Future of ferroelectric tunnel junction - nonvolatile or volatile?

Posted on In ,

Hafnia-based ferroelectric tunnel junctions (FTJs) have been suggested as one of the most promising candidates for emerging memory and computing technologies, thanks to their non-destructive readout, low power consumption, and CMOS and BEOL compatibility. Among all the reported FTJ structures, the Metal-Ferroelectric-Insulator-Metal (MFIM) one features a high endurance, excellent linearity of conductance modulations, and friendly 3D stackability.

FTJs exhibit either nonvolatile or volatile behavior, depending on the device stacks and the resulting depolarization field applied to the ferroelectric (FE) layer. A powerful FTJ model with more comprehensive physical mechanisms that can accurately capture the dynamics responses to depolarization, while maintaining computational efficiency, is essential for guiding device engineering to meet specific requirements of non-volatile (e.g., storage memory and neuromorphic synaptic device) and volatile (e.g., artificial neuron and reservoir computing) applications.

In this work, we present our efforts to develop such a high-efficient and comprehensive FTJ model based on an MFIM structure. Compared with the previously reported FTJ models, i) the multi-domain characteristics of polycrystalline HfO2 FE films are modeled in a more efficient manner, by parameterizing the inhomogeneous mechanism into a partial differential equation (PDE), accurately capturing the polarization reversal dynamics; ii) the potential profile is calculated by a more thorough physical model including both the screening charge (SC) and image force (IF) effects; iii) for the first time, the dynamic retention behaviors and their dependence on key material parameters are systematically studied and discussed, in addition to the modeling of ON-state current density (JON) and the tunneling electroresistance (TER) ratio, which are crucial for both non-volatile and volatile applications.

Reference
High-efficient and Comprehensive Modeling of MFIM Ferroelectric Tunnel Junctions for Non-volatile/Volatile Applications
Yu Li, Hao Jiang*, Jie Yu, Xuanyu Zhao, Xiaodong Wang, Qihan Liu, Yingfen Wei*, Qi Liu, and Ming Liu
2024 IEEE International Memory Workshop (IMW).