A recent study has revealed a groundbreaking approach to enhancing the design of polyimide films, which are crucial in aerospace, flexible electronics, and micro-display technologies due to their thermal stability and insulation properties. Researchers from East China University of Science and Technology have developed an AI-assisted materials-genome approach that significantly accelerates the optimization of these materials, addressing long-standing challenges in balancing stiffness, strength, and toughness.
Traditionally, materials scientists have faced difficulties in achieving high mechanical performance in thermosetting polyimide films, as enhancing one property often diminishes another. The conventional trial-and-error method for synthesizing these films has proven to be slow and costly, limiting the exploration of complex molecular structures. However, the introduction of materials-genome strategies, which integrate computational methods, experimental data, and artificial intelligence (AI), provides a promising alternative.
Published online on September 2, 2025, in the Chinese Journal of Polymer Science, the researchers” study presents a machine-learning model capable of predicting three essential mechanical characteristics: Young”s modulus, tensile strength, and elongation at break. The study screened over 1,700 phenylethynyl-terminated polyimide candidates, identifying a new formulation named PPI-TB that outperformed established benchmark polyimides.
The research team employed Gaussian process regression models, trained on more than 120 experimental datasets of polyimide films. They regarded each polymer”s structural fragments—such as dianhydride, diamine, and end-capping units—as “genes,” thus defining a comprehensive chemical space. The models demonstrated impressive predictive accuracy for all three mechanical metrics, achieving R² values between 0.70 and 0.74. This allowed the researchers to evaluate each candidate”s mechanical performance comprehensively.
Molecular dynamics simulations confirmed the model”s predictions, revealing that PPI-TB exhibited a modulus of 3.48 GPa along with superior toughness and strength compared to the established systems PETI-1 and O-O-3. Subsequent laboratory experiments validated the consistency between the predicted and actual measurements.
Moreover, further analysis of the “genes” and feature importance highlighted key design principles. The findings indicate that conjugated aromatic structures enhance stiffness, while the inclusion of heteroatoms and heterocycles strengthens molecular interactions. Additionally, flexible silicon or sulfur-containing units improve elongation at break. These insights underscore the potential of integrating AI with molecular interpretation to uncover critical structure–property relationships and expedite polymer innovation.
“By translating polymer fragments into genetic-like descriptors, we can treat molecular design like decoding a genome,” stated Prof. Li-Quan Wang, a corresponding author of the study. “Machine learning not only predicts performance but also reveals which chemical “genes” drive it. This synergy between data science and chemistry allows us to explore material possibilities that would take decades by conventional means.” The success of PPI-TB exemplifies how AI can redefine the discovery process for next-generation high-temperature polymers.
The AI-driven materials-genome strategy provides a scalable framework for designing polymers that achieve targeted combinations of stiffness, strength, and flexibility—all crucial traits for applications in microelectronics, aerospace composites, and flexible circuit substrates. By replacing years of experimental iterations with predictive modeling and virtual screening, this innovative approach drastically cuts both costs and development time. Furthermore, this methodology holds potential for adaptation in other high-performance polymer classes, paving the way for the creation of lightweight, durable, and thermally stable materials essential for future technological advancements.
