Ionoelastomer Synapses With Configurable Synaptic Plasticity
Sijie Zheng, Zhong‐Da Zhang, Xiaowei Wang, Xiuyang Zou, Ziyang Liu, Qingning Li, Ya‐Nan Zhong, Sui‐Dong Wang, Feng Yan
Abstract
An organic heterostructure, composed of an ionoelastomer that comprises polycationic chains and mobile anions, and a semiconducting polymer, forms a new class of artificial synapses. These ionoelastomer synapses update their synaptic weights by modifying device conductance through the spatial redistribution of anions in response to electrical stimuli. The memory effect of these synapses is highly dependent on the specific anion species present, providing a unique means to modulate synaptic plasticity simply through anion selection. The continuously programmable and nonvolatile states of the selected ionoelastomer synapse are harnessed to emulate a soft neural network and perform the handwritten digit and fashion image recognition tasks, which achieve high recognition accuracies comparable to ideal numeric model but only using 16 discrete device states.