Luis 🦙
Luis 🦙¶
Title: Determining new dielectric materials to replace SiO2 using machine learning
Description: Dielectric materials are a subset of materials that are typically found in microelectronics due to their insulator-like properties. Their properties are partly due to the fact that dielectrics have wide band gaps, typically above 4 eV. Dielectrics are crucial in reducing the size of transistors and following the trend on Moore’s Law. Finding alternative dielectrics to SiO2 is crucial since the current plan for transistor reduction results in quantum tunneling. In order to more efficiently find new materials that don’t yet have dielectric constants in the Materials Project database, we turned to machine learning to help us predict dielectrics that can act as a substitute to SiO2. We ran linear regression and random forest models against known dielectrics and were able to produce compounds that had been proven to be dielectrics and others that haven’t been tested experimentally.