Bias-Switchable ZnO/AlScN Heterojunction for Self-Powered UV Photodetection and Neuromorphic Learning.

Abstract

The convergence of ultraviolet (UV) communication and neuromorphic computing demands multifunctional devices that integrate high-speed photodetection with synaptic learning, yet the intrinsic mismatch between fast photoresponse and slow memory retention poses a fundamental challenge for monolithic integration. Here, we report a dual-mode UV optoelectronic device based on a ZnO/AlScN heterojunction, in which flexible switching between a self-powered photodetection mode and a neuromorphic mode is achieved through bias modulation. In mode I, operating at zero bias, the device exploits the type-II band alignment and built-in electric field of the heterojunction to realize a rapid response to 365 nm UV illumination, exhibiting a pA-level dark current, an on/off ratio of ∼ 4.2 × 103, rise/decay times of 279/64 ms, a responsivity of 101.06 mA/W, and a specific detectivity of 1.15 × 1012 Jones. It also demonstrates UV optical communication based on ASCII encoding. In mode II, operating under a 1 V bias, enhanced carrier separation together with oxygen-vacancy-induced persistent photoconductivity (PPC) in ZnO enables a series of neuromorphic behaviors, including excitatory postsynaptic current, paired-pulse facilitation, the transition from short-term to long-term memory, memory-forgetting characteristics, and Pavlovian associative learning. Our findings demonstrate the potential of the ZnO/AlScN heterojunction as a versatile two-terminal platform integrating self-powered UV detection and neuromorphic learning, enabled by interfacial band engineering and bias-controlled oxygen-vacancy dynamics for dual-mode switching. This strategy offers an approach for developing multifunctional optoelectronic devices for UV communication, biomimetic visual perception, and neuromorphic information processing.