Optoelectronic array of photodiodes integrated with RRAMs for energy-efficient in-sensor computing

Area of Science:

• Materials Science
• Electrical Engineering
• Computer Science

Background:

• The Internet of Things (IoT) demands miniaturized, high-efficiency, low-power computing devices at the edge.
• In-sensor computing offers a solution for in-situ data processing directly within sensor arrays, reducing data transmission bottlenecks.

Purpose of the Study:

• To develop an optoelectronic array for in-sensor computing by integrating photodiodes (PDs) with resistive random-access memories (RRAMs).
• To demonstrate the potential of this integrated array for efficient optical image recognition in IoT devices.

Main Methods:

• Fabrication of a PD-RRAM unit cell capable of reconfigurable optoelectronic output and photo-responsivity through RRAM programming.
• Construction and testing of a 3x3 PD-RRAM array to perform optical image recognition tasks.

Main Results:

• The PD-RRAM unit cell demonstrated tunable optoelectronic properties based on RRAM resistance states.
• The 3x3 array successfully performed optical image recognition with ultralow latency and minimal power consumption.

Conclusions:

• The integrated PD-RRAM optoelectronic array presents a promising primitive for energy-efficient in-sensor computing.
• This technology holds significant potential for advancing future IoT applications requiring edge computing capabilities.