† Corresponding author. E-mail:
‡ Corresponding author. E-mail:
Project supported by the National Basic Research Program of China (Grant No. 2015CB921401), the National Natural Science Foundation of China (Grant Nos. 51331002, 51371027, 51431009, 51471183, and 11274371), the National Instrumentation Program of China (Grant No. 2012YQ120048), and the Instrument Development Program of Chinese Academy of Sciences (Grant No. YZ201345).
The interplay of the Rashba effect and the spin Hall effect originating from current induced spin–orbit coupling was investigated in the as-deposited and annealed Pt/Co/MgO stacks with perpendicular magnetic anisotropy. The above two effects were analyzed based on Hall measurements under external magnetic fields longitudinal and vertical to dc current, respectively. The coercive field as a function of dc current in vertical mode with only the Rashba effect involved decreases due to thermal annealing. Meanwhile, spin orbit torques calculated from Hall resistance with only the spin Hall effect involved in the longitudinal mode decrease in the annealed sample. The experimental results prove that the bottom Pt/Co interface rather than the Co/MgO top one plays a more critical role in both Rashba effect and spin Hall effect.
Spin transfer torque generated by spin-polarized current provides another powerful method to manipulate the magnetization direction compared with the traditional way due to Oersted fields.[1–8] More recently, torque related to the Rashba effect and spin Hall effect (SHE), so called spin–orbit torque herein, originating from the spin–orbit coupling (SOC) induced by dc electric current has been demonstrated to realize the magnetization switch.[9–14] It offers for the current-in-plane device more advantages in applications with lower power consumption and higher reliability such as the three-terminal magnetic tunnel junctions with the separated read/write wires.[12] The entangled correlation due to the Rashba effect and SHE is not fully understood although much progress has been made in heavy metal (HM)/ferromagnetic (FM)/oxide heterostructures with structural inversion asymmetry (SIA), such as Pt/Co/MgO,[15] Pt/Co/AlOx,[9,10,13] Ta/CoFeB/MgO,[16–18] and W/CoFeB/MgO.[19] The realization including the experimental setup and novel multilayers to separate the measured parameters from the Rashba effect and the SHE is considerably critical. In this letter, several methods have been carried out to solve this issue, including direction-variable Hall measurements, tunable interfacial structures, together with thermal annealing. With respect to the Pt/Co/MgO sample in vertical mode with only the Rashba effect involved, the coercivity (Hc) as a function of dc current decreases after it is annealed. For the longitudinal mode with only the spin Hall effect involved, calculated SHE torques from the Hall resistance curve show a clear drop after annealing as well. The results from samples with tunable Pt/Co structures prove that the bottom Pt/Co interface rather than the Co/MgO top one has a critical influence here, which can give a reasonable explanation of the reduced effect due to annealing.
Three multilayered structures consisting of (from substrate) Pt(3)/Co(0.9)/MgO(2) (labeled as S-1), Pt(2)/Co(0.45)/Pt(1)/Co(0.45)/MgO(2) (S-2), and Pt(2)/[Co(0.3)/Pt(0.5)]2/Co(0.3)/MgO(2) (S-3) (all thickness in nm) were prepared on Si wafer by magnetron sputtering. A 2-nm-thick Pt layer was deposited on the top as the capping layer to avoid degradation. The films were patterned into 10-μm-wide Hall bars using photolithography and Ar-ion etching for the transport measurements. Thermal annealing of the patterned samples were carried out in a vacuum furnace (less than 5 × 10−7 Torr) for 40 minutes at 150 °C and 300 °C, respectively.
A Keithley 2400 SourceMeter as dc current injection and a 2182A NanovoltMeter for Hall voltage were used for transport measurement. The temperature dependence of resistance was measured by a Quantum Design Physical Property Measurement System (PPMS). The details about the sample preparation, microfabrication, annealing process, and transport measurements can be found in our previous work.[20,21]
The experimental setup is illustrated in Fig.
We focus on the current dependence of coercivity under vertical mode for deeper understanding. Figure
The HM/FM interfacial structures have been proven to be considerably critical for the Rashba effect in multilayers in former studies.[23–25] In order to manipulate the interfacial structures, two samples were deposited, S-2: Pt(2)/Co(0.45)/Pt(1)/Co(0.45)/MgO(2), and S-3: Pt(2)/[Co(0.3)/Pt(0.5)]2/Co(0.3)/MgO(2) (in nm). It is necessary to emphasize that the total thickness of Pt and Co is the same for three samples with different layer structures. In other words, the repeat number of Pt/Co is one, two, and three, respectively. Figure
With respect to SHE in Pt/Co/MgO, the longitudinal mode is used for the investigation, where RHall is measured by applying a positive or negative current. The setup is shown by the inset of Fig.
As stated above, the Pt/Co interface plays a dominant role in the transport behavior, where the Rashba effect and SHE are entangled due to spin orbit coupling. However, this interfacial role is quite different for the Rashba effect and SHE, which is investigated for the vertical and longitudinal modes, respectively. More exactly, for the vertical mode with only the Rashba effect involved, the measured AHE Hall resistance is closely related to PMA. It means the magnetic structures at the Pt/Co interface together with the perpendicular Co layer will determine the measured values due to the Rashba effect. Regarding the longitudinal mode with only SHE involved, the Hall resistance is measured. In this case, the Pt/Co interface plays a more critical role in the injection of spin current from Pt layer to Co layer. That is to say, the interfacial states, such as intermixing, roughness, and defects, are quite important for the SHE in longitudinal mode, rather than the magnetic states such as magnetized Pt partial layers for the Rashba effect in vertical mode.
Recently, the experimental evidence for the induced magnetization of HM films adjacent to a magnetic layer due to a strong proximity effect has been provided by x-ray magnetic circular dichroism measurements.[26] For the Pt/Co bilayer, it is reasonable to assume a magnetic interfacial layer consisting of magnetic Pt atoms and their adjacent Co atoms. Therefore, for simplicity, the Pt/Co bilayer can be considered as three stacks, including traditional Pt and layers separated by a magnetic interface. For the vertical mode with only the Rashba effect involved, the value of RHall shown in Fig.
Based on the above discussion about the interplay of the Rashba effect and SHE both due to spin orbit coupling, the Pt/Co interfacial structure plays an important role. However, the roles of the interface take place in different ways. For the vertical mode with only the Rashba effect involved, the magnetic structures at the interface are quite critical. Meanwhile, for the longitudinal mode with only SHE involved, the morphological states such as intermixing, roughness, and defects are more important.
In summary, the Rashba effect and SHE originating from current-induced spin orbit coupling were investigated in as-deposited and annealed Pt/Co/MgO with PMA based on transport measurements under direction-variable external magnetic field. According to the results of vertical mode Hall measurement, the coercivity as a function of dc current related to the Rashba effect decreases in annealed sample S-1: Pt(3)/Co(0.9)/MgO(2). This can be attributed to variation of the magnetic structure of the Pt/Co bilayer due to thermal annealing. Meanwhile, the results from samples with tunable interfaces (S-2 and S-3) can also prove the rationality of this interpretation. With respect to the SHE studied in longitudinal-mode Hall measurement, the calculated spin–orbit torque due to the SHE shows a clear drop after annealing, especially at 300 °C. This means that the morphological states such as intermixing, roughness, and defects are more important for the spin current injection.
1 | |
2 | |
3 | |
4 | |
5 | |
6 | |
7 | |
8 | |
9 | |
10 | |
11 | |
12 | |
13 | |
14 | |
15 | |
16 | |
17 | |
18 | |
19 | |
20 | |
21 | |
22 | |
23 | |
24 | |
25 | |
26 | |
27 |