Parametric spin pumping into an antiferromagnetic insulator

We report the observation of spin transport through antiferromagnetic insulators using spin pumping driven by short-wavelength magnons. The short-wavelength magnons are excited by parametric pumping in a $\mathrm{Pt}/\mathrm{NiO}/{\mathrm{Y}}_3{\mathrm{Fe}}_5{\mathrm{O}}_{12}$ trilayer. The parametrically excited magnons in the ${\mathrm{Y}}_3{\mathrm{Fe}}_5{\mathrm{O}}_{12}$ layer inject a spin current into the NiO layer, which is detected electrically using the inverse spin Hall effect in the Pt layer. We found that the spin-pumping efficiency of the parametrically excited magnons in the $\mathrm{Pt}/\mathrm{NiO}/{\mathrm{Y}}_3{\mathrm{Fe}}_5{\mathrm{O}}_{12}$ film is nearly an order of magnitude smaller than that in the $\mathrm{Pt}/{\mathrm{Y}}_3{\mathrm{Fe}}_5{\mathrm{O}}_{12}$ film. The suppression of the parametric spin-pumping efficiency induced by inserting the NiO layer is comparable to that of the FMR spin-pumping efficiency. We further found that the suppression of the spin-pumping efficiency due to the NiO insertion for the dipole-exchange magnons with the wave number of $k\sim {10}^3-{10}^4{\mathrm{cm}}^{-1}$ is more significant than that for the exchange magnons with $k\sim {10}^5{\mathrm{cm}}^{-1}$. This difference is attributed to two-magnon scattering and spin pinning due to antiferromagnetic grains in the NiO layer. These results provide insight into the understanding of the spin pumping into antiferromagnetic insulators.

Figure 1

(a) The magnon dispersion for a YIG film with a schematic of parametric excitation. $f_0$ is the applied microwave frequency. (b) A schematic illustration of the experimental setup, where $H$ is the external magnetic field and $h_{\text{RF}}$ is the microwave magnetic field. The microwave absorption intensity $P_{\text{abs}}$ in the sample was determined by measuring ${\mathrm{S}}_{21}$ using a vector network analyzer, $P_{\mathrm{abs}}=(\mathrm{\Delta }|S_{21}{|}^2/|S_{21}^0|)P_{\mathrm{in}}$, where $|S_{21}^0|$ represents the ratio between the incident and transmitted microwave power without magnetic resonance. $\mathrm{\Delta }|S_{21}{|}^2$ is the change ratio of the transmitted microwave power [18]. (c) Magnetic field $H$ dependence of the microwave absorption intensity $P_{\text{abs}}$ and ISHE voltage $V_{\text{ISHE}}$ measured for the Pt/NiO/YIG trilayer at the microwave frequency of $f_0=6.5$ GHz with the power of $P_{\mathrm{in}}=100\mathrm{mW}$ (the solid curve in blue) and $398\mathrm{mW}$ (the solid curve in red).

Figure 2

Magnetic field $H$ dependence of the microwave absorption intensity $P_{\text{abs}}$ and ISHE voltage $V_{\text{ISHE}}$ for the (a) Pt/YIG bilayer and (b) Pt/NiO/YIG trilayer measured at $f_0=6.5$ GHz with various excitation powers $P_{\mathrm{in}}$. In the present study, the microwave excitation power $P_{\text{in}}$ is defined as the power provided by the network analyzer. The microwave power applied to the sample, $P_{\text{app}}$, is suppressed from $P_{\text{in}}$: $P_{\text{app}}=\beta P_{\text{in}}$. Since the suppression factor $\beta$ is hard to control in our experimental setup, $\beta$ is different in different experiments, and therefore it is difficult to compare the data measured at the same $P_{\text{in}}$ for different samples. This is the reason why the $P_{\text{abs}}$ and $V_{\text{ISHE}}$ signals measured at different $P_{\text{in}}$ for the Pt/YIG bilayer and Pt/NiO/YIG trilayer are shown.

Figure 3

(a) The calculated ${\mu }_{0}H$ dependence of the wave number $k$, and the angle ${\theta }_k$ between $\mathbf{k}$ and $\mathbf{H}$ of the parametrically excited magnons at the threshold power under the 6.5 GHz microwave application. The calculated curves were obtained using the saturation magnetization ${\mu }_0{M}_{\text{s}}=198$ mT, $B/A=3,k_c=8.5\times {10}^7{\text{m}}^{-1}$, and the exchange stiffness constant $D=1.05\times {10}^{-5}{\mathrm{m}}^2/\mathrm{s}$ [26] (for details, see Ref. [27]). (b) ${\mu }_{0}H$ dependence of the magnon damping ${\eta }_{\mathbf{k}}$ at 6.5 GHz. ${\eta }_{\mathbf{k}}\left(5\text{mT}\right)$ is the damping at ${\mu }_{0}H=5$ mT.

Figure 4

Magnetic field $H$ dependence of the spin-pumping efficiency $\kappa /{\kappa }_{\max }$ for the Pt/YIG bilayer at $P_{\mathrm{in}}=365\mathrm{mW}$ (the solid curve in blue) and the Pt/NiO/YIG trilayer at $P_{\mathrm{in}}=600\mathrm{mW}$ (the solid curve in red) measured at $f_0=6.5$ GHz. Here, $\kappa =V_{\text{ISHE}}/P_{\text{abs}}$, and ${\kappa }_{\text{max}}$ is the maximum value of $\kappa$. To directly compare $\kappa$, we plot $\kappa$ measured at the microwave powers at which the $P_{\text{abs}}$ and $V_{\text{ISHE}}$ signals are observed in the field range of $30\text{mT}<{\mu }_{0}H<110\text{mT}$ in the Pt/NiO/YIG and Pt/YIG films. At $P_{\text{in}}=365$ mW for the Pt/YIG and 600 mW for the Pt/NiO/YIG, $H_{\text{jump}}\left(P_{\text{in}}\right)$, as well as the field range where the spin-pumping signal is observed, is the same between the Pt/NiO/YIG and Pt/YIG films. This allows us to directly compare the spin-pumping efficiency of exchange and dipole-exchange magnons in the two films. Although the frequency of the magnons excited in the present experiment is within the frequency range of the surface mode (see the inset), the transverse parametric pumping excites exchange magnons with large wave numbers, rather than the surface mode with relatively small wave numbers [27, 33].

Figure 5

A two-dimensional plot of the spin-pumping efficiency $\kappa$ as a function of ${\mu }_{0}H$ and $P_{\mathrm{in}}$ for the (a) Pt/YIG bilayer and (b) Pt/NiO/YIG trilayer. The dashed curve in red denotes the threshold power of the parametric excitation. The dashed curve in black is $H_{\mathrm{jump}}\left(P_{\mathrm{in}}\right)$, the field where the primary excited magnons switch between dipole-exchange and exchange magnons.