Committee Members:     
Prof. Rampi Ramprasad, Advisor, MSE
Prof. Seung Soon Jang, Co-Advisor, MSE
Prof. Karl I. Jacob, MSE
Prof. Ryan P. Lively, ChBE
Prof. Chao Zhang, CSE

Advances in artificial intelligence for accelerated discovery of energy storage polymers

Abstract:

Electrostatic capacitors are critical energy storage components in advanced electrical systems in the defense, aerospace, energy, and transportation sectors. Compared with other, more vigorously discussed, energy storage devices—such as batteries, fuel cells and supercapacitors—electrostatic capacitors offer unparalleled power density (107 W kg−1). This attribute renders electrostatic capacitors particularly advantageous for deployment in diverse fields, including wind pitch control (with maximum temperatures around 125 °C), hybrid, all-electric and rail vehicles (∼ 150 °C), pulsed power systems (∼ 180 °C), aircraft and aircraft launchers (∼ 300 °C), and space exploration (∼ 480 °C). A persistent hurdle facing electrostatic capacitor technology involves improving their energy density Ue, especially at elevated temperatures. Overcoming this barrier would allow for considerable reductions in both mass and volume requirements. Exploring previously unknown polymers, and optimizing their chemical structure for use as a dielectric, is one route to the discovery of capacitors that address such inefficiencies.

However, locating these polymers within the vast chemical expanse is akin to searching for a needle in a haystack. Artificial intelligence (AI) may be used to accelerate the process. To date, state-of-the-art AI algorithms have primarily been developed and optimized for game play, computer vision, natural language processing, and social networking, rather than for physics, chemistry, and materials science. The first part of my dissertation fills this gap by developing AI tools in two areas: chemical structure generation and structure-property modeling. These include rule-based systems based purely on known chemistry, as well as data-driven machine learning algorithms such as graph neural networks.

The second part of the dissertation aims the developed tools at the discovery of polymer-based capacitor dielectrics. Hundreds of thousands of previously unknown polymers were canvassed. A handful—belonging to the polynorbornene, polyimide, and other families—were explored via molecular modeling (MD, DFT), chemical synthesis, and physical characterization. Among these, one particular polymer named PONB-2Me5Cl displays remarkable Ue values at room temperature and above. At 200 °C, the polymer has an extraordinary Ue of 8.3 J/cc—over an order of magnitude higher than that of any commercial alternative. This discovery widens the scope of potential applications for electrostatic capacitors at high temperatures, such as wind pitch control, hybrid vehicles, and pulsed power systems and also reveals new structure-property relationships.

Despite the emphasis on polymers for electrostatic energy storage, the AI tools developed in this dissertation generalize to a broad spectrum of materials discovery challenges. I have used the tools for the discovery of non-polymeric (MOFs) and polymeric materials for other applications as well (gas storage and separation). Overall, this research demonstrates the impact of AI on chemical structure generation and structure-property modeling, highlighting the potential for materials design advancement.