Conductive Islands Assisted Resistive Switching in Biomimetic Artificial Synapse for Associative Learning and Image Recognition

Abstract

Solution-processed memristor devices hold significant potential for advancing synaptic applications, offering scalable, cost-effective, and efficient solutions for next-generation neuromorphic systems. These devices replicate the behavior of biological synapses with their gradual and continuous changes in resistance, making them promising for neuromorphic computing and artificial neural networks. However, understanding the temporal characteristics and cumulative effect of successive pulses on the devices is essential for emulating the dynamics of biological synapses. This study proposes a hybrid materials-based conductive islands-assisted resistive switching (CIARS) in a solution-processed active layer consisting of thermally exfoliated graphitic carbon nitride (g-C₃N₄ abbreviated as CN) nanosheets embedded with silver nanoparticles (Ag NPs). The fabricated devices demonstrate repeatable and bipolar memory features with a lower operating voltage (≤0.5 V), emulating the frequency and amplitude-dependent synaptic plasticities. The threshold switching occurs due to the formation of conduction filaments (CFs) of silver ion (Ag⁺), whereas the analog behavior results from the voltage pulse-dependent modulation of Schottky barrier height combined with metallic CFs formation. The experimental findings demonstrate the efficacy of the CIARS mechanism within the material platform, highlighting its potential for applications in associative learning, Morse code detection, and image recognition.