[Paper Review] Antiferromagnetic Skyrmion based Energy-Efficient Leaky Integrate and Fire Neuron Device

Namita Bindal, Ravish Kumar Raj+3 more|arXiv (Cornell University)|Jul 21, 2022

Magnetic properties of thin films4 citations

TL;DR

This paper proposes an antiferromagnetic skyrmion-based neuromorphic device that emulates leaky-integrate-and-fire (LIF) neuron behavior using thermal or perpendicular magnetic anisotropy (PMA) gradients to induce skyrmion motion. The device achieves energy-efficient operation with 4.32 fJ per LIF cycle and enables readout via tunnel magnetoresistance (TMR) changes up to 9.2%, demonstrating a promising path for low-power spintronic neuromorphic computing.

ABSTRACT

The development of energy-efficient neuromorphic hardware using spintronic devices based on antiferromagnetic (AFM) skyrmion motion on nanotracks has gained considerable interest. Owing to its properties such as robustness against external magnetic fields, negligible stray fields, and zero net topological charge, AFM skyrmions follow straight trajectories that prevent their annihilation at nanoscale racetrack edges. This makes the AFM skyrmions a more favorable candidate over the ferromagnetic (FM) skyrmion for future spintronic applications. This work proposes an AFM skyrmion-based neuron device exhibiting the leaky-integrate-fire (LIF) functionality by exploiting thermal gradient or alternatively perpendicular magnetic anisotropy (PMA) gradient in the nanotrack for leaky behavior by moving the skyrmion in the direction to minimize the system energy. Furthermore, it is shown that the AFM skyrmion couples efficiently to the soft ferromagnetic layer of a magnetic tunnel junction enabling efficient read-out of the skyrmion. The maximum change of 9.2% in tunnel magnetoresistance (TMR) is estimated for detecting the AFM skyrmion. Moreover, the proposed neuron device has the energy dissipation of 4.32 fJ per LIF operation thus, paving the path for developing energy-efficient devices in antiferromagnetic spintronics for neuromorphic computing.

Motivation & Objective

• To develop an energy-efficient neuromorphic device using antiferromagnetic skyrmions for brain-inspired computing.
• To implement leaky-integrate-and-fire (LIF) neuron functionality in a spintronic platform using skyrmion dynamics.
• To enable efficient readout of skyrmion states via tunnel magnetoresistance (TMR) in a magnetic tunnel junction.
• To minimize energy dissipation in skyrmion-based neuromorphic devices through intrinsic skyrmion stability and low-energy control mechanisms.
• To demonstrate the feasibility of antiferromagnetic skyrmions as a superior alternative to ferromagnetic skyrmions in nanoscale racetrack devices.

Proposed method

• Utilizes thermal gradient or perpendicular magnetic anisotropy (PMA) gradient along a nanotrack to induce controlled skyrmion motion toward lower energy states, enabling leaky integration.
• Employs a magnetic tunnel junction (MTJ) with a soft ferromagnetic layer to detect skyrmion presence via changes in tunnel magnetoresistance (TMR).
• Designs the nanotrack geometry to ensure straight skyrmion trajectories due to zero net topological charge, preventing annihilation at edges.
• Models skyrmion dynamics under applied gradients to simulate integration and firing behavior mimicking biological LIF neurons.
• Calculates energy dissipation per LIF operation based on skyrmion motion and switching energy in the MTJ.
• Analyzes TMR response to skyrmion presence, estimating a maximum change of 9.2% for reliable signal detection.

Experimental results

Research questions

• RQ1 Can antiferromagnetic skyrmions be used to implement leaky-integrate-and-fire (LIF) neuron behavior in a spintronic device?
• RQ2 How can thermal or PMA gradients be leveraged to induce controlled skyrmion motion for integration functionality?
• RQ3 What is the achievable energy dissipation per LIF operation in an AFM skyrmion-based neuron device?
• RQ4 Can the presence of an antiferromagnetic skyrmion be efficiently detected using tunnel magnetoresistance (TMR) in an MTJ?
• RQ5 How does the stability and trajectory of AFM skyrmions compare to FM skyrmions in nanoscale racetrack architectures?

Key findings

• The proposed device achieves an energy dissipation of 4.32 fJ per leaky-integrate-and-fire (LIF) operation, demonstrating high energy efficiency.
• A maximum tunnel magnetoresistance (TMR) change of 9.2% is estimated when an antiferromagnetic skyrmion is detected, enabling reliable readout.
• Antiferromagnetic skyrmions exhibit straight trajectories due to zero net topological charge, reducing the risk of annihilation at track edges.
• The use of thermal or PMA gradients enables controlled skyrmion motion that mimics the integration phase of LIF neurons.
• The device leverages the intrinsic robustness of AFM skyrmions against external magnetic fields and stray fields, enhancing stability.
• The integration of skyrmion motion with MTJ readout provides a scalable and energy-efficient pathway for antiferromagnetic spintronic neuromorphic computing.