Photonic synaptic devices have attracted increasing attention as key components for artificial vision systems and neuromorphic computing because they can simultaneously sense, process, and memorize optical stimuli. However, single-layer oxide semicond...
Photonic synaptic devices have attracted increasing attention as key components for artificial vision systems and neuromorphic computing because they can simultaneously sense, process, and memorize optical stimuli. However, single-layer oxide semiconductor devices often exhibit limited visible-light absorption or unstable memory behavior, making it difficult to achieve both broadband photoresponse and reliable synaptic characteristics. In this study, double-layered ZnON/InGaZnO photonic synaptic thin-film transistors were investigated to improve RGB photoresponse and memory characteristics through oxygen-defect and interface-defect engineering.
ZnON was used as a visible-light absorption layer owing to its narrow optical bandgap, while InGaZnO was introduced as a capping and auxiliary channel layer to improve electrical stability and provide interface-related charge trapping states. Compared with single-layer ZnON and InGaZnO devices, the ZnON/InGaZnO double-layer device exhibited enhanced RGB photoresponse, improved short-term and long-term memory characteristics, and clear potentiation/depression behavior under repeated optical and gate erase pulses.
The origin of these characteristics was analyzed using spectroscopic ellipsometry, X-ray photoelectron spectroscopy, photo-induced current transient spectroscopy, and transistor measurements under RGB illumination. By controlling the O₂ flow rate during ZnON deposition, the distribution of shallow-level and deep-level defects was modulated. Shallow-level defects and D1 transition near the absorption edge contributed to fast photocurrent generation and enhanced RGB photoresponse, whereas deep-level defects acted as charge trapping states and improved current retention after light removal. Among the investigated conditions, the ZnON-O2-3.6/InGaZnO device showed the most balanced photoresponse and memory characteristics due to the effective formation of both shallow-level defects for photocurrent generation and deep-level defects for charge storage.
These results demonstrate that the photonic synaptic characteristics of ZnON/InGaZnO thin-film transistors can be controlled by engineering oxygen-related and interface-induced defect states. Therefore, double-layer oxide semiconductor structures provide an effective strategy for realizing visible-light-driven photonic synaptic devices for artificial vision systems and neuromorphic optoelectronics.