Coexistence of Long‐Term Memory and Short‐Term Memory in an SiN_x‐Based Memristor

Abstract

Neuromorphic computing has the capacity to emulate the neural structure and the operation of the human brain with energyefficient and flexible processing. [1] When implementing a neuromorphic system on the hardware level, an array of memristors can be used as adaptive synapses. [2,3] In particular, vector matrix multiplication in a memristor has the advantage of greatly reducing the computational burden of the graphics processing unit. Basically, long-term retention is preferred for hardware-based neuromorphic systems in terms of energy efficiency. On the other hand, short-term memory (STM) is suitable for temporal/sequential data processing, which can be emulated by the loss of retention in a memristor. [4,5] Until now, the artificial synaptic devices of memristor type have been reported from a lot of materials to implement the bio-inspired computing. [6-13] SiN x is one of the core materials in the complementary metal oxide semiconductor (CMOS) process, where it is used as a passivating, buffer, masking, and chemical mechanical polishing stopper layer. It has also been used in nonvolatile memory applications, such as in NAND flash [14] and memristor devices. [15-24] The SiN x layer has a different form in flash memory as in charge trapping. The SiN x layer in a memristor acts as a resistive switching layer by controlling conductance with the applied electric

Retina‐Inspired Two‐Terminal Optoelectronic Neuromorphic Devices with Light‐Tunable Short‐Term Plasticity for Self‐Adjusting Sensing

“…Currently, STP that uses electrical stimulation to achieve both STD and STF has appeared in synaptic devices, such as transistors and memristors. [31][32][33][34] Wan et al used electrical stimulation in an IGZO-based transistor to simulate STD and STF. The dynamic double-electron-layer modulation of the ion-conducting electrolyte by protons ensures that the connection scheme determines the high-pass filtering and low-pass filtering functions.…”

Artificial Neurons and Synapses Based on Al/a-SiNxOy:H/P+-Si Device with Tunable Resistive Switching from Threshold to Memory

“…According to the references [ 8 , 9 ], the precise control of ion migration in the resistive switching devices is the performance selection criteria for neuromorphic applications. However, the realization of resistive switching from TS to MS by tuning the Si dangling bond conductive pathway in Si-based RRAM devices is less reported [ 25 , 26 , 27 , 28 , 29 , 30 ].…”

Mechanism and Conductance Range Enhancement of TiN/AlOx/AlOy/ITO Synaptic Devices

“…[ 5–7 ] In particular, the properties of progressive switching memristor I–V have received increasing attention from researchers to simulate the weight update of biological synapses. [ 8,9 ] Recent studies have verified the realizability of memristors in this field, thereby enabling to overcome the limitations of traditional von Neumann architecture. [ 10–12 ]…”

“…[5][6][7] In particular, the properties of progressive switching memristor I-V have received increasing attention from researchers to simulate the weight update of biological synapses. [8,9] Recent studies have verified the realizability of memristors in this field, thereby enabling to overcome the limitations of traditional von Neumann architecture. [10][11][12] Binary transition metal oxides have emerged as candidate materials for artificial synaptic functional layers owing to their advantages such as good scalability, simple structure, low power consumption, and great compatibility with complementary metal oxide semiconductor processes.…”