Artificial Optoelectronic Synapse Based on 18R-Phase SnSe2 for Neuromorphic Computing

In recent years, optoelectronic synapses made from 2D materials like black phosphorus, MoS₂, InSe, and organic compounds have rapidly developed. A suitable bandgap enables them to respond to light stimuli in a manner similar to the responses of the human eye's visual neurons. However, most synapses made from these materials suffer from drawbacks such as high costs, complex device structures, and narrow spectral response ranges. This paper introduces a low-energy consumption artificial optoelectronic synapse based on 18R-SnSe₂, prepared using mechanical exfoliation, which demonstrates excellent synaptic functions within the visible to near-infrared range. The modulation of optical pulses achieves the conversion from short-term memory (STM) to long-term memory (LTM). Furthermore, through simulations based on convolutional neural network (CNN) algorithms, the device achieves high-accuracy recognition of handwritten digit images and has strong fault tolerance against noise. Even at a noise level of 40%, it maintains an accuracy of over 89%, revealing great application potential in neuromorphic computing.