A single layer spin-orbit torque nano-oscillator

Mohammad Haidar,Mykola Dvornik,Afshin Houshang,Roman Khymyn,Ahmad A Awad,Johan Åkerman

doi:10.1038/s41467-019-10120-4

Abstract

Spin torque and spin Hall effect nano-oscillators generate high intensity spin wave auto-oscillations on the nanoscale enabling novel microwave applications in spintronics, magnonics, and neuromorphic computing. For their operation, these devices require externally generated spin currents either from an additional ferromagnetic layer or a material with a high spin Hall angle. Here we demonstrate highly coherent field and current tunable microwave signals from nano-constrictions in single 15–20 nm thick permalloy layers with oxide interfaces. Using a combination of spin torque ferromagnetic resonance measurements, scanning micro-Brillouin light scattering microscopy, and micromagnetic simulations, we identify the auto-oscillations as emanating from a localized edge mode of the nano-constriction driven by spin-orbit torques. Our results pave the way for greatly simplified designs of auto-oscillating nano-magnetic systems only requiring single ferromagnetic layers with oxide interfaces.

Highlights

Spin torque and spin Hall effect nano-oscillators generate high intensity spin wave autooscillations on the nanoscale enabling novel microwave applications in spintronics, magnonics, and neuromorphic computing
The spin Hall effect[4,5,6] was instead used to create pure spin currents injected into an adjacent ferromagnetic layer, in devices known as spin Hall nano-oscillators (SHNOs)[3]
SHNOs still suffer from a number of drawbacks and conflicting requirements: (i) the non-magnetic layer generating the spin current dramatically increases the zerocurrent spin wave damping of the ferromagnetic layer, (ii) the surface nature of the spin Hall effect requires ultrathin ferromagnetic layers for reasonable threshold currents, which further increases the spin wave damping, and (iii) since the current is shared between the driving layer and the magnetoresistive ferromagnetic layer, neither driving nor signal generation benefits from the total current, leading to non-optimal threshold currents, low output power, unnecessary dissipation, and heating

Summary

Introduction

Spin torque and spin Hall effect nano-oscillators generate high intensity spin wave autooscillations on the nanoscale enabling novel microwave applications in spintronics, magnonics, and neuromorphic computing. SHNOs still suffer from a number of drawbacks and conflicting requirements: (i) the non-magnetic layer generating the spin current dramatically increases the zerocurrent spin wave damping of the ferromagnetic layer (up to 3× for NiFe/Pt13), (ii) the surface nature of the spin Hall effect requires ultrathin ferromagnetic layers for reasonable threshold currents, which further increases the spin wave damping, and (iii) since the current is shared between the driving layer and the magnetoresistive ferromagnetic layer, neither driving nor signal generation benefits from the total current, leading to non-optimal threshold currents, low output power, unnecessary dissipation, and heating It would, be highly advantageous if one could do away with the additional metal layers for operation. By performing additional thickness dependent spin-torque FMR measurements, we conclude that the anti-damping spin-torque originates from the interfaces in the devices

Methods

Results

Conclusion