Madhubanti Mukherjee1,Shivank Shukla1,Rishi Gurnani1,Joseph Kern1,Harikrishna Sahu1,Rampi Ramprasad1
Georgia Institute of Technology1
Madhubanti Mukherjee1,Shivank Shukla1,Rishi Gurnani1,Joseph Kern1,Harikrishna Sahu1,Rampi Ramprasad1
Georgia Institute of Technology1
In the domain of electrical and electronic applications, the demand for flexible polymer dielectrics capable of withstanding extreme temperatures and electric fields is pressing. The dielectric breakdown strength, signifying the maximum electric field a polymer can endure while retaining excellent insulating qualities (i.e., sufficiently high bandgap (E<sub>g</sub>)) is a pivotal parameter. Together with the dielectric constant of the polymer, it dictates the upper limit of electrostatic energy storage in a capacitor. Another key parameter for high-temperature energy storage is the high glass transition temperature (T<sub>g</sub>), which ensures thermal and electromechanical stability, thereby preserving capacitive performance. Simultaneously obtaining high T<sub>g</sub>, a high dielectric constant, and a high E<sub>g</sub> for achieving enhanced energy density proves to be a formidable challenge due to the observed negative correlation between T<sub>g</sub> and E<sub>g</sub>, further complicated by potentially increased dielectric loss at higher T<sub>g</sub>. Identifying such suitable candidates in the vast chemical landscape of polymers, entangled with the complexities of dielectric breakdown mechanisms is daunting. Artificial intelligence-driven screening methods have emerged as proficient tools for sifting through expanding polymer libraries, streamlining the selection process for experimental exploration [1]. The application of screening criteria based on readily accessible proxy properties has expedited the search process in contrast to the laborious manual extraction of chemo-structural attributes. These criteria include a high bandgap to ensure an insulating phase, a high glass transition temperature to ensure thermal stability, and a substantial dielectric constant that enhances energy density [2, 3]. Employing these criteria in conjunction with machine learning models, we conducted an extensive screening initiative encompassing over 14,000 previously synthesized polymers and an additional 6 million hypothetical polymers crafted from diverse reaction templates. This systematic approach has unraveled a multitude of promising candidates, boasting attributes such as a glass transition temperature reaching 400°C, a bandgap of 4 eV, and a dielectric constant within the range of 3-5. These characteristics hold the promise of remarkably high energy density. Furthermore, our study leverages density functional theory and molecular dynamics simulations to further validate the bandgap and glass transition temperature for a significant portion of these polymers. This study also provides valuable insights into the structural elements essential for designing robust, high-temperature polymer dielectrics with exceptional energy storage capabilities.<br/><br/>[1] Chem. Mater. 35, 4, 1560–1567 (2023)<br/>[2] ACS Appl. Mater. & Interfaces <i>12,</i> 33, 37182-37187 (2020)<br/>[3] Matter, 5(9), 2615-2623 (2022)