Multifunctional ZnO-based optical memristors for synapse-neuron integration and neuromorphic vision systems

Area of Science:

• Materials Science
• Neuroscience
• Electrical Engineering

Background:

• Memristive devices offer promising avenues for neuromorphic computing.
• Integrating optical and synaptic functionalities is crucial for advanced AI systems.
• ZnO-based materials present unique properties for electronic applications.

Purpose of the Study:

• To analyze ZnO-based memristive devices with ITO electrodes for optical and neuromorphic applications.
• To investigate the synaptic properties and nociceptive functionalities of the ITO/ZnO/ITO stack.
• To evaluate the device's performance in image classification and neural network modeling.

Main Methods:

• Fabrication and characterization of ITO/ZnO/ITO memristive devices.
• Cross-sectional Transmission Electron Microscopy (TEM) and energy-dispersive X-ray (EDX) analysis.
• Optical stimulation (405 nm) to investigate synaptic plasticity, nociception, and computational capabilities (Restricted Boltzmann Machine).
• Reservoir computing for MNIST image classification.

Main Results:

• The device demonstrated short-term memory (STM) and paired-pulse facilitation (PPF) under high light intensity.
• Key nociceptor features like threshold, non-adaptation, and sensitization were reproduced.
• Achieved 97.35% accuracy in MNIST image classification using reservoir computing.
• Exhibited neuron-level computation via Restricted Boltzmann Machine (RBM) at low light intensity.

Conclusions:

• ZnO-based memristive devices show potential as visual nociceptors, enhancing neuromorphic vision with danger-signaling.
• The devices integrate neuromorphic, nociceptive, and computational features for multifunctional applications.
• This research paves the way for bio-inspired devices capable of self-protection and adaptability.