Spin Hall magnetoresistence and spin orbit torque efficiency in (Pt,Ta)/FeCoB bilayers

Abstract

The current induced spin-orbit-torque (SOT) [1-4] is the key physical parameter describing the generation of magnetization dynamics in bilayers composed by a heavy metal (HM) layer and a ferromagnetic (FM) layer. Such bilayers, especially when associated with a magnetic tunnel junction at the FM layer (then forming the so called 3-terminal device), are considered as promising elements to be applied in novel magnetic memories, microwave generators and even bio-inspired types of computing [1]. Therefore, the robust and accurate determination of the SOT in HM/FM bilayers is of high relevance for optimizing future applications.In HM/FM bilayers the SOT is mainly caused by the spin Hall effect (SHE) of the HM layer, by which a spin current is injected into the FM layer. 3-terminal devices with application as magnetic memories adopt mainly FM films with perpendicular magnetic anisotropy, while for sensor applications also the configuration with in-plane magnetic anisotropy is of interest. The determination and quantification of the SOT can be performed in both cases by different techniques. While in a 3-terminal device the magnetization dynamics of observed via the tunnel magnetoresistance effect, in HM/FM bilayers other methods can be employed, such as the SOT-ferromagnetic resonance (FMR), the harmonic Hall voltage measurement and the spin Hall magnetoresistance (SMR). The SMR is a straightforward method when the FM layer is an insulator, while with metallic FM, in which the testing electric current in not confined in the HM layer only, it is accompanied by additional effects, such as the anisotropic magnetoresistance or the anomalous Nernst effect [5]. Therefore, it is important to analyse the measurement results by accurate models taking into account all aspects. The dominant theoretical approach for SHE-SOT is the drift-diffusion model considering spin accumulation at the interface between the FM and HM by introducing a spin mixing conductance. However, the agreement with experiments may be poor if spin dissipation at the interface (spin memory loss, spin dependent scattering, etc.) is disregarded [2].In this paper we analyze transport measurements on FM/HM bilayers as a function of the HM thickness by employing a non-equilibrium thermodynamic approach to SHE-SOT [6] with the aim to reach a semi-quantitative agreement. The study was performed on HM/FM bilayers with HM=Pt (wedge 4-19nm), Ta (wedge 4-19nm), and FM=Fe60Co20B20 (2nm) prepared at Singulus Technologies AG. All samples were annealed for 2h at 310°C in a magnetic field of 1T in order to develop an in-plane magnetic anisotropy (y axis). For the transport measurements the films were patterned into µm-size Hall bars/crosses (along x axis) by optical lithography and ion beam etching with Ti/Au contact pads. Magneto-optic Kerr effect (MOKE) and SMR measurements were performed following the protocol of [5], while SOT-FMR and harmonic Hall voltage measurements and analysis were based on references [3,4].Fig.1 shows the comparison between MOKE loops and the SMR signal. The loop along the y axis is more rectangular and indicates the magnetic easy anisotropy axis. The SMR signal is much more pronounced with the field along the x axis, while it shows a small peak close to the coercivity when the field is along the y axis. With MOKE the anisotropy was determined and a dependence on HM thickness was observed.Fig.2 presents the summary of the SOT-FMR and harmonic Hall voltage measurements for HM/FM bilayers with different thickness of HM. FMR analysis enabled determination of effective magnetization and magnetization damping (α = 0.015 for Pt and 0.0075 for Ta), while anomalous Hall effect and harmonic Hall measurements resulted in SOT efficiencies dependence vs. HM thickness. While damping-like SOT efficiency for Ta reaches ξDL-Ta = -0.35 and ξDL-Pt = 0.15 for Pt, with high resistivity difference (ρTa > 200 µΩcm and 24 < ρPt < 52 µΩcm) taken into account, the spin Hall conductivity for Pt of σSH-Pt = 2.9 (105 S/m) exceeds the one for Ta σSH-Ta = 1.7 (105 S/m) and is among the highest values reported [8].The HM thickness dependence of the SMR is analyzed by considering both the finite conductance at the interface [7] (an approach valid for small thicknesses) and thermodynamic approach [6] (valid for larger thicknesses). The results are compared with the harmonic Hall voltage measurements the SOT and the SOT efficiency and discussed in relation to the key parameters as the spin Hall angle and the spin diffusion length.Acknowledgements: WS acknowledges National Science Centre Grant No. UMO-2015/17/D/ST3/00500, Poland. Microfabrication was performed at the Academic Centre for Materials and Nanotechnology of AGH. **