0D/1D mixed-dimensional heterojunction synaptic transistor for visual information processing

Abstract

Neuromorphic hardware based on artificial synaptic devices offers a promising solution to the speed and power efficiency bottlenecks of conventional von Neumann architectures, paving the way for next-generation intelligent computing. In this study, a artificial synaptic transistor was developed based on a mixed-dimensional heterojunction consisting of zero-dimensional (0D) molybdenum disulfide quantum dots and one-dimensional (1D) InZnO nanowires. Due to the unique 0D/1D heterostructure that can enhance the interfacial carrier separation and trapping efficiency, linear conductance modulation is thereby improved remarkably. This strategy not only emulates essential biological synaptic behaviors, including short-term plasticity, long-term plasticity, and spike-timing-dependent plasticity, but also achieves highly linear synaptic weight updates, resulting in a handwritten digit recognition accuracy of 97.2% in an artificial neural network. The modulation of interfacial charge transfer in the mixed-dimensional heterostructure offers a pathway for high-performance artificial synapses and holds significant potential for energy-efficient visual information processing systems.