Utilizing Machine Learning and Diode Physics to Optimize Photovoltaic Performance in Sequentially Processed Perovskite Solar Cells

Organic-inorganic lead halide perovskite photovoltaics are well-known for their exceptional solution processability. However, achieving uniformly crystalline perovskite films often requires complex deposition methods. To solve this challenge of the perovskite deposition method, several studies have been performed to utilize deposition parameters such as concentrations of the annealing temperatures, precursor solution, and spin-coating speeds. Unfortunately, most of these studies have been conducted on trial-and-error approaches, which are resource-intensive and time-consuming. Nowdays, machine learning techniques have considered as powerful tools for utilizing complex experimental parameters and predicting the device performance, making more efficient analysis of parameter spaces.<br/>In our study, we combined Shockley diode-based numerical analysis with machine learning techniques to analyze the photovoltaic characteristics of the device and utilize their photovoltaic performance by considering experimental variables. The application of the Shockley diode equation allowed us to extract photovoltaic parameters and predict power conversion efficiencies, contributing to our understanding of charge recombination and device physics. Using machine learning techniques, we trained a machine learning model using current-voltage curves that are sensitive to changes in manufacturing conditions. This enabled us to identify the optimal settings for improved device performance.<br/>Our harmonized approach not only reveals the relationship between experimental conditions and device performance, but also simplifies the optimization process, reducing the need for extensive trial-and-error experimentation. This methodology shows great promise in advancing the development and fine-tuning for next-generation perovskite solar cells.