An MGI approach to new dielectric polymers for wound film capacitors

Energy-dense, easily manufactured, affordable capacitors are essential components for climate-sensitive power distribution. They must also be reliable and exhibit graceful failure, which means that if they are damaged the response is a measured reduction in performance rather than a sudden extinction. Metallized polymer-wound film capacitors are a natural choice. The energy stored in a film capacitor is proportional to the product of the dielectric constant and the square of the dielectric breakdown strength. The state-of-the-art polymer is biaxially oriented polypropylene, BOPP. It has a very high breakdown strength due to oriented crystallites, but also has a low dielectric constant due to the non-polar nature of the alkyl chains. The Office of Naval Research invested in the pursuit of other polymers that could store more energy. However, the chemical space of organic polymers is huge, raising the question, “Where does one start?”

In a recent Multidisciplinary University Research Initiative project led by the University of Connecticut, an MGI approach was employed, emphasizing computation with strong feedback from experimentation (polymer synthesis and chemical, morphological, and electronic characterization). Researchers used high throughput density functional theory to probe chemical space and identify materials with high dielectric constant and high dielectric breakdown. With computations on hundreds of materials, they noted an expected trend as nearly all materials fell on a line that showed as the dielectric constant increased, the breakdown strength decreased. More interestingly, there were outliers.

Key to the success of the research effort was the balanced, iterative input between computation and experimentation. Synthesis pathways for some compounds did not exist. Feedback from synthesis experts identified closely related materials that could be made, and new computations centered on these were carried out to identify the most promising materials. These materials were synthesized and characterized, which in turn led to more cycles of computation, synthesis, and characterization, and eventually to promising families of new dielectric materials.

For a film to be useful for capacitors, it needs to be processable to very thin thicknesses (4–20 micrometers) without pinholes, flaws, and with less than 5% thickness variation. Reel-to-reel processing used to manufacture large volumes of films provides many parameters that can be monitored in-line and manipulated. With quantitative understanding of the parameters that can be controlled during fabrication, the computation/synthesis/characterization process was applied to the most promising new materials to find optimal properties that, in turn, provided predictable processing windows.

Recently, with the benefit of the data generated to date, machine learning approaches to materials design are being incorporated into the discovery-to-manufacturing continuum for advanced dielectric films and advanced capacitors.