Toward sustainable efficient photovoltaics: a first-principles, device engineering, and machine learning study of 30.38% efficient Rb2AgBiI6 solar cells
摘要
Rb2AgBiI6 is an emerging lead-free double halide perovskite with promising optoelectronic characteristics, yet its photovoltaic potential remains largely unexplored. In this work, we combine first-principles, SCAPS-1D device modelling, and machine-learning (ML) analysis to evaluate the feasibility of Rb2AgBiI6 as a high-performance and environmentally benign solar absorber. DFT calculations confirm the structural stability of the cubic Fm3̅m phase and reveal an indirect bandgap of 1.49 eV, favourable carrier effective masses, and strong optical absorption in the visible region. Device simulations show that appropriate band alignment with TiO2 (ETL) and Cu2O (HTL) yields efficient carrier extraction and suppressed recombination. The optimized device structure, FTO/TiO2/Rb2AgBiI6/Cu2O/Au, achieves a power-conversion efficiency (PCE) of 30.38%, with Voc = 1.27 V, Jsc = 28.73 mA/cm2, and FF = 82.97%. Sensitivity analysis of absorber doping, thickness, interface defect density, and parasitic resistances identifies key performance-limiting mechanisms. Additionally, a Random Forest ML model trained on simulation data achieves up to 97% prediction accuracy, enabling rapid estimation of photovoltaic parameters. Overall, our findings demonstrate that Rb2AgBiI6 is a highly promising, stable, and lead-free absorber for next-generation perovskite photovoltaics.