Related Experiment Video
Updated: Jan 19, 2026
09:14
Flow-assisted Dielectrophoresis: A Low Cost Method for the Fabrication of High Performance Solution-processable Nanowire Devices
Published on: December 7, 2017
8.2K
Design of a Power Efficient Artificial Neuron Using Superconducting Nanowires
Emily Toomey1, Ken Segall2, Karl K Berggren1
1Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, Cambridge, MA, United States.
Frontiers in Neuroscience
|September 26, 2019
Summary
Superconducting nanowires can mimic biological neurons and synapses for efficient artificial intelligence hardware. This novel approach offers a promising path towards low-power, high-performance computing systems inspired by the brain.
Area of Science:
- Neuroscience and Artificial Intelligence
- Superconducting electronics
Background:
- Growing demand for high-performance, low-power information processing.
- Spiking neural networks (SNNs) offer bio-realistic, efficient computation.
- Hardware SNNs are an attractive alternative to software-based approaches.
Purpose of the Study:
- To develop an artificial neuron and synapse using superconducting nanowires.
- To demonstrate the biological characteristics and functional capabilities of the proposed hardware.
Main Methods:
- Utilized coupled superconducting nanowires to emulate neuron spiking behavior.
- Employed electrothermal circuit simulations to validate neuron characteristics.
- Designed a variable inductive synapse leveraging nanowire inductance non-linearity.
Main Results:
- The nanowire neuron successfully reproduced key characteristics of biological neurons.
- The inductive synapse demonstrated excitatory and inhibitory control with direct fan-out.
- The energy performance of the nanowire neuron is competitive with existing technologies.
Conclusions:
- Superconducting nanowires provide a viable platform for brain-inspired computing hardware.
- The proposed neuron and synapse designs advance the development of efficient SNNs.
- This research paves the way for next-generation, low-power AI.
