Gate-Tunable Highly Linear Bipolar Photoresponse in Se@SWCNT Adaptive Neurons for Dynamically Programmable...

Area of Science:

• Materials Science
• Neuroscience
• Electrical Engineering

Background:

• Neuromorphic systems require tunable and controllable photoconductive devices.
• Existing devices suffer from asymmetric and nonlinear properties, limiting training tasks.
• Brain-inspired computing needs simplified integration and fabrication for complex visual environments.

Purpose of the Study:

• To present a programmable synaptic transistor for advanced optical neuromorphic systems.
• To overcome limitations of existing devices by enabling gate-controlled positive and negative responses.
• To demonstrate a simplified approach for array integration and wafer-scale fabrication.

Main Methods:

• Fabrication of a 1D van der Waals heterojunction using Selenium@Single-Walled Carbon Nanotubes (Se@SWCNT).
• Utilizing gate control for tunable positive and negative responses in the phototransistor.
• Characterization of weight variation linearity, persistent photoconductivity, and memory states under optical stimulation.

Main Results:

• Achieved improved symmetry and linearity (R² > 0.99) in weight variation.
• Demonstrated linear persistent photoconductivity and negative photoconductivity with over 128 memory states.
• Successfully processed three tasks of increasing complexity by adjusting light intensity and wavelength.

Conclusions:

• The Se@SWCNT transistor offers a simplified, highly controllable solution for optical neuromorphic systems.
• The device's adaptability to dynamic visual environments facilitates transitions between bio-inspired brain regions.
• This innovation significantly advances brain-like computing and bio-inspired vision with enhanced accuracy and dynamic switching.