High-refractive-index materials screening from machine learning and ab initio methods

In this study we analyze the dielectric properties of a recently published dataset to identify high-refractiveindex and high-band-gap materials that are crucial for modern optoelectronic applications. We employ advanced crystal graph convolutional neural networks and density functional perturbation theory calculations to accelerate the discovery of such materials. Our analysis confirms the traditional inverse relationship between band gap and dielectric constant, which persists even in this large dataset. However, our study reveals several promising materials that possess competitive properties compared to current industry standards. Our findings provide valuable insights into the field of dielectric materials and demonstrate the potential of advanced machine learning and computational techniques for accelerating materials discovery.