Machine-learning–driven exploration of HTL effects in Mg3SbBr3 perovskite solar cells for ultra-high efficiency

Abstract

Creating efficient and durable lead-free perovskite solar cells continues to be a major obstacle in advancing environmentally sustainable photovoltaic solutions. In this study, an extensive computational analysis of Mg₃SbBr₃-based perovskite solar cells is performed by combining SCAPS-1D device modeling with machine learning approaches to uncover key efficiency-limiting factors and practical routes for performance enhancement. The impact of various non-ideal factors, such as light absorption losses, defect-driven recombination in the bulk and at interfaces, electrical resistance losses, and temperature-related effects, is examined in a structured and detailed manner. Device configurations incorporating different hole transport materials, specifically CNTS, PbS–TBAI, and spiro-OMeTAD, are systematically compared and assessed. When optimized within physically reasonable limits, the CNTS-based device structure (FTO/TiO₂/Mg₃SbBr₃/CNTS/Au) delivers the best overall performance, showing higher open-circuit voltage, improved fill factor, and better long-term stability as a result of superior energy band matching and reduced recombination at the interfaces. Machine learning models developed using datasets generated from SCAPS simulations show strong predictive reliability and offer clear, interpretable understanding of how defect concentrations, energy band matching, and charge transport properties influence device performance. Notably, all simulated solar cell metrics are interpreted in the context of well-known thermodynamic constraints and the Shockley–Queisser limit, with the stated efficiency values reflecting theoretical maxima under ideal conditions rather than results that can be directly achieved in experiments. This work identifies Mg₃SbBr₃ as a strong candidate for lead-free light absorption and emphasizes that integrating physics-driven simulations with transparent machine learning methods offers a practical and effective approach for steering the informed development of future eco-friendly perovskite solar cells.