Controlling Gilbert damping in a YIG film using nonlocal spin currents

We demonstrate the control of Gilbert damping in 65-nm-thick yttrium iron garnet (YIG) films using a spin-polarized current generated by a direct current through a nanocontact, spin filtered by a thin Co layer. The magnetodynamics of both the YIG and the Co layers can be excited by a pulse-modulated microwave current injected through the nanocontact and the response detected as a lock-in amplified voltage over the device. The spectra show three clear peaks, two associated with the ferromagnetic resonance (FMR) in each layer, and an additional Co mode with a higher wave vector proportional to the inverse of the nanocontact diameter. By varying the sign and magnitude of the direct nanocontact current, we can either increase or decrease the linewidth of the YIG FMR peak consistent with additional positive or negative damping being exerted by the nonlocal spin current injected into the YIG film. Our nanocontact approach thus offers an alternative route in the search for auto-oscillations in YIG films.

Figure 1

(a) Sketch of the spin valve stack consisting of YIG/Ag/Co and the circuit used for the ST-FMR measurements. (b) Variation of the FMR linewidth as a function of the frequency for YIG (blue) and YIG/Ag bilayer (red). (c) MR measurements as a function of the in-plane applied field angle $\theta$, measured under an external field $H_{\mathrm{ext}}=1000$ Oe.

Figure 2

ST-FMR results at $I=0$ mA. (a) A typical ST-FMR spectrum (black circles) measured at $f=13$ GHz showing peaks of YIG, Co, and a third mode. The solid red line is a Lorentzian fit. (b) Field dependence of the resonance frequency of YIG (blue dots), Co (red dots), and the third mode (violet dots) in the ST-FMR spectra. (c) Variation of the linewidth as a function of frequency for YIG and Co. Lines are linear fits.

Figure 3

ST-FMR results in the presence of a dc current. Variation in the linewidth of YIG as a function of the dc current at different frequencies for (a) device 1 and (b) device 2.

Figure 4

Angular dependence of the linewidth of YIG measured at $f=8$ GHz on device 1 (a) at $I=0$ mA and (b) as a function of the dc current.