Large-area Y2O3-based memristive crossbar array for neuromorphic computation

Abstract

In the last one decade, memristor which is also known as ‘fourth fundamental circuit element’ has attracted extensive attention as it offers numerous potentials applications in the next-generation non-volatile memory (NVM) technology. Memristor is the prominent candidate to replace conventional change-based flash memory technology in futuristic applications such as data storage and neuromorphic computation. Primarily, memristive devices offers several advantages as compared to conventional memories technologies such as its simple structure, ultrafast operational speed, energy efficient, high endurance and retention, high device production yield, three-dimensional (3D) integration capability to fabricate high density memory, low device-to-device (D2D), and ultralow cycle-to-cycle (C2C) variabilities. On the one hand, the design and validation of detailed non-linear analytical models for transition metal oxide (TMO) especially Y2O3 are extensively required to emulate the fundamental resistive switching response along with several neuromorphic characteristics of the memristive devices.