Transparent Optoelectronic Synapse Device Based on the Bi2O3/ZnO Heterojunction for Neuromorphic Computing

Advancements in neuromorphic engineering rely on the development of efficient and scalable synaptic devices capable of emulating the complex functionalities of biological synapses. Here, we propose an optoelectronic synapse device based on a transparent Bi2O3/ZnO heterostructure, which exhibits biological synapse behaviors under the stimuli of 365 nm ultraviolet (UV) light. The light intensity, duration, and pulse number of the 365 nm UV light pulse are adjusted to simulate the synaptic behavior, including excitatory postsynaptic current, paired-pulse facilitation, short-term plasticity, and long-term plasticity. Furthermore, optical potentiation/electrical depression behaviors were conducted. Based on the synaptic plasticity characteristics, the Bi2O3/ZnO optoelectronic synapse demonstrates learning-memory functions and capability of MNIST handwritten digit recognition classification. The variation mechanism of nonvoltage-induced conductivity under UV light stimulus can be attributed to persistent photoconductivity effect in ZnO and Bi2O3, as well as the modulation of heterojunction interface trapping. This work presents a semiconductor microfabrication compatible artificial synapse, suggesting potential applications in visual neuromorphic systems.