Time-Domain Simulation and Optimization of the Memory Window for HZO-Based FeFETs Using the NLS Model

Hafnium-zirconium oxide (HZO)-based ferroelectric field-effect transistors (FeFETs) are expected to become core devices for new embedded memory and compute-in-memory systems. However, existing simulations rely on finite-element-based TCAD tools, which are computationally intensive and time-consuming, and they struggle to account for the dynamic flipping of ferroelectric domains. This paper utilizes a time-domain simulation framework based on the nucleation-limited switching (NLS) model coupled with the surface potential of a MOSFET, enabling a self-consistent solution for polarization and electrical characteristics; a Monte Carlo method is employed to simulate device variability, and Shmoo plots are used to identify optimal programming and erasure process windows; an integrated solution is proposed for 22 nm FDSOI devices, addressing geometric scaling, modification of the Landau-Khalatnikov (L-K) dynamic model for ultrathin ferroelectric layers, and suppression of short-channel effects. Model validation is limited to selected operating metrics, and predictive accuracy outside the calibrated cases requires additional independent datasets. This method enables end-to-end simulation of FeFETs, from material polarization and device electrical characteristics to performance optimization, thereby providing model-based analytical and design support for the development of advanced, ultra-low-power FeFETs.