Proposing magnetoimpedance effect for neuromorphic computing

Jamilpanah, Loghman; Chiolerio, Alessandro; Crepaldi, Marco; Adamatzky, Andrew; Mohseni, Majid

Abstract

Oscillation of physical parameters in materials can result in a peak signal in the frequency spectrum of the voltage measured from the materials. This spectrum and its amplitude/frequency tunability, through the application of bias voltage or current, can be used to perform neuron-like cognitive tasks. Magnetic materials, after achieving broad distribution for data storage applications in classical Von Neumann computer architectures, are under intense investigation for their neuromorphic computing capabilities. A recent successful demonstration regards magnetisation oscillation in magnetic thin films by spin transfer or spin orbit torques accompanied by magnetoresistance (MR) effect that can give a voltage peak in the frequency spectrum of voltage with bias current dependence of both peak frequency and amplitude. Here we use classical magnetoimpedance (MI) effect in a magnetic wire to produce such a peak and manipulate its frequency and amplitude by means of the bias voltage. We applied a noise signal to a magnetic wire with high magnetic permeability and owing to the frequency dependence of the magnetic permeability we got frequency dependent impedance with a peak at the maximum permeability. Frequency dependence of the MI effect results in different changes in the voltage amplitude at each frequency when a bias voltage is applied and therefore a shift in the peak position and amplitude can be obtained. The presented method and material provide optimal features in structural simplicity, low-frequency operation (tens of MHz-order) and high robustness at different environmental conditions. Our universal approach can be applied to any system with frequency dependent bias responses.

Citation

Jamilpanah, L., Chiolerio, A., Crepaldi, M., Adamatzky, A., & Mohseni, M. (2023). Proposing magnetoimpedance effect for neuromorphic computing. Scientific Reports, 13(1), Article 8635. https://doi.org/10.1038/s41598-023-35876-0

• Journal Article Type: Article
• Acceptance Date: May 25, 2023
• Online Publication Date: May 27, 2023
• Publication Date: May 27, 2023
• Deposit Date: Jun 6, 2023
• Publicly Available Date: Jun 7, 2023
• Journal: Scientific Reports
• Electronic ISSN: 2045-2322
• Publisher: Nature Research (part of Springer Nature)
• Peer Reviewed: Peer Reviewed
• Volume: 13
• Issue: 1
• Article Number: 8635
• DOI: https://doi.org/10.1038/s41598-023-35876-0
• Keywords: Applied physics; Electronics; photonics; device physics; Materials for devices
• Public URL: https://uwe-repository.worktribe.com/output/10835789
• Publisher URL: https://www.nature.com/articles/s41598-023-35876-0