Spin waves inside a nano-oscillator imaged for the first time

For the first time, researchers have directly imaged the magnetisation dynamics inside a spin Hall nano-oscillator — a nanoscale device that converts direct current into tunable microwave signals and is a promising building block for energy-efficient wireless communication and brain-inspired computing. A Swedish–German team led by the University of Gothenburg and Helmholtz-Zentrum Berlin (HZB) achieved this using time-resolved scanning transmission X-ray microscopy at the MAXYMUS instrument at BESSY II. The results, now published in Advanced Materials, reveal spin-wave features that had escaped previous, indirect measurement techniques.

Spin Hall nano-oscillators (SHNOs) are among the most versatile devices in spintronics: a direct current driven through a nanometre-sized constriction sets the local magnetisation into steady precession, turning a DC input into a tunable radio-frequency output. Because SHNOs are easy to fabricate, CMOS-compatible and can be mutually synchronised in large arrays, they are considered a model platform for tunable microwave sources and for neuromorphic, wave-based computing. Yet what actually happens inside a single oscillator had never been observed directly — the dynamics unfold on nanometre length scales and within fractions of a nanosecond.

The team at the University of Gothenburg fabricated SHNOs based on ultrathin CoFeB layers and characterised their microwave dynamics. To look inside the devices, the group teamed up with the team at HZB, together with colleagues from the Max Planck Institute for Intelligent Systems in Stuttgart. At the scanning transmission X-ray microscope MAXYMUS, operated at the BESSY II electron storage ring in Berlin, the researchers filmed the magnetisation dynamics stroboscopically, using X-ray magnetic circular dichroism as contrast mechanism — with a spatial resolution of a few tens of nanometres and a time resolution well below the period of the oscillations at about 6 GHz.

The X-ray movies show that the spin-wave auto-oscillations concentrate at the two edges of the nanoconstriction — but with a pronounced asymmetry favouring one edge — and that the emitted spin waves propagate in a strongly anisotropic way, moving out perpendicular to the applied magnetic field.  Micromagnetic simulations could only reproduce these observations after three effects that are usually neglected were taken into account. First, the magnetic film is not perfectly uniform, but consists of tiny crystal grains whose boundaries affect the spin waves. Second, nanofabrication slightly weakens the perpendicular magnetic anisotropy — the built-in tendency of the magnetisation to point out of the film plane — which shifts the spots from which the waves are emitted. And third, a subtle interface effect, the Dzyaloshinskii–Moriya interaction, makes spin waves travelling in opposite directions behave differently, which explains the one-sidedness of the emission. Only with all three ingredients did simulation and experiment agree — a clear sign that the physical models used to design SHNOs and their synchronised networks need to be refined.

The measurements also produced an unexpected side observation: under prolonged illumination with intense soft X-rays, the magnetic properties of the samples changed gradually and permanently. This effect concerns the particular material combination studied here — an ultrathin CoFeB layer covered with MgO — and not magnetic materials in general. Yet, precisely this CoFeB/MgO interface is a standard ingredient of today’s spintronic devices, which are often considered for use in high-radiation environments such as satellites and other spacecraft. How such devices respond to large radiation doses therefore deserves a closer look.

“Only by seeing the spin waves directly could we understand which ingredients really govern these oscillators. Effects like grain boundaries and the Dzyaloshinskii–Moriya interaction turned out to be essential, not optional,” says Sebastian Wintz, who led the measurements at BESSY II.

 “We know that SHNOs can synchronize in very large networks. What this direct imaging gives us is a much firmer microscopic basis for understanding and controlling how spin waves set the coupling and phase relationships within those networks,” adds Johan Åkerman who is leading the team at University of Gothenburg.

Note:

The work was supported, among others, by the European Research Council (ERC Advanced Grant “TOPSPIN”, Grant No. 835068), the Knut and Alice Wallenberg Foundation through the Wallenberg Scholars programme (KAW 2022.0079), the Swedish Research Council through a Distinguished Professor Grant (Grant No. 2024-01943) and the Helmholtz Young Investigator Group programme.