Recent advances in data-driven and artificial intelligence-integrated perovskite solar cells: From design to self-driving laboratories

Abstract

Perovskite solar cells (PSCs) have rapidly advanced owing to their excellent optoelectronic properties such as high absorption, long diffusion length, and high carrier mobility, achieving power conversion efficiencies of up to 27%. The ABX₃ crystal structure of perovskites and their various possible material combinations provide broad compositional and dimensional tunability, enabling tailored bandgaps, controlled stability, and targeted optoelectronic features. However, the efficiency of conventional trial-and-error approaches in discovering new materials is limited by the interplay between the compositional, material, and processing variables, highlighting the need for reproducible synthetic protocols and reliable datasets to support the high-throughput exploration of material combinations. Artificial intelligence (AI) technologies, including machine learning and large language models, leverage such datasets to provide predictive and generative capabilities for performance forecasting, compositional and process optimization, inverse design of novel materials, and literature knowledge extraction. Furthermore, the combination of an automated protocol setup, fabrication, high-throughput characterization using AI, and large device-level datasets has paved the way for building autonomous research platforms. Specifically, automation and robotics are integrated with in situ metrology and algorithmic guidance to reduce the build–measure–learn cycle from weeks to hours, thereby accelerating discovery and stability assessment. This review focuses on three central pillars of data-driven and AI research: data platforms, AI methodologies, and self-driving laboratories, which could collectively reshape PSC research into a systematic, autonomous, and scalable framework. By reviewing advances across these domains, we demonstrate how data-driven strategies can transform PSC development from intuition-based exploration to accelerated and reliable innovation, paving the way for practical deployment and commercialization.