Chalcogenide-Based Brain-Inspired Photo-Synapses for Neuromorphic Vision Sensor: An Experimental and Theoretical...

Area of Science:

• Materials Science
• Neuromorphic Engineering
• Artificial Intelligence

Background:

• 2D chalcogenide memristors offer optically and electrically tunable synaptic behavior for artificial biological visual systems.
• 2D van der Waals materials like SnS2 enable multifunctional optoelectronic devices through rational design.

Purpose of the Study:

• To simulate a human biological visual system using multifunctional optoelectronic synaptic devices based on SnS2.
• To investigate the potential of SnS2 memristors for memory and logic functions relevant to the brain's visual cortex.
• To evaluate the performance of SnS2-based devices in machine vision tasks, including image recognition.

Main Methods:

• First-principles-based Density Functional Theory (DFT) calculations to determine SnS2 semiconductor properties (bandgap: 2.47 eV).
• Fabrication and characterization of SnS2 optoelectronic synaptic devices.
• Machine vision simulations using SnS2 retina devices for MNIST dataset recognition.
• Experimental determination of synaptic characteristics to guide hybrid AI framework training.

Main Results:

• SnS2 exhibits excellent photosensitivity, enabling wavelength-sensitive responses for letter recognition and image memory.
• The SnS2 retina device achieved 98.51% accuracy for MNIST datasets in machine vision simulations.
• Devices operate at ultralow voltage (0.1 V) with low energy consumption (0.345 nJ/event).
• SnS2 photo-synaptic devices perform OR and AND logic operations by varying optical input wavelengths.

Conclusions:

• SnS2-based optoelectronic synaptic devices show significant promise for developing advanced robotic vision systems.
• The integration with hybrid AI frameworks enables innovative neuromorphic computing capabilities.
• These findings pave the way for next-generation artificial vision systems mimicking biological retina functions.