Design Guidelines of Hafnia Ferroelectrics and Gate-Stack for Multilevel-Cell FeFET

In this work, we demonstrate a novel approach to superior multilevel-cell (MLC) ferroelectric field-effect transistor (FeFET) with a large memory window (MW) and negligible ${V}_{T}$ variation toward MLC operation. We realized high ferroelectricity in a relatively thick HZO ferroelectric (FE) layer for FeFET with a large MW [MW $\propto $ thickness of FE layer ( ${T}_{\text {FE}}$ )] based on our understanding of thermodynamics and kinetics. Moreover, we employed the MFMIS gate-stack with a floating gate for FeFET to minimize the ${V}_{T}$ variation with respect to different distributions of phase and grain size. We applied experimentally obtained materials and electrical data from HZO to TCAD simulation to statistically analyze the impact of materials and gate-stack on the MW and the ${V}_{T}$ variation of FeFET. Consequently, we found that increasing the Zr content of HZO effectively reduces the ${V}_{T}$ variation while significantly enhancing the MW. Also, compared to conventional metal–FEs–insulator–silicon (MFIS) FeFET, the (metal–FEs–metal–insulator–silicon) MFMIS FeFET shows significantly reduced ${V}_{T}$ variation and an enlarged MW by inducing uniform channel conductivity due to the equalization effect of the inserted floating gate even for the spatial distribution of FE grains in the HZO layer. Our experimental and simulation methodologies covering materials engineering and gate-stack provide a visible solution for the design of future FeFETs with outstanding MLC operation.

I. Introduction

Along with the dramatically increasing demand in high-performance computing for data analyses and processing in real time [1], [2], emerging memories based on hafnia ferroelectrics (FEs) have attracted a great deal of attention as potential solutions to the demand for superior computing power [3], [4], [5]. A deep understanding of the fundamental physics of hafnia-based FEs has been achieved over the past ten years with the assistance of sophisticated material analysis techniques and first-principles calculation-based estimation [4], [6], [7], [8], [9], [10]. Matured understanding of hafnia FE enabled the accelerated research to enhance the computing power of FE field-effect transistor (FeFET), which is the most representative hafnia FE-based memory application [11].