Proceedings of International Conference on Perovskite Thin Film Photovoltaics and Perovskite Photonics and Optoelectronics (NIPHO20)
Publication date: 25th November 2019
Hybrid halide perovskites are one of the new era solar cells and photocatalysts those are expected to solve global energy problems. Lead halogen compounds with Pb2+ at the B site, such as methylammonium lead iodines (MAPbI3, MA = CH3NH3) and formamidine lead iodines (FAPbI3, FA = HC(NH2)2), are easily produced at low cost, however, have disadvantages such as chemical instability and toxicity.
In this study, we aimed to search for new materials for lead-free perovskite solar cells and photocatalysts exploring the properties and elements of perovskite materials using statistics and machine learning. First-principles calculations on the perovskite ABX3 and the double perovskite material A2BB'X6, and 11,025 high-throughput screening data obtained by Nakajima and Sawada [1] were used as a data set. These depend on the number of elements contained in the B site and the combination of those groups, covering all 49 types of atoms in Groups 2 to 15. We used Python and Excel to summarize the data and build a multiple regression analysis prediction model using (a) crystal structure, (b) total energy, (c) value of direct transition gap, (d) value of indirect transition gap, (e) position of conduction band minimum and valence band maximum, (f) effective electron and hole masses, (g) existence of toxic elements, and (h) stability, as target properties of machine learning.
This research was supported by MEXT as "Priority Issue on Post-K computer” (Development of new fundamental technologies for high-efficiency energy creation, conversion/storage and use) and by JSPS KAKENHI in Scientific Research on Innovative Areas "Innovations for Light-Energy Conversion (I4LEC)".