Data-driven insights to tackle photo-induced phase segregation for mixed-halide perovskite solar cells

Perovskite-based tandem solar cells have emerged as one of the most promising photovoltaic (PV) technologies due to the continuously improving power conversion efficiency (PCE). However, in tandem cells, the wide-bandgap perovskites are prone to phase segregation under illumination, which results in significantly reduced photostability. To explore the mechanism of phase segregation in mixed-halides perovskites and develop photostable wide-bandgap devices, we employ a data-driven approach to systematically analyze the impact of perovskite compositions and charge transport layers on photo-induced phase segregation. In this work, we build a Gaussian Process regression (GPR) model to identify the relationship between the A-site (FA/MA/Cs) ratios and changes in photoluminescence (PL) peak positions after 500-hour light soaking. Based on the analysis of model, we select the most stable perovskite composition (Cs_0.24MA_0.02FA_0.74Pb(I_0.76Br_0.24)₃) (1.69 eV) for device fabrication. With this stable perovskite composition, we further screen twelve different charge transport layers. Our results indicate that the mixed self-assembled monolayers (SAMs) as hole transport layers exhibited superior stability compared to other layers. Finally, we achieved a high-performance and high-stable photovoltaic device (without other molecular additives), with a PCE of 21.18 % and an open-circuit voltage (V_OC) of 1.255 V. The device maintains 85 % of its original efficiency even after 1000 h of continuous exposure to light under open-circuit conditions.