Ta<sub>2</sub>PdS<sub>6</sub>/MoS<sub>2</sub> Heterojunction Phototransistor for High-Performance Photoelectric Synapses and Graphic Identification
Haijuan Wu, Zhicheng Lin, Jinxiu Liu, Chunchi Zhang, Chao Tan, Zegao Wang
Abstract
Artificial synapses and neurons with efficient, high-speed, and highly parallel information processing capabilities are considered to be a new direction for the next generation of learning, cognition, and data storage. In this work, we have integrated photodetectors and photoelectric synapse in Ta 2 PdS 6 /MoS 2 van der Waals heterostructures, which can be used in photodetection and optical artificial neural networks. We have systematically studied the photoelectric characteristics of the blue-violet to near-infrared (405 ∼ 1550 nm) band. At 633 nm, the responsivity and specific detectivity are as high as 590.36 AW –1 and 5.63 × 10 11 Jones, respectively. In addition, the heterojunction acquired a persistent photoconductivity behavior due to the presence of interfacial defect states, which was used to simulate the synaptic properties of the human brain, such as transition from short-term memory to long-term memory, paired-pulse facilitation, “learning–forgetting–relearning” behavior, and excitatory-postsynaptic current. In addition, on the basis of photoelectric synapse, the recognition of handwritten digits with different noise levels by an artificial neural network was simulated, which shows a high training accuracy (90%). This study lays a foundation for the development of high-performance heterojunctions and artificial synapses using two-dimensional materials.