Control of spin current by a magnetic YIG substrate in NiFe/Al nonlocal spin valves

Rapid Communication

Control of spin current by a magnetic YIG substrate in NiFe/Al nonlocal spin valves

F. K. Dejene, N. Vlietstra, D. Luc, X. Waintal, J. Ben Youssef, and B. J. van Wees

Phys. Rev. B 91, 100404(R) – Published 24 March 2015

Abstract

We study the effect of a magnetic insulator [yttrium iron garnet (YIG)] substrate on the spin-transport properties of ${Ni}_{80} {Fe}_{20} / Al$ nonlocal spin valve (NLSV) devices. The NLSV signal on the YIG substrate is about two to three times lower than that on a nonmagnetic ${SiO}_{2}$ substrate, indicating that a significant fraction of the spin current is absorbed at the Al/YIG interface. By measuring the NLSV signal for varying injector-to-detector distances and using a three-dimensional spin-transport model that takes spin-current absorption at the Al/YIG interface into account, we obtain an effective spin-mixing conductance $G_{↑ ↓} ≃ 5 - 8 \times 10^{13} Ω^{- 1} m^{- 2}$ . We also observe a small, but clear, modulation of the NLSV signal when rotating the YIG magnetization direction with respect to the fixed spin polarization of the spin accumulation in the Al. Spin relaxation due to thermal magnons or roughness of the YIG surface may be responsible for the observed small modulation of the NLSV signal.

Received 19 December 2014
Revised 11 February 2015

DOI:https://doi.org/10.1103/PhysRevB.91.100404

Authors & Affiliations

F. K. Dejene^1,*, N. Vlietstra¹, D. Luc², X. Waintal², J. Ben Youssef³, and B. J. van Wees¹

¹Physics of Nanodevices, Zernike Institute for Advanced Materials, University of Groningen, NL-9747 AG Groningen, The Netherlands
²CEA-INAC/UJF Grenoble 1, SPSMS UMR-E 9001, F-38054 Grenoble, France
³Laboratoire de Magnetisme de Bretagne CNRS, Université de Bretagne Occidentale, 6 Avenue Le Gorgeu, F-29285 Brest, France

^*f.k.dejene@gmail.com

Article Text (Subscription Required)

Click to Expand

Supplemental Material (Subscription Required)

Click to Expand

References (Subscription Required)

Click to Expand

Issue

Vol. 91, Iss. 10 — 1 March 2015

Reuse & Permissions

Access Options

Author publication services for translation and copyediting assistance advertisement

Images

Figure 1
Concept of the experiment for $\hat{m} ∥ μ_{s}$ . (a) A charge current through the $F_{1}$ /Al interface creates a spin accumulation $μ_{s}$ in the Al. The diffusion of $μ_{s}$ to the $F_{2}$ /Al interface is affected by spin-flip relaxation at the Al/YIG interface. Scattering of a spin up electron ( $s = ℏ / 2$ ) into a spin down electron ( $s = - ℏ / 2$ ) is accompanied by magnon emission ( $s = ℏ$ ), creating a spin current into the YIG that is minimum (maximum) when ${\hat{μ}}_{s}$ is parallel (perpendicular) to the magnetization of the YIG. (b) Profile of $μ_{s}$ along the Al strip on a ${SiO}_{2}$ (red) and YIG (blue) substrate. The spin accumulation at the $F_{2}$ /Al is lower for the YIG substrate compared to that on ${SiO}_{2}$ .
Reuse & Permissions
Figure 2
(a) Scanning electron microscope image of the measured type-A device. Two Py wires (indicated by green arrows) are connected by an Al cross. A charge current $I$ from contact 1 to 2 creates a spin accumulation at the $F_{1}$ /Al interface that is detected as a nonlocal spin voltage $V_{nl}$ using contacts 3 and 4. (b) The NLSV resistance $R_{nl} = V_{nl} / I$ for representative YIG (blue) and ${SiO}_{2}$ (red and orange) NLSV samples. For comparison, a constant background resistance has been subtracted from each measurement. (c) Dependence of the NLSV signal on the spacing $d$ between the injecting and detecting ferromagnetic wires together with calculated spin signal values using a 1D (dashed lines) and 3D (solid lines) spin-transport model. For each distance $d$ between the injector and detector, several devices were measured, with the error bars indicating the spread in the measured signal.
Reuse & Permissions
Figure 3
(a) Nonlocal spin valve resistance $R_{nl}$ of a type-B device with $d = 500$ nm between injecting and detecting Py wires and $t_{Al} = 130$ nm. A constant background resistance of 117 $m Ω$ was subtracted from the original data. (b) Angular dependence of the NLSV signal in the parallel and antiparallel configurations. The AP curve is an average of ten measurements and that of the P state is a single scan. Both resistance states exhibit a $cos (2 α)$ dependence on the angle between $\hat{m}$ and $μ_{s}$ . The black solid lines are calculated using the 3D-FEM model for $G_{r} = 1 \times 10^{13} Ω^{- 1} m^{- 2}$ that show a percentage modulation of only 12% corresponding to the green curve in (c) $δ R_{SV} / R_{SV}$ . The angular dependent measurement in (b) is from a device for which a complete set of measurements was performed. A spin valve measurement as in (a) was also performed for another device with $d = 300$ nm.
Reuse & Permissions
Figure 4
Calculated NLSV signals showing the (a) $\hat{x}$ component and (b) $\hat{z}$ component of the NLSV signal $R_{nl}$ in the parallel (red) and antiparallel (blue) magnetization configurations of the injector and detector ferromagnetic contacts for $G_{r} = 1 \times 10^{13} Ω^{- 1} m^{- 2}$ and $G_{i} = 0.1 G_{r}$ . Even if the injected spin accumulation is polarized along the magnetization direction of the injecting electrode $F_{1}$ , its interaction with the magnons via the spin-mixing conductance induces these spin accumulation components.
Reuse & Permissions

Physical Review B

covering condensed matter and materials physics