Research on high electron-affinity molecule-grafted modified polyetherimide energy storage dielectrics based on machine learning
Liu, Jiaqi (2026)
Kandidaatintyö
2026
School of Energy Systems, Energiatekniikka
Kaikki oikeudet pidätetään.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi-fe2026051949592
https://urn.fi/URN:NBN:fi-fe2026051949592
Tiivistelmä
Polymer dielectric films are key materials for film capacitors used in advanced power systems, such as flexible high-voltage direct current transmission, electric vehicle power electronics, and pulsed-power devices. Nevertheless, traditional biaxially oriented polypropylene have sharply increased energy loss above 105 °C, and suffer limited energy density, which makes it difficult to achieve the demands of high-temperature, compact, and reliable energy-storage applications. Polyetherimide (PEI) has good thermal stability, electrical insulation properties, yet it is still affected by higher dielectric and conduction losses at temperatures above 150 °C.
To solve this issue, this thesis proposes a molecular design strategy of combining machine-learning-assisted screening and high-electron-affinity molecular grafting modification. To select promising molecules with high electron affinity, favorable grafting reactivity, and a deep-trap profile were sought in a molecular library by using a machine-learning workflow to find a candidate molecule. Following this, the selected molecule was incorporated into a semi-empirical simulation framework to determine the potential effect of the selected molecule on the dielectric and energy-storage performance of grafted PEI at a high temperature. Relative permittivity, dielectric loss, dc conductivity, leakage current, breakdown strength, charge-discharge efficiency, and recoverable energy density were taken into account in the simulation.
The results suggest that the chosen grafting molecule, 2-trifluoromethyl-3, 5-dicyanoaniline, can be used to improve the high-temperature dielectric performance of PEI through the introduction of electron-trapping sites and the inhibition of thermally activated carrier transport. Within the present simulation model, the predicted optimum grafting content is around 0.15 wt%, at which the recoverable energy density reaches 5.493 J cm-3 and the charge-discharge efficiency reaches 90.8%, both significantly higher than those of pristine PEI. Such results imply that machine-learning-directed molecular screening paired with the grafting-based trap engineering is a potential pathway to the development of high-temperature polymer dielectrics to use in capacitor applications.
To solve this issue, this thesis proposes a molecular design strategy of combining machine-learning-assisted screening and high-electron-affinity molecular grafting modification. To select promising molecules with high electron affinity, favorable grafting reactivity, and a deep-trap profile were sought in a molecular library by using a machine-learning workflow to find a candidate molecule. Following this, the selected molecule was incorporated into a semi-empirical simulation framework to determine the potential effect of the selected molecule on the dielectric and energy-storage performance of grafted PEI at a high temperature. Relative permittivity, dielectric loss, dc conductivity, leakage current, breakdown strength, charge-discharge efficiency, and recoverable energy density were taken into account in the simulation.
The results suggest that the chosen grafting molecule, 2-trifluoromethyl-3, 5-dicyanoaniline, can be used to improve the high-temperature dielectric performance of PEI through the introduction of electron-trapping sites and the inhibition of thermally activated carrier transport. Within the present simulation model, the predicted optimum grafting content is around 0.15 wt%, at which the recoverable energy density reaches 5.493 J cm-3 and the charge-discharge efficiency reaches 90.8%, both significantly higher than those of pristine PEI. Such results imply that machine-learning-directed molecular screening paired with the grafting-based trap engineering is a potential pathway to the development of high-temperature polymer dielectrics to use in capacitor applications.