A Novel Ambipolar Ferroelectric Tunnel FinFET based Content Addressable Memory with Ultra-low Hardware Cost and High Energy Efficiency for Machine Learning
Jin Luo, Weikai Xu, Boyi Fu, Zheru Yu, Mengxuan Yang, Yiqing Li, Qianqian Huang, Ru Huang
Abstract
In this work, a novel ferroelectric tunnel FET (FeTFET) based content addressable memory (CAM) cell with only one transistor is proposed and experimentally demonstrated based on 14-nm FinFET technology node for the first time. By exploiting and modulating the non-volatile ferroelectric polarization for entry storage and the unique feature of ambipolar tunneling current for input searching query, XNOR-like matching operation of ternary CAM can be realized in one Fe-FinTFET without the need of twin complementary circuit branches. Moreover, benefiting from the ferroelectric multi-domain feature and the intrinsic steep slope from tunneling mechanism, multi-bit CAM function for high density can also be experimentally implemented in the fabricated singe Fe-FinTFET device. Based on the proposed FeTFET CAM design, Hamming and Manhattan distance computing are demonstrated with high energy efficiency, showing its great potential for area- and energy-efficient machine learning.