Materials Informatics Applied to Ultraviolet Light Emitting Diodes
Materials Data Science (MDS) is an interdisciplinary field that aims to use techniques traditionally associated with Computer Science for Materials Science. We define this area by combining Materials Informatics with Computational Materials, along with the field of Artificial Intelligence. The goal is to bring together computational methods that enable the simulation, generation, description, and prediction of new materials and properties, enhancing our understanding of these physical systems and expediting the technology production process based on materials. Light Emitting Diodes (LEDs), on the other hand, are electronic devices based on semiconductor materials. Particularly, to achieve emissions in the ultraviolet (UV) spectrum in these devices, materials with ultra-wide bandgaps (UWBG) are employed, meaning they have an energy gap greater than 3.2 eV. Although UV-LEDs already exist, their physical efficiencies and costs remain high, especially when it comes to emissions in the UVC range (wavelength between 280nm and 100nm). The objective of this work, therefore, was the development and application of MDS techniques for the discovery and synthesis of promising UWBG materials for UVC LED construction. To achieve this, materials simulation techniques, such as Density Functional Theory (DFT) with a high-throughput (HT) approach, data mining techniques like database screening, and artificial intelligence techniques, particularly those known as machine learning (ML), were employed. Preliminary results indicated a mineral with an ultra-wide bandgap, Jakobssonite, with the chemical formula CaAlF5, a perovskite in which the octahedra AlF6–2 are corner connected, forming one-dimensional chains (1D). By applying data augmentation techniques, we expanded the family of this material through chemical substitutions. These new materials were studied using HT-DFT and heuristic models, such as ML surrogate models.