Variable-sensitivity photodetector that uses a metal-semiconductor-metal structure for optical neural networks

Area of Science:

• Optoelectronics
• Artificial Intelligence Hardware

Background:

• Optical neural networks require efficient and dynamic components.
• Traditional photodetectors lack tunable sensitivity for complex computations.

Purpose of the Study:

• To develop a novel photodetector with variable sensitivity for optical neural networks.
• To demonstrate its capability for implementing dynamic synaptic interconnections.

Main Methods:

• Utilized a metal-semiconductor-metal structure for the photodetector.
• Modulated quantum efficiency using an applied bias voltage.
• Investigated bipolar current flow and sensitivity dependence on voltage.

Main Results:

• Achieved a linear dependence of photodetector sensitivity on bias voltage.
• Demonstrated the device functions as a multiplier of incident light intensity and bias voltage.
• Successfully fabricated and tested a 4x4 array of these photodetectors.

Conclusions:

• The variable-sensitivity photodetector is suitable for optical neural network applications.
• The device enables dynamic synaptic interconnections crucial for advanced AI.
• The demonstrated 4x4 array confirms the practical feasibility of this technology.