Spiking neurons from tunable Gaussian heterojunction transistors

Spiking neural networks (SNNs) offer energy-efficient neuromorphic computing by leveraging spatiotemporal processing, spiking sparsity, and high inter-neuron bandwidth. While silicon-based technology can be used, it requires complex multi-transistor circuits, limiting integration density. This work demonstrates highly tunable dual-gated Gaussian heterojunction transistors (GHeTs) for simplified spiking neuron implementation using monolayer molybdenum disulfide (MoS2) and single-walled carbon nanotubes (CNTs). These dual-gated GHeTs emulate biological neuron spike-generating ion channels, enabling various spiking responses (phasic, delayed, tonic bursting). This tunable Gaussian response has significant potential for neuromorphic computing and other applications like telecommunications, computer vision, and natural language processing.

Megan E. Beck, Ahish Shylendra, Vinod K. Sangwan, Silu Guo, William A. Gaviria Rojas, Hocheon Yoo, Hadallia Bergeron, Katherine Su, Amit R. Trivedi, Mark C. Hersam