A synaptic phototransistor for image processing and memory

With the integration of the Internet of Things and artificial intelligence technology, the amount of data is increasing exponentially. However, the traditional von Neumann architecture makes the processing of nonstructural visual data inefficient. In view of this, many neuromorphic visual systems have been proposed. However, the data transmission process through the photoreceptors to the computing unit still results in data redundancy and additional power consumption. Therefore, it is necessary to develop intelligent visual systems capable of performing calculations within pixels, which requires the development of photodetectors with sensing, memory and processing functions. Here, a neuromorphic synaptic phototransistor based on MoS2 is prepared. The proposed phototransistor shows synaptic properties, including short-term/long-term synaptic plasticity, paired-pulse facilitation and forgetting properties. As a proof of concept, it is demonstrated that the image preprocessing and memory ability of the intelligent visual system based on this neuromorphic phototransistor. The proposed phototransistor can be used as a photoreceptor and a biological synapse in an intelligent visual system, which shows its potential application in the efficient processing of visual information.