Polarization-sensitive in-sensor computing in chiral organic integrated 2D p-n heterostructures for mixed-multimodal...

Area of Science:

• Optoelectronics
• Materials Science
• Computer Engineering

Background:

• Sensor-based computing reduces latency and energy use by processing data at the source.
• Existing systems face limitations in real-time data handling and computational flexibility.

Purpose of the Study:

• To develop advanced in-sensor computing capabilities using polarization-sensitive detectors.
• To enable mixed-multimodal image processing within a single, non-reconfigurable circuit.

Main Methods:

• Integration of cholesteric liquid crystal reflectors with 2D van der Waals p-n heterostructures.
• Development of circularly polarized light detectors with high dissymmetry factor and rapid photoresponse.
• Implementation of mixed-multimodal in-sensor computing by controlling detector responsivity via light chirality.

Main Results:

• Achieved a high dissymmetry factor (1.90) for effective separation of circularly polarized images.
• Demonstrated a rapid photoresponse (4 μs) and wide linear dynamic range (114.1 dB).
• Enabled dynamic control of responsivity for blending two image processing modes.

Conclusions:

• The proposed polarization-sensitive detectors are suitable for analog multiply-and-accumulate operations in in-sensor computing.
• Mixed-multimodal in-sensor computing simplifies circuit complexity while preserving kernel optimization.
• This approach advances real-time decision-making and efficient data processing in sensor systems.