Artificial Visual Systems Fabricated with Ferroelectric van der Waals Heterostructure for In-Memory Computing...

Area of Science:

• Materials Science
• Neuroscience
• Electrical Engineering

Background:

• Artificial neural networks and retina-inspired systems require integrated sensing, processing, and memory.
• Existing devices often lack the ability to perform all these functions in a single unit.

Purpose of the Study:

• To propose and demonstrate a ferroelectric field-effect transistor-based artificial visual system.
• To investigate the device's capability for integrated optical signal management.

Main Methods:

• Fabrication of a van der Waals heterostructure using CuInP2S6 and α-In2Se3.
• Utilizing coupled dipole polarizations in α-In2Se3 controlled by electrical and optical signals.

Main Results:

• Demonstrated a retina-inspired artificial visual system with integrated sensing, processing, and memory.
• Device operates across 405-850 nm, detects low light (0.03 mW/cm²), and shows color recognition and high paired-pulse facilitation (~170%).
• Exhibits short- to long-term memory transitions, pattern recognition, light adaptation, optical logic, and event learning.

Conclusions:

• The ferroelectric heterostructure device functions as an artificial visual system.
• This technology serves as a crucial step towards all-in-one sensing and memory-processing devices.