Plasma‐Engineered AlGaN/GaN Optoelectronic Synapses for Low‐Power Neuromorphic Computing

摘要

Abstract The development of hardware‐level synaptic systems is crucial for enabling the advancement of modern humanoid robotics and artificial intelligence systems to emulate biological vision. While optoelectronic synapses offer a promising solution, current implementations often require complex multi‐material integration and floating gate configurations, hindering scalability and energy efficiency. Here, a two‐terminal optoelectronic synaptic device based on AlGaN/GaN heterostructures that integrates sensing and memory functionalities into a single device is introduced. By optimizing the device architecture through oxygen plasma treatment, UV sensitivity is significantly enhanced and carrier recombination is reduced, enabling the device to perceive weak UV light (0.15 µW cm⁻ 2 ) with ultralow energy consumption (25.3 fJ per synaptic event). This device exhibits rich synaptic behaviors, including paired‐pulse facilitation, short‐term plasticity, and long‐term plasticity, essential for emulating biological learning and memory processes. Furthermore, its potential for diverse information processing tasks such as photonic information encryption, associative learning, and forgetting is demonstrated. This work paves the way for the development of highly integrated, energy‐efficient neuromorphic systems capable of mimicking the complexity and efficiency of the human brain.