Bias-Tunable Two-Terminal Organic Photodetector for Intelligent Imaging.

Area of Science:

• Organic electronics
• Photodetector technology
• Machine vision

Background:

• In-sensor computing reduces computational load in machine vision by processing data directly on image sensors.
• Reconfigurable photodetectors are key for in-sensor computing, enabling pixel-level signal modulation.
• Current approaches using novel materials or complex transistor architectures face scalability and resolution limitations.

Purpose of the Study:

• To develop a simple, scalable reconfigurable photodetector for in-sensor computing applications.
• To demonstrate voltage-controlled, quasi-linear tuning of photodetector responsivity.
• To showcase the device's potential for neural network emulation and image preprocessing.

Main Methods:

• Fabrication of a vertically stacked, two-terminal organic photodetector using a donor/acceptor/donor tri-layer active material.
• Characterization of the photodetector's electrical and optical properties, including dynamic range and responsivity tuning.
• Implementation of a 4x4 photodetector array to perform photocurrent mapping and kernel operations for image processing emulation.

Main Results:

• The device achieved an 80 dB linear dynamic range.
• Quasi-linear responsivity tuning from -26 mA/W to 77 mA/W was achieved with bias variation from -1 V to 1 V.
• The photodetector array successfully emulated image preprocessing filters by performing kernel operations.

Conclusions:

• The simple, vertically stacked organic photodetector offers significant advantages over complex existing designs.
• Voltage-controlled tunable responsivity enables direct implementation of neural network weights.
• The straightforward fabrication process facilitates monolithic integration, advancing compact and efficient intelligent image sensors.