High-Performance Artificial Synapse Developed by HZO on (110) NSTO

Abstract

The neuromorphic computing paradigm, characterized by nonvolatile artificial synapses, is seen as crucial for overcoming the limitations of the von Neumann architecture. Moreover, the effective implementation on a large scale necessitates enhanced performance, reduced energy consumption, and increased robustness of synaptic devices. In this work, a bipartite artificial synapse was developed using Hf 0.5 Zr 0.5 O 2 (HZO) ferroelectric thin films on Nb:SrTiO 3 (NSTO) (110) substrates achieving a substantial storage window and an exceptionally high switching ratio spanning 7 orders of magnitude. Notably, the device demonstrates the capability to transition between multiple resistive states, facilitating multilevel data storage. Additionally, the fabricated device emulates the behavior of biological synapses, exhibiting synaptic plasticity feathers such as long-term potentiation/depression (LTP/LTD), paired-pulse facilitation (PPF), and spike-timing dependent plasticity (STDP). These findings underscore the significant potential of HZO artificial synapse devices for neuromorphic computing applications.