Y2O3-Based Crossbar Array for Analog and Neuromorphic Computation

Abstract

we report an implementation of (8 x 8) Y2O3-based memristive crossbar array (MCA) out of a total dimension of (30 x 25) array fabricated by utilizing a dual ion beam sputtering (DIBS) system. The selected (8 x 8) MCA is further used to electrically write random alphabets and perform synaptic learning characteristics to perform analog and neuromorphic computing applications. The MCA effectively exhibits multiple current levels and mimics var-ious artificial synaptic properties with superior bidirec-tional switching responses. The MCA mimics potentiation, depression, and different Hebbian learning-based spike-time-dependent plasticity rules, suggesting the importance of the Y2O3-based MCA for large-scale neuromorphic and analog computations. This work provides different insights into the design of an artificial synapse by utilizing Y2O3 as a switching oxide in memristors.