Abstract
The development of energy-efficient neuromorphic systems can be enabled by novel devices that can emulate biological dynamics with minimal power overhead. Negative differential resistance (NDR) devices are a promising direction within neuromorphic research due to their intrinsically sharp nonlinear behavior, which is a key characteristic of biological neurons. In this work, we present a graphene-hBN interlayer tunneling field-effect transistor (ITFET), whose tunable NDR current-voltage characteristics enable efficient spike generation in a neuron circuit. Using an experimentally-calibrated Verilog-A model, results demonstrate up to 2.9 × energy savings per spike over a CMOS analog spike generator. We further show that the same device can be used to create bistable synaptic memory cells. These results demonstrate ITFETs as a robust candidate for next-generation neuromorphic architectures.
Funding Information
  • Army Research Office (W911NF-22-1-0160)

This publication has 23 references indexed in Scilit: