FeFETs for Near-Memory and In-Memory Compute

Ferroelectric Field Effect Transistors (FeFETs), where a ferroelectric material is placed on the gate of a transistor has seen resurgence in the recent years with the advent of doped HfO2 as a ferroelectric material [1]. Here we will discuss the potential and challenges for FeFETs as a memory solution for near-memory and in-memory computing. The ability to integrate the ferroelectric oxide on high-performance Si channels means that fast READ speed is achievable. In addition, recent demonstrations have shown that HfO2 based Ferroelectrics scale down to very small footprint area, which should enable achieving very large density. These aspects together with the fact that write energy in the device is low (<1fJ), point to the potential of a very high bandwidth, low power, embedded memory that could be highly beneficial to AI applications.