Metal Oxide Nano-Interface Boosting the Deep Ultraviolet Adjustable Noise-Filtering In-Sensor Computing

Area of Science:

• Materials Science
• Optoelectronics
• In-sensor Computing

Background:

• Long-afterglow light-emitting devices (LALEDs) offer integrated sensing, memory, processing, and display capabilities for optical and electrical in-sensor computing.
• Conventional deep-ultraviolet (DUV) materials have low conductivity, limiting their application in DUV-responsive LALEDs (DUV-LALEDs).

Purpose of the Study:

• To address the conductivity limitations of DUV materials for DUV-LALEDs.
• To demonstrate a novel material for efficient DUV photon absorption and charge transport.
• To achieve advanced in-sensor computing functionalities using DUV-LALEDs.

Main Methods:

• Fabrication of sol-gel-fractured indium-magnesium oxide (InMgO) for DUV-LALEDs.
• Characterization of InMgO properties, including mobility, memory dynamic range, and responsivity.
• Implementation of DUV-LALEDs in fusion-node reservoir computing networks for recognition tasks.

Main Results:

• InMgO exhibits ideal nano-interface for DUV absorption and efficient charge transport via hopping, achieving high mobility (0.6 cm² V⁻¹ s⁻¹).
• The InMgO-based DUV-LALEDs demonstrated excellent memory dynamic range (70 dB) and responsivity (523.7 A/W).
• Adjustable hardware-level noise filtering and 99% recognition accuracy in multi-dimensional tasks were achieved using the DUV-LALEDs.

Conclusions:

• Sol-gel-fractured InMgO is a promising material for DUV-LALEDs, overcoming conductivity limitations.
• The developed DUV-LALEDs enable efficient in-sensor computing with noise-filtering capabilities.
• Joint material and device optimization offers a cost-effective strategy for fabricating DUV analytical chips.