Machine learning-based screening of two-dimensional perovskite organic spacers

Abstract

Perovskite solar cells (PSCs) are prominent devices that have attracted much attention in future photovoltaic technology. It has been investigated that adding organic spacers to PSCs is an effective way to enhance their power conversion efficiency (PCE), but how to quickly screen out suitable organic spacers is still a great challenge. In this paper, we investigated the link between the feature descriptors of ammonium salts that can be used as organic spacers and PCE using a machine learning (ML) approach to screen the materials. We used a dataset of 27 common ammonium iodides that was built for a machine learning model to predict the ratio of improved PCE to its highest certified efficiency in the same year. Among other things, molecular weight and the number of hydrogen bond donors were calculated as important characteristics for selection for use in organic spacers. The 112 organic spacers collected from the database were screened using the best machine learning model, and the best ammonium iodide salt was predicted to be PEAI. Our work establishes criteria for efficient screening of organic spacers for use in two-dimensional perovskite solar cells.