Revealing degradation mechanisms in YSZ ceramics through machine learning-guided aging and multiscale characterization

Abstract

The long-term performance of yttria-stabilized zirconia (YSZ) based energy and biomedical devices is compromised by low-temperature degradation (LTD). This study presents a novel integration of machine learning-guided hydrothermal aging with multiscale characterization to resolve a two-stage degradation mechanism in 3 mol% YSZ. Stage 1 (0 to 30 hrs) features initial surface relief building, which transitions to partial refinement and relief distribution in stage 2 (30 to 60 hrs), alongside a rise in monoclinic phase content. The evolving microstructure increases triple-junction grain boundary density, and these junctions act as degradation hotspots, where vacancy exchange and water access accelerate the transformation. These findings highlight grain boundary chemistry, rather than grain size alone, as a key LTD driver, suggesting boundary engineering as a strategy to enhance YSZ stability for energy, biomedical, and thermal applications.