Thickness scaling in hafnium ferroelectrics: insights into oxygen vacancy clustering and its impact on ferroelectricity and reliability

Abstract Hafnium-based ferroelectrics have emerged as promising candidates for next-generation non-volatile memory technologies owing to their excellent complementary metal–oxide–semiconductor (CMOS) compatibility and exceptional thickness scalability. However, the defect mechanisms governing the evolution of ferroelectricity and reliability during thickness scaling remain poorly understood. In particular, the role of oxygen-vacancy-related defects has primarily been discussed from the perspective of defect concentration, while their spatial organization has received far less attention. Here, we systematically investigate the impact of thickness scaling on the ferroelectric properties and reliability of Hf 0.5 Zr 0.5 O 2 (HZO) capacitors. Although thinner HZO films exhibit reduced remanent polarization due to enhanced dead-layer effects, increased depolarization fields, and stabilization of nonpolar phases, they surprisingly demonstrate superior breakdown tolerance and endurance performance. Notably, X-ray photoelectron spectroscopy (XPS) analysis revealed a higher oxygen vacancy concentration in thinner films, indicating that defect concentration alone cannot explain the observed reliability evolution. Time-dependent dielectric breakdown (TDDB) analysis using the time-dependent clustering model, together with density functional theory (DFT) calculations and atomic force microscopy (AFM) characterization, suggests that thickness scaling modifies the spatial organization of oxygen-vacancy-related defects. Grain refinement in ultrathin films increases the grain-boundary density, thereby suppressing severe defect clustering and promoting a more dispersed defect organization, which delays breakdown and improves long-term reliability. These findings identify defect organization as a critical factor governing the thickness-scaling behavior of hafnium ferroelectrics and provide new insights for the design of highly scaled and reliable ferroelectric memory devices.