Interaction between ferromagnetic resonance and spin currents in nanostructures

We report here on the detection of ferromagnetic resonance in a permalloy nanostructure using the inverse spin-Hall effect in a platinum layer in contact. The tiny spin currents driven out of the precessing magnetization of a micron square sized structure generate an electrical voltage in the platinum layer because of spin-orbit scattering. We have achieved isolating this signal from other resistive contributions and show that it dominates in certain field geometries. This detection technique can therefore be applied in ferromagnetic nanostructured materials under certain experimental precautions. We also have been able to modify the damping of our Py nanostructures by injecting spin polarized currents using the spin-Hall effect in Pt.

Figure 1

SEM image of the device consisting of a 15-nm-thick Pt stripe including four contacts with two 20-nm-thick Py structures deposited on top. These are composed of a 2-$\mu$m-diameter circle and an ellipse with a 2-$\mu$m-long axis parallel to the stripe ($x$) and a 600-nm-short axis in the transverse ($y$) direction and separated by 800 nm. The short end of a coplanar strip waveguide is adjacent to the bilayers at respectively 1.1 $\mu$m and 1.4 $\mu$m from the ellipse and the circle centers. It is made in a Ti(5 nm)/Au(140 nm) bilayer by optical lithography, e-beam deposition, and lift-off techniques, in a shape optimized for our kapton substrate.

Figure 2

Schematics of the inverse spin-Hall effect signal. A spin current is generated in the nanostructure because of magnetization damping. It is converted into an electrical signal in the Pt by spin-orbit scattering and can be measured as a dc voltage peak at resonance.

Figure 3

(a) Field dependence of the measured voltage for the ellipse ($V_3-V_2$) at two different frequencies. The applied angle between H${}_0$ and the current lines (i.e., the main direction of the ellipse or the $x$ axis) is 60°, thus setting the magnetization at an angle close to 45°. (b) The odd (solid lines) and even (dashed lines) contributions are extracted by adding/subtracting the positive and negative field curves (11 GHz and 15 GHz curves are shifted for clarity).

Figure 4

Transverse field dependence of the measured voltage for the ellipse at two frequencies: 11 GHz and 18 GHz. The solid and dashed curves correspond respectively to the odd and even contributions. All peaks on the odd parts are negative and have a similar shape. They are due to the ISHE, whereas the even contributions stem from a residual AMR effect (the AMR contribution to the odd part is estimated below the noise level).

Figure 5

Transverse field dependence of the measured voltage for the ellipse in the Au/Py sample at a frequency of 15 GHz (on the same scale as that of Fig. 4). The solid curve is the odd contribution (from ISHE) and the dashed curve the even part (AMR contribution).

Figure 6

Effect of a dc current in the Pt stripe on the resonance shape of the elliptical Py nanostructure in a transverse field. The orange curve is the raw data measured when a positive current density of $6.{10}^6{\mathrm{A}/\mathrm{cm}}^2$ flows in the Pt layer, and the blue curve shows the subtraction of positive to negative current data. The amplification/reduction of resonance peak amplitude for negative/positive fields is consistent with a 4% change of the damping parameter due to spin injection.