Skip to main content
Browse by:


In this talk, we will present the recent progress on doped HfO2 based ferroelectric devices. First, we will discuss fundamental device physics including the minor loop switching and history effect, the drain-erase scheme and the variability and scalability of ferroelectric field effect transistor (FeFET). Machine learning assisted predictive modeling framework for phase variation is proposed. Second, we will present compute-in-memory (CIM) paradigm for deep neural network acceleration. For MB-level image classification workloads, a new concept on ferroelectric non-volatile capacitor (nvCap) for charge-domain computing is proposed, and the related capacitive crossbar array is experimentally demonstrated. Hybrid synapse that combines charging/discharging mechanism and non-volatile storage is proposed for in-situ training. For GB-level bioinformatics workloads, 3D NAND architecture based on FeFET is proposed for implementing hyperdimensional computing with heterogeneous integration.

Contact: Cynthia Rice