All‐In‐One Hardware Devices with Event‐Based Vision Sensor Arrays for Image Sensing, Computing, and Learning
Sen Zhang,Pingdan Xiao,Xitong Hong,Ruohao Hong,Chang Liu,Qianlei Tian,Wanhan Su,Chao Ma,Xingqiang Liu,Kenli Li,Johnny C. Ho,Yawei Lv,Qinghui Hong,Lei Liao,Xuming Zou
DOI: https://doi.org/10.1002/adfm.202306173
IF: 19
2023-08-31
Advanced Functional Materials
Abstract:Here, a flexible artificial vision system integrated with perception, computation, and learning functionalities is constructed using ambipolar SnO optoelectronic synaptic transistor. By introducing HfO2 passivation layer, the SnO transistors exhibit gate‐tunable bidirectional photoresponse, and the proposed learning circuit achieves fast recognition at a 16% Gaussian noise level and high recognition accuracy up to 95.2% for pattern letters. Metal oxide semiconductors (MOSs) are considered as potential candidates for the low‐cost, large‐area fabrication of flexible optoelectronic devices. However, the current optoelectronic devices based on MOSs are limited to unidirectional photoresponse, which constrains the performance of MOSs‐based vision sensors for artificial vision systems. Herein, for the first time, a flexible artificial vision system integrated with optical perception, computation, and learning functionalities is demonstrated using SnO optoelectronic synaptic transistor‐based event‐driven vision sensors to enable dynamic image perception, noise reduction, detection, and recognition. Specifically, an ambipolar SnO transistor is demonstrated by introducing HfO2 passivation layer, which facilitates the movement of O atoms around Sn‐vacancy sites to the HfO2 layer to achieve the transformation from p‐type to ambipolar transport behaviors. More importantly, the HfO2‐passivated SnO transistors exhibit gate‐tunable bidirectional photoresponse behavior, which is essential to simulate the neurobiological functionalities of bipolar cells. This way, the multilayer neural network learning circuit built from SnO transistors achieves fast recognition at a 16% Gaussian noise level and high recognition accuracy up to 95.2% for pattern letters. Under the bending states, recognition accuracies are still retained at 91.2%. These properties are well retained even under the influence of 100% offset of the synaptic programming value.
materials science, multidisciplinary,chemistry, physical,physics, applied, condensed matter,nanoscience & nanotechnology