Bidirectional Photoresponse in a Mixed-Dimensional MoS2/Ge Heterostructure and Its Optic-Neural Synaptic Behavior for Colored Pattern Recognition

Realization of multi-functional synaptic devices is imperative to deploy high-performance brain-like vision systems. Here, a junction field-effect transistor based on a two-dimensional MoS2/three-dimensional Ge heterojunction structure is presented. The heterojunction gate ensures efficient electrostatic gate control and eliminates hysteresis, which highlights its capability for high-performance electronic circuit applications. With the visible absorptive MoS2 channel and the infrared absorptive Ge gate, a positive photoresponse to visible light and a negative photoresponse to infrared light are obtained simultaneously. Responsivities reach as high as 24.9 and -0.40 A/W at 532 and 1550 nm, corresponding to a detectivity of 7.9 x 1011 and -1.3 x 1010 Jones. Meanwhile, based on its optical stimulation-controlled synaptic behaviors, an optoelectronic neural network for colored digit pattern recognition is developed. The recognition accuracy reaches as high as 89.6 and 91.6% for visible and near-infrared patterns, respectively. These results demonstrate a new pathway for realizing novel multi -wavelength neuromorphic visual systems.

Automated summary

In this study, a junction field-effect transistor based on a two-dimensional MoS2/three-dimensional Ge heterojunction structure is introduced. The heterojunction gate allows efficient electrostatic gate control and eliminates hysteresis, making it suitable for high-performance electronic circuit applications. By utilizing the visible absorptive MoS2 channel and the infrared absorptive Ge gate, simultaneous positive photoresponse to visible light and negative photoresponse to infrared light are achieved. The developed optoelectronic neural network based on the device's optical stimulation-controlled synaptic behaviors shows high recognition accuracy for visible and near-infrared patterns. These findings provide a new pathway for realizing novel multi-wavelength neuromorphic visual systems.