Design of alloys of halide perovskites: an approach by combination of machine learning and density functional theory

Abstract

Halide perovskites belongs to the semiconductor materials class with a high potential for applications in photovoltaics devices, especially in solar cells. The low cost of production, easy fabrication and the high performance makes this materials the possible candidates to replace the actual technologic based on silicon. However, high degradation issue in moisture environment, high temperature and other, at the same time the composition that contains Pb are main drawbacks to use halide perovskites in large scale industry. Alloys of halide perovskites have been an alternative to reduce these main issues, though the large number of possible systems that could explored surpasses the scale of tens of bilious compounds. From the experimental point of view this exploration is impactable, because it requires a large trial-and-error methods, manpower and large time consuming. In this project, we propose a way to explore the large numbers of alloys of halide perovskites through a combination of machine learning and density functional theory calculations. With this method it is possible to predict systems that are stables and arrange according to the target application, for example, solar cells or light emission diodes. For this study, we will use a pre selection techniques, such as the Goldschmidt tolerance factor and the bond valence method. In these system, the machine learning will be used to select the top 10% better candidates where the electronic structure will be explored through the density functional theory calculations. We expect to provide a large number of alloys of halide perovskites that are stable, with high performance in target applications and environment friendly. This will give a pathway to experimentalists to explore the halide perovskites world. (AU)