Observation of magnon-mediated electric current drag at room temperature

Spin-based electronic devices such as magnetic memory and spin logic rely on spin information transport. Conduction electrons, due to their intrinsic spin angular momentum, become an obvious choice for spin information carriers. Here, we experimentally demonstrate that magnons, quasiparticles representing low-energy excitations of ferromagnetic materials, can serve as effective spin information carriers as well. Specifically, we consider two nonmagnetic heavy metals (HMs) that are separated by an electric leak-free ferrimagnetic insulator. When an electric current is applied in one of the HM layers, magnons in the ferrimagnetic insulator are excited and become an effective medium to couple the spin currents in two HMs. As a result, the charge/spin current in one HM layer can drag a charge/spin current in the other HM layer. This work provides a route for spin-based electronic devices where the spin transport is carried by quasiparticles other than electrons.

Figure 1

(a) A schematic illustration of the sample structure and the measurement method. An electric current $I_{\mathrm{b}}$ is applied along the $x$ axis in the bottom Pt layer, the magnetic field $H$ is applied along the $y$ axis, and the voltage $V_{\mathrm{t}}$ is measured along the $x$ axis in the top Pt layer. (b) The proposed spin transport processes in the Pt/YIG/Pt structure. The spin current is generated by the charge current in the bottom Pt layer via the spin Hall effect (SHE), resulting in a spin accumulation at the bottom Pt/YIG interface. Subsequently, the magnon current excited by the spin accumulation diffuses from the bottom to the top interface, and then converts back to the electron spin current in the top Pt layer. Finally, the spin current in the top Pt is detected by the voltage via the inverse spin Hall effect (ISHE). (c) and (d) The cross-sectional transmission electron microscopy (TEM) image of the Pt(10)/YIG(100)/Pt(10) (thickness in nanometers) multilayer and the cross-sectional high resolution TEM (HRTEM) image of the Pt/YIG interface, respectively.

Figure 2

(a) $V_{\mathrm{t}}/R_{\mathrm{t}}-I_{\mathrm{b}}$ characteristics measured in Pt/YIG/Pt, Cu/YIG/Pt, and Pt/YIG/Cu samples. A 1 kOe magnetic field is applied along the $y$ direction during the measurement. (b) Temperature-dependent $V_{\mathrm{t}}/R_{\mathrm{t}}-I_{\mathrm{b}}$ curves of the Pt/YIG/Pt sample. A 1 kOe magnetic field is applied along the $y$ direction during the measurement. (c) The magnetic field dependence of the magnon drag voltage in Pt/YIG/Pt and the spin Seebeck voltage in Cu/YIG/Pt, which are in accordance with the magnetic properties of YIG as shown in the $M-H$ loop.

Figure 3

Top: The definition of the angular $\alpha , \beta$, and $\gamma$ in this figure. A 3 kOe constant magnitude of the magnetic field is applied to saturate the magnetic moment $\mathit{M}$ during the sample rotation. (a) The magnetization direction dependence of the magnon drag voltage in Pt/YIG/Pt. A 10 mA electric current is applied in the bottom Pt layer, and the voltage is measured in the top Pt layer. (b) The magnetization direction dependence of the spin Seebeck voltage in Cu/YIG/Pt. The electric current applied in the bottom Cu is 20 mA, and the voltage is measured in the top Pt layer.

Figure 4

(a) $V_{\mathrm{t}}/R_{\mathrm{t}}-I_{\mathrm{b}}$ curves of Pt(10)/YIG(t)/Pt(10) samples with different thicknesses ($t=40$, 60, 80, 100 nm) of YIG. A 1 kOe magnetic field is applied along the $y$ direction during the measurement. (b) The YIG thickness dependence of the electrical drag coefficient $a$, while the open squares show the measured data, and the red line shows the fitting curve $a=a_0{e}^{-t/l_{\mathrm{m}}}$. (c) The magnetization direction dependence of the magnon drag voltage in Pt/YIG/Pt samples with different thicknesses of YIG. A 10 mA electric current is applied in the bottom Pt layer, and the voltage is measured in the top Pt layer. A 3 kOe constant magnitude of the magnetic field is applied to saturate the magnetic moment $\mathit{M}$ during the sample rotation.