Controllable photoelectric dual-output artificial light-emitting synapse device

Artificial synapse devices serve as a critical hardware foundation for neuromorphic computing. However, traditional devices predominantly rely on single electrical regulation, which limits the intuitive acquisition of synaptic weights and lacks independent inhibitory regulation mechanisms. This study introduces a charge -trapping layer of poly (4-vinylphenol) (PVP) into the quantum dot light-emitting diode (QLED) structure to fabricate an artificial light-emitting synapse device capable of achieving synchronous coupling between electrical conductance modulation and light output. The device demonstrates typical synaptic plasticity behavior under electrical pulse stimulation. The amplitude of the excitatory postsynaptic current (EPSC) can be continuously modulated by varying the pulse frequency and amplitude, while the paired-pulse facilitation (PPF) index exhibits an exponential decay characteristic with increasing pulse intervals. Furthermore, the introduction of ultraviolet (UV) light as an independent suppression signal effectively reduces the device's conductance within seconds, providing precise regulation of the synaptic weight accumulation induced by electrical pulses. An artificial neural network constructed using this device achieved a classification accuracy of 93.05% in the MNIST handwritten digit recognition task, which is 5.88% higher than the accuracy achieved without light regulation. These results highlight the significant application potential of this device in neuromorphic computing and visual information processing.