Bi-wavelength-Gated Plasticity in Defect-Engineered β - Ga2O3 Optoelectronic Synapses for Bioinspired Visual Systems

Abstract Optoelectronic synapses that integrate light sensing, memory, and preprocessing are highly desirable for neuromorphic vision. However, most reported optoelectronic synapses rely on a single optical stimulus and mainly exhibit passive modulation of the photoresponse, making it difficult to actively distinguish transient interference from information that should be retained. Here, we report a wavelength-gated optoelectronic synapse based on defect-engineered β-Ga2O3, in which short-term (STP) and long-term synaptic plasticity (LTP) can be selectively programmed by dual-wavelength ultraviolet illumination. X-ray photoelectron spectroscopy (XPS) and temperature-dependent photoluminescence (PL) measurements indicate that N incorporation change the defect states of β-Ga2O3 by introducing N-related defect states and altering oxygen vacancy (VO), which provide effective pathways for wavelength-dependent carrier trapping, release, and recombination dynamics. Under 254 nm illumination, near-band-edge excitation leads to fast carrier recombination and STP, whereas 365 nm photons activate sub-bandgap defect states, enabling persistent photoconductivity and LTP. Benefiting from this wavelength-selective memory behavior, a physical-layer encrypted image transmission strategy is further demonstrated. Furthermore, a wavelength-gated reservoir-computing framework is constructed for noise-robust face recognition by coupling device-level preprocessing with reservoir readout. Owing to the transient erasure of noise information and the long-term retention of valid visual features, the system with preprocessing achieves a recognition accuracy of 94%, outperforming the system without preprocessing, which achieves 83%. This work provides a defect-engineering strategy for wavelength-programmable β-Ga2O3 optoelectronic synapses and highlights their potential for selective memory, secure optical communication, and neuromorphic visual.