Self-assembled 3D Interconnected Magnetic Nanowire Networks for Neuromorphic Computing.

Area of Science:

• Materials Science
• Nanotechnology
• Computer Engineering

Background:

• Three-dimensional (3D) nanomagnetic systems are promising for neuromorphic computing due to complex spin textures and dynamics.
• Fabrication and characterization challenges limit the investigation of 3D nanomagnetic systems.

Purpose of the Study:

• To introduce interconnected networks of self-assembled magnetic nanowires (NW) as a novel 3D platform for neuromorphic computing.
• To demonstrate the electrical control and programming capabilities of these NW networks for artificial intelligence applications.

Main Methods:

• Fabrication of self-assembled magnetic nanowire networks.
• Electrical characterization of transport pathways and magnetization states.
• Application of current pulses for selective addressing and state switching.

Main Results:

• Networks exhibit multiple unique transport pathways with discrete, electrically controllable magnetization states.
• Pathways function as programmable synaptic weights, allowing diverse programming via current pulses.
• Observed history-dependent magnetic state switching and interconnected transport path functionalities.

Conclusions:

• Interconnected magnetic NW networks offer an efficient platform for neuromorphic computing.
• These networks demonstrate potential as reservoir layers in neural network architectures.
• The demonstrated capabilities pave the way for advanced AI hardware development.