This study analyzes a large dataset of perovskite solar cell (PSC) ageing data to investigate the relationship between initial efficiency and long-term stability. A statistical analysis reveals a correlation between maximum power conversion efficiency (PCE) and relative PCE loss after 150 hours, with more efficient cells exhibiting higher stability. Unsupervised machine learning (self-organizing map) clustered degradation curves into distinct shapes, revealing a correlation between curve shape and maximum PCE. The findings suggest that higher efficiency is statistically linked to improved stability, potentially due to energy conservation or reduced defect density.

Noor Titan Putri Hartono, Hans Köbler, Paolo Graniero, Mark Khenkin, Rutger Schlatmann, Carolin Ulbrich, Antonio Abate