Dynamic Landau–Khalatnikov Modeling of Ferroelectric Hafnium–Zirconium Oxides Across Volatile and Non‐Volatile Switching

Abstract

The design of energy‐efficient neuromorphic devices demands compact, physically grounded models that seamlessly connect microscopic ferroelectric dynamics to macroscopic circuit behavior. Hafnium–zirconium oxides (, HZO) are compelling candidates owing to their CMOS compatibility, scalability below 10 nm, and tunable transition between ferroelectric and volatile responses. Yet, conventional first‐order Landau–Khalatnikov (LK) formulations cannot capture key experimental features such as negative differential capacitance, pseudo‐inductive transients, and leakage interplay. Here we introduce an extended LK framework incorporating inertial second‐order terms, nonlinear charge‐dependent damping, and a modular leakage branch. This unified, simulation‐ready model quantitatively reproduces experimental Q–V characteristics across Hf‐rich, equiatomic, and Zr‐rich HZO compositions, defining operational regimes from non‐volatile memory to volatile, reservoir‐like dynamics. By linking free‐energy landscapes to compact circuit representations, the approach establishes a predictive device–model–circuit co‐design framework that connects atomistic energetics with system‐level simulation, enabling composition‐specific design rules for scalable, low‐power neuromorphic architectures.