Comparison of spin-orbit torques and spin pumping across NiFe/Pt and NiFe/Cu/Pt interfaces

We experimentally investigate spin-orbit torques and spin pumping in NiFe/Pt bilayers with direct and interrupted interfaces. The dampinglike and fieldlike torques are simultaneously measured with spin-torque ferromagnetic resonance tuned by a dc-bias current, whereas spin pumping is measured electrically through the inverse spin-Hall effect using a microwave cavity. Insertion of an atomically thin Cu dusting layer at the interface reduces the dampinglike torque, fieldlike torque, and spin pumping by nearly the same factor of $\approx 1.4$. This finding confirms that the observed spin-orbit torques predominantly arise from diffusive transport of spin current generated by the spin-Hall effect. We also find that spin-current scattering at the NiFe/Pt interface contributes to additional enhancement in magnetization damping that is distinct from spin pumping.

Figure 1

(a) Schematic of the dc-tuned ST-FMR setup and the symmetry of torques acting on the magnetization $\mathbf{m}$. Through spin-orbit effects, the charge current in the normal metal generates two torques in the ferromagnet: dampinglike torque (DLT) and fieldlike torque (FLT). (b) and (c) ST-FMR spectra of NiFe/Pt at (b) different frequencies and (c) dc-bias currents.

Figure 2

(a) Schematic of the dc spin-pumping (inverse spin-Hall effect) voltage measurement. (b) Representative dc voltage spectrum. The inverse spin-Hall signal $V_{ISH}$ dominates the anomalous Hall effect signal $V_{AHE}$.

Figure 3

(a) Resonance linewidth $W$ versus frequency $f$ at different dc-bias currents. (b) and (c) Resonance field $H_{FMR}$ versus frequency $f$ at different dc-bias currents for (b) NiFe/Pt and (c) NiFe/Cu/Pt.

Figure 4

(a) Resonance linewidth $W$ versus dc-bias current $I_{dc}$ at $f=5\mathrm{GHz}$. (b) Effective spin-Hall angle ${\theta }_{DL}$ calculated at several frequencies.

Figure 5

(a) and (b) dc voltage $V_{dc}$ spectra, dominated by the inverse spin-Hall voltage $V_{SH}$, measured around resonance in (a) NiFe/Pt and (b) NiFe/Cu/Pt.

Figure 6

Net current-induced effective field, derived from resonance field shift $\mathrm{\Delta }{H}_{FMR}$ normalized by the field direction angle $|sin\varphi |=1/\sqrt{2}$. The solid lines denote the estimated Oersted field.

Figure 7

(a) and (b) Effective spin-Hall angle ${\theta }_{SH,rf}^{\mathrm{eff}}$ extracted from ST-FMR line-shape analysis, disregarding the (a) fieldlike torque and taking into account the (b) fieldlike torque.

Figure 8

(a) Change in resonance linewidth $W$ versus dc-bias current $I_{dc}$ in Ta/NiFe at $f=6.5\mathrm{GHz}$. (b) Effective spin-Hall angle ${\theta }_{DL}$ calculated at several frequencies.

Figure 9

Empirical damping parameter ${\alpha }_{W/f}$ as a function of dc-bias current $I_{dc}$.