Microresonator-ferromagnetic resonance investigation of thermal spin-transfer torque in Co2FeAl/MgO/CoFeB magnetic tunnel junctions

Similar to electrical currents flowing through magnetic multilayers [1,2], thermal gradients applied across the barrier of a magnetic tunnel junction may induce pure spin currents and generate ‘thermal’ spin-transfer torques large enough to induce magnetization dynamics on the free layer [3, 4]. The relation of spin current, charge current and heat current was theoretically described by Bauer et al. using Onsager’s reciprocity rule [5]. According to Onsager’s law, spin currents can be produced by bias voltages or thermal gradients and investigated in terms of spin-Seebeck effect in magnetic multilayers.
First, Hatami et al. theoretically studied the spin-Seebeck effect in spin-valves and introduced the concept of thermal spin-transfer torques. They predicted that the thermally induced spin current creates an imbalance on the interface between non-magnetic and ferromagnetic layers due to collisions (electron-electron and electron-phonon interactions) [3]. Thermal spin-transfer torques were studied experimentally within asymmetric Co/Cu/Co nanowire spin-valves which exhibit switching field changes under varying a.c. currents causing Joule heating [6]. In magnetic tunnel junctions, it was theoretically predicted that temperature differences of around 10 K over an ultrathin barrier (1 nm) can create magnetization dynamics in Fe/MgO/Fe magnetic tunnel junctions [4]. The spin-Seebeck effect has been studied on CoFeB/MgO/CoFeB magnetic tunnel junctions using different heating methods such as Joule heating, heating with Peltier elements, as well as laser heating [8-14]. Recently, it was shown that using Co2FeAl as a reference layer improves tunneling magneto-Seebeck (TMS) in magnetic tunnel junctions [7].
Here, we describe a novel experimental approach and setup to observe effects of thermal gradients within magnetic tunnel junctions with Heusler compounds by using the microresonator ferromagnetic resonance (µR-FMR) method under laser heating. Initially, microresonators (shown in figure 1) were introduced by Narkowicz et al. for electron paramagnetic resonance (EPR) experiments to achieve optimal sensitivity for small objects [8]. Detecting the FMR signal of nano- to micron-sized samples in conventional cavities (cm3) is not possible, due to the too small ferromagnetic volume, and therefore low filling factor. A planar microresonator, by definition, is a two-dimensional structure, its diameter can be tailored to match the order of the sample’s size (shown as a black ellipse in the microresonator loop in figure 1). Two stubs are attached to the inductive loop. The capacitive radial stub in first approximation may be viewed as an element to tune the loop to the operation frequency, while the rectangular stub matches the structure to the 50 Ω impedance of the microstrip feedline.

Figure 1: Layout of a planar microresonator with simulated electric field distribution at the resonance frequency. The inset shows the current and magnetic field distribution (out-of-plane direction) in the loop containing a sample (black ellipse).
We investigated magnetic tunnel junctions (MTJs) fabricated out of Co2FeAl/MgO/CoFeB stacks. The sample and microresonator fabrication consist of multiple steps of lithography, ion etching and lift-off processes. The sample is finally patterned into a 6x9 µm2 elliptical shape using electron beam lithography (EBL) and ion beam etching is used to etch down the sample to the substrate. Microresonators are then fabricated around the sample using UV lithography. For laser heating, a continuous-wave (CW) laser at 532 nm wavelength and with tunable power up to 33 mW is focused on the sample.
“Hot-FMR” measurements were performed on unpatterned multilayers between 300 K and 450 K (figure 2) to understand the effect of global heating. It is clearly seen that the FMR signal of Co2FeAl exhibits a shift with increasing temperature. As seen in the inset graph, it is difficult to quantify the changes for the CoFeB signal, due to its small intensity. Subsequent, measurements in the presence of a thermal gradient were performed on 6x9 µm2 MTJs, integrated into microresonator loops with an inner diameter of 20 µm. The MTJs were submitted to laser irradiation, up to a maximum power of 33 mW. Unlike the Hot-FMR measurements, the resonance field and linewidth did not show clear changes with increasing laser power. The results suggest that the laser power is neither sufficient to induce magnetization dynamics via thermal gradients across the barrier, nor lead to significant changes of the magnetic parameters due to global heating of the sample.
Figure 2: FMR spectra of the extended films of Co2FeAl / MgO / CoFeB measured in the in-plane direction at different temperatures
As a conclusion, the effect of a global temperature change on the resonance frequency and linewidth of Co2FeAl was analyzed. With regards to the µR-FMR results, higher laser power is needed to induce magnetization dynamics. Moreover, the lateral heat transport might reduce the vertical thermal gradients, thus similar measurements on smaller structures are required.
This study was funded by the German Research Foundation (DFG) via priority program SpinCaT (SPP 1538). We thank H. Schultheiss for helping with the optical part of the experimental setup and S. Zhou for giving the access to the VSM setup.
[1] J.C. Slonczewski, J. Magn. Magn. Mater. 159, L1, (1996).
[2] L. Berger, Phys. Rev. B 54, 9353, (1996).
[3] M. Hatami, G.E.W. Bauer, Q. Zhang and P.J. Kelly, Phys. Rev. Lett. 99, 066603 (2007).
[4] X. Jia, K. Xia and G.E.W. Bauer, Phys. Rev. Lett. 107, 176603 (2011).
[5] G.E.W. Bauer, E. Saitoh and B.J. van Wees, Nature Mater. 11, 391, (2012).
[6] H. Yu, S. Granville, D.P. Yu and J.-Ph. Ansermet, Phys. Rev. Lett. 104, 146601 (2010).
[7] A. Boehnke, U. Martens, C. Sterwerf, A. Niesen, T. Huebner, M. von der Ehe, M. Meinert, T. Kuschel, A. Thomas, C. Heiliger, M. Münzenberg M, Nature Communications 8,(1),1626, (2017).
[8] R. Narkowicz, D. Suter and R. Stonies, J. Magn. Reson.175, 275 (2005).