable polymers with balanced mechanical properties could lead to more efficient aircraft components and space vehicle materials. In electronics manufacturing, the methodology enables faster development of flexible circuit substrates and micro-display technologies that require both thermal stability and mechanical resilience.
By replacing years of experimental iteration with predictive modeling and virtual screening, this AI-driven approach drastically reduces development costs and timeframes. The materials-genome strategy provides a universal, scalable framework that could be adapted for other high-performance polymer classes, potentially accelerating innovation across multiple materials science domains. This represents a fundamental shift in how advanced materials are discovered and optimized, moving from labor-intensive experimental approaches to data-driven computational methods.
