Comprehensive Variation Model of Hafnium-Based Ferroelectric Capacitors for High-Sigma Memory Window Simulation

Abstract

The polycrystalline nature of emerging ferroelectric materials introduces inherent variability in switching behavior, which poses a fundamental challenge for future high-density memory arrays. Predicting and mitigating this variability, particularly the worst-case “tail bits,” is critical for performance yet difficult with conventional approaches. Here, we present FEVAR, a synergistic framework that integrates multiscale material characterization with a physics-based modeling pipeline to accurately predict the performance distribution of hafnium-based ferroelectric random-access memory (FeRAM) arrays. Leveraging this approach, we further delineate how material properties govern array-level performance. This work not only provides guidance for optimizing FeRAM but also establishes a generalizable methodology for linking material structure to device performance in the face of intrinsic variability.