Design Space Exploration of Ferroelectric Tunnel Junction Toward Crossbar Memories

Abstract

We perform a simulation-based analysis on the potential of emerging ferroelectric tunnel junctions (FTJ) as a memory device for crossbar arrays. Though FTJs are promising due to their low power switching characteristics compared to other emerging technologies, the greatest challenge for FTJs is the trade-off between integration density and read performance. Our analysis highlights the need to co-optimize the ferroelectric thickness of the FTJ and read/write voltages to achieve proper functionality at large array sizes. Our analysis shows that FTJbased crossbar achieves 93% higher sense margin at iso-read power of 116nW (per bit) but this FTJ design comes at cost of 9.28× higher write power at iso-write time of 250 ns. In response, we study the potential trade-offs of design points outside the feasible region to understand what device characteristics are desired overcome such challenges.

Ferroelectric tunnel junctions: promise, achievements and challenges

“…However, inherent characteristics of FTJs pose challenges as individual devices are scaled down for integration, resulting in a proportional reduction in tunneling current, making rapid current detection challenging. Additionally, accurate sensing of ON and OFF states is complicated by the leakage current generated within the crossbar array [269]. While incorporating additional selector devices can address these issues, the associated cost and process complexity are considered impediments.…”

Ferroelectric synaptic devices based on CMOS-compatible HfAlOxfor neuromorphic and reservoir computing applications

“…In this context, various emerging semiconductor devices such as resistive random-access memory (ReRAM), magnetoresistive RAM, FTJ, and phase-change RAM have attracted attention. [32][33][34][35][36] Among these devices, FTJs-which are based on domain switching [37][38][39] -have the advantage of enabling non-destructive operation, [40][41][42] unlike ReRAMs are most widely studied. Therefore, based on this advantage, imitating synaptic plasticity through gradual conductance changes is possible, and research on neuromorphic applications with FTJs can be considered.…”