Optical Detection of Spin Transport in Nonmagnetic Metals

We determine the dynamic magnetization induced in nonmagnetic metal wedges composed of silver, copper, and platinum by means of Brillouin light scattering microscopy. The magnetization is transferred from a ferromagnetic ${\mathrm{Ni}}_{80}{\mathrm{Fe}}_{20}$ layer to the metal wedge via the spin pumping effect. The spin pumping efficiency can be controlled by adding an insulating interlayer between the magnetic and nonmagnetic layer. By comparing the experimental results to a dynamical macroscopic spin-transport model we determine the transverse relaxation time of the pumped spin current which is much smaller than the longitudinal relaxation time.

Figure 1

(a) Scheme of the sample layout and (b) SEM picture of the waveguide. A multilayer structure is prepared on top of a coplanar waveguide. A static magnetic field ${\mu }_0{H}_{\mathrm{static}}$ of 20 mT is applied parallel to the signal line and perpendicular to the dynamic magnetic field $h_{\mathrm{RF}}$, which is caused by an alternating microwave current flowing through the coplanar waveguide. The magnetization in the ${\mathrm{Ni}}_{80}{\mathrm{Fe}}_{20}$ layer is excited by $h_{\mathrm{RF}}$ and spins are pumped into the metal wedge. (c) MOKE hysteresis loop to determine the saturation field in the $x$ direction. (d) BLS spectrum taken on a pure ${\mathrm{Ni}}_{80}{\mathrm{Fe}}_{20}$ film at a static magnetic field of 20 mT. (e) BLS scans across the structure for the first two maxima of (d) at microwave frequencies of 5.5 GHz and 7.8 GHz. The profiles correspond to the first and the third laterally standing spin wave mode. The second mode is not excited.

Figure 2

BLS scan data of the silver (a), the copper (b), and the platinum (c) wedged sample. Each graph shows the measurement of the main sample with active spin pumping (dots) and the respective reference sample with blocked spin pumping (triangles). A difference between main and reference sample is only visible for silver and copper but not for platinum (see text). The error bars reflect the uncertainty in thickness determination by the mechanical profilometer. The solid and dashed lines are fits of the scan data according to Eq. (1). The evolution of the pure spin part of the BLS signal (dotted line) is derived from the fitting parameters. The inset in (a) shows schematically that the optical absorption length $z_{\mathrm{opt}}$ as well as the decay length of the transverse component $l_2$ contribute to the total BLS intensity probed by the laser.