記住我
Here, we report the current-induced magnetization switching in the interlayer exchange-coupled Co/Ir/Co system sandwiched by the Pt layers. The switching processes for different magnetization alignments, i.e., ferromagnetically coupled or antiferromagnetically coupled configuration, were investigated by changing Ir layer thickness (tIr) through the electrical measurement and the magnetic domain imaging. The local picture for the switching process triggered by the dual spin–orbit torque was also clarified for both antiferromagnetically and ferromagnetically coupled cases by the numerical calculation based on the macrospin model.
Thin films consisting of Pt (2 nm)/Co (0.65 nm)/Ir (tIr = 0.45, 0.50, 0.80, or 1.30 nm)/Co (0.9 nm)/Pt (2 nm) were deposited on a thermally oxidized Si substrate using magnetron sputtering at room temperature with a base pressure below 6.0 × 10−6 Pa. Ta was deposited as 1 nm thick buffer and capping layers before and after deposition of the Pt/Co/Ir/Co/Pt, respectively. We expected the interlayer exchange coupling in the Co/Ir/Co tri-layer and the perpendicular magnetic anisotropy induced by Pt layers on magnetizations in Co layers. Magnetization curves are shown in Figs. 1(a)–1(d), which were measured by the vibrating sample magnetometer at room temperature. At tIr = 0.45 nm, both the out-of-plane and the in-plane magnetization curves exhibited the remanent magnetization, indicating that the magnitude of perpendicular magnetic anisotropy is insufficient for achieving the fully perpendicularly magnetized state. This is possibly due to the non-perfect film growth of the top Co layer on the thin Ir layer. However, at tIr ≥ 0.50 nm, the perpendicular magnetic anisotropy overcomes the out-of-plane demagnetizing field, resulting in the perpendicularly magnetized state. Low (large) remanent magnetization and large (small) saturation field at tIr = 0.50 and 1.30 nm (at tIr = 0.80 nm) indicate the antiferromagnetic coupling (ferromagnetic coupling) in the Co/Ir/Co structure, being consistent with our previous report.4949. H. Masuda, T. Seki, Y. Yamane, R. Modak, K. Uchida, J. Ieda, Y.-C. Lau, S. Fukami, and K. Takanashi, Phys. Rev. Appl. 17, 054036 (2022). https://doi.org/10.1103/PhysRevApplied.17.054036For the current-induced magnetization switching measurement, thin films were patterned into Hall-bar-shaped devices with 5 μm wire width using photolithography and Ar-ion milling. Each device has the electrical contact pads of Cr (20 nm)/Au (200 nm) deposited by ion-beam sputtering. Figure 2(a) schematically illustrates the damping-like spin–orbit torque in the antiferromagnetically and the ferromagnetically coupled Pt/Co/Ir/Co/Pt structures with magnetization m for each Co layer. The application of a charge current with a current density jc in the x direction induces the spin current density js in the z direction in Pt layers due to the spin Hall effect. The ratio of conversion is expressed as the spin Hall angle: θSH = (2e/ℏ) js/jc, where e and ℏ denote the elementary charge and the Dirac's constant, respectively. The spin moments σ accumulate at the interfaces pointing in the ± y direction, generating the damping-like spin–orbit torque (τDL) proportional to m × (m × σ).5959. L. Zhu, D. C. Ralph, and R. A. Buhrman, Appl. Phys. Rev. 8, 031308 (2021). https://doi.org/10.1063/5.0059171 The effective field of damping-like spin–orbit torque (BDL) then works parallel to m × σ on magnetizations. For the antiferromagnetically coupled system, BDL acting on both top and bottom Co layers point in the –x direction. On the other hand, for the case of the ferromagnetically coupled system, BDL acts in the –x direction for the top side and in the +x direction for the bottom side. One may anticipate that BDL for the top Co and BDL for the bottom Co cancel out in the case that two Co layers are strongly coupled, which is regarded as the condition that dual spin–orbit torque with opposite signs simultaneously acts on the single ferromagnetic layer. As it will be shown, however, the ferromagnetically coupled system having a moderate coupling strength exhibits a different behavior from that of the single ferromagnetic layer.The measurement setup for the four-probe method is depicted in Fig. 2(b). A DC charge current (Idc) with pulse width of ∼100 ms was applied in the x direction using a Keithley 2400 DC source meter. The Hall resistance (Rxy) was detected by applying an AC charge current of 20 μA with a frequency of 9997 Hz using a lock-in amplifier SR830 after each application of Idc. An external magnetic field of 80 mT was also applied simultaneously in the x direction to assist the magnetization switching. The values of Rxy as a function of Idc for the devices with tIr = 0.45, 0.50, 0.80, and 1.30 nm are shown in Figs. 2(c)–2(f), respectively. The Rxy–Idc curves with hysteresis were observed for all the samples, suggesting that magnetization switching is induced by the Idc application. The Rxy–Idc curves for tIr = 0.45 and 0.80 nm show the gradual change in Rxy with positive Idc in the transition from a low resistance state to a high resistance state. Compared with these Rxy–Idc curves, Rxy was sharply changed at Idc ∼16.5 and −23.6 mA (12.6 and −17.4 mA) for the antiferromagnetically coupled sample with tIr = 0.50 nm (1.30 nm). The difference in Rxy–Idc curves indicates that the switching process depends on the magnetization alignment via interlayer exchange coupling. It is noted that the asymmetry in the positive and negative switching currents means that the observed Rxy–Idc curves are minor loops. This is attributed to the domain-wall trapping at the intersected region between the channel and the branches of Hall-bar-shaped device which will be discussed in a later paragraph.In order to understand the switching process in detail, the domain structures were visualized using a commercial Evico magnetics GmbH Kerr microscopy with an in-plane magnetic field coil. Figures 3(a)–3(d) display the domain structures under μ0Hx = 80 mT, visualized after the application of Idc for tIr = 0.45, 0.50, 0.80, and 1.30 nm, respectively. First, Idc > 22 mA was applied in the +x direction, which led to the saturated magnetic state. The images at Idc > 22 mA were used as the background signal and subtracted from the obtained images. For all the samples, the black and white contrast changes were observed with varying Idc. For tIr = 0.45 nm, the complicated white regions were observed at Idc = –5.3 mA. The black regions also appeared at Idc ≤ –14 mA. The complicated and intermixed contrasts for tIr = 0.45 nm are attributed to the existence of not only out-of-plane magnetized component but in-plane magnetized component as well, which is in agreement with the magnetization curves. For the antiferromagnetically coupled samples with tIr = 0.50 and 1.30 nm, the abrupt expansion of black region was observed, being consistent with the Rxy−Idc curves showing the sharp change in Rxy. On the other hand, the ferromagnetically coupled sample with tIr = 0.80 nm showed a sparse nucleation at Idc = –9.6 mA, and the switched region gradually increased with increasing |Idc|. This is an apparent difference in the magnetic domain structures formed during the spin–orbit torque switching between the antiferromagnetic coupling case and the ferromagnetic coupling case. It is worth noting that the domain-wall trappings were observed in the intersected regions between the channel and the branches of Hall device for all the samples. This behavior is reminiscent of the magnetic ratchet effect found in the device with asymmetric configuration,6060. A. Himeno, S. Kasai, and T. Ono, Appl. Phys. Lett. 87, 243108 (2005). https://doi.org/10.1063/1.2140884 which would induce the asymmetric hysteresis in the minor loop as observed in Fig. 2. One may anticipate the formation of skyrmion bubbles in the Pt/Co/Ir structure,6161. C. Moreau-Luchaire, C. Moutafis, N. Reyren, J. Sampaio, C. A. F. Vaz, N. Van Horne, K. Bouzehouane, K. Garcia, C. Deranlot, P. Warnicke, P. Wohlhüter, J.-M. George, M. Weigand, J. Raabe, V. Cros, and A. Fert, Nat. Nanotechnol. 11, 444–448 (2016). https://doi.org/10.1038/nnano.2015.313 but we did not observe it possibly due to the decrease in the interfacial Dzyaloshinskii–Moriya interaction3131. Y. Ishikuro, M. Kawaguchi, N. Kato, Y.-C. Lau, and M. Hayashi, Phys. Rev. B 99, 134421 (2019). https://doi.org/10.1103/PhysRevB.99.134421 and the relatively large perpendicular magnetic anisotropy in the present Pt/Co/Ir/Co/Pt films.Considering that the magnetization switching occurs through the nucleation of a reversed domain and the subsequent domain expansion as visualized in Fig. 3, the nucleation of a reversed domain is triggered by the spin–orbit torque, and we may assume that the spin–orbit torque locally acts on the Co magnetization for some parts of the device. To understand the local picture of current-induced magnetization switching in Pt/Co/Ir/Co/Pt systems, we numerically simulated the spin–orbit torque switching dynamics. Let us consider the magnetic multilayer systems consisting of two coupled ferromagnetic layers, FM1 and FM2, with a common saturation magnetization (Ms) and thicknesses tm1 and tm2, sandwiched by the spin Hall layers. The surface magnetic energy density (E) was assumed to be E=JIECm1·m2−Ktm1m1z2+tm2m2z2−μ0Mshx·tm1m1+tm2m2,(1)where JIEC is the interlayer exchange coupling strength, K is the perpendicular magnetic anisotropy constant, m1 (m2) is the unit vector representing the magnetic moment in FM1 (FM2) layer, and hx is the external magnetic field in the x direction. Here, we assumed that JIEC > 0 (mμ (μ = 1 and 2) were described by the coupled Landau–Lifshitz–Gilbert equations as follows: ∂mμ∂t=−γmμ×hμ+αmμ×∂mμ∂t−γmμ×mμ×σμ,(2)where γ is the gyromagnetic ratio, and α is the damping constant. The effective magnetic fields hμ were defined by hμ=−1μ0MstmμδE∂mμ,(3)and the spin–orbit torques were characterized by σμ=±ℏθSH2eμ0Mstmμjcy,(4)where the upper (lower) sign corresponds to μ = 1 (2). Figures 4(a) and 4(b) show the calculated switching dynamics for the antiferromagnetically and the ferromagnetically coupled systems, respectively, with parameters of μ0Hx = –10 mT, tm1 = 0.8 nm, tm2 = 0.5 nm, |JIEC| = 1.5×10−3 J m−2,4949. H. Masuda, T. Seki, Y. Yamane, R. Modak, K. Uchida, J. Ieda, Y.-C. Lau, S. Fukami, and K. Takanashi, Phys. Rev. Appl. 17, 054036 (2022). https://doi.org/10.1103/PhysRevApplied.17.054036 K = 5.7 × 105 J m−3,4949. H. Masuda, T. Seki, Y. Yamane, R. Modak, K. Uchida, J. Ieda, Y.-C. Lau, S. Fukami, and K. Takanashi, Phys. Rev. Appl. 17, 054036 (2022). https://doi.org/10.1103/PhysRevApplied.17.054036 Ms = 1.1 × 106 A m−1, α = 0.03,6262. N. Fujita, N. Inaba, F. Kirino, S. Igarashi, K. Koike, and H. Kato, J. Magn. Magn. Mater. 320, 3019 (2008). https://doi.org/10.1016/j.jmmm.2008.08.012 and θSH = 0.15.6363. M.-H. Nguyen, D. C. Ralph, and R. A. Buhrman, Phys. Rev. Lett. 116, 126601 (2016). https://doi.org/10.1103/PhysRevLett.116.126601 When a charge current with jc = 3.1 × 1013 A m−2 was applied to the antiferromagnetically coupled system, m1 and m2 were first tilted toward the –x direction and then almost aligned antiparallelly along the y axis. Note that m1 and m2 acquire the small z component opposite to their initial directions. The turning-off of jc then resulted in the full switching of the antiferromagnetic order. This switching scheme is consistent with a CoFeB/Ta/CoFeB system.2222. G. Y. Shi, C. H. Wan, Y. S. Chang, F. Li, X. J. Zhou, P. X. Zhang, J. W. Cai, X. F. Han, F. Pan, and C. Song, Phys. Rev. B 95, 104435 (2017). https://doi.org/10.1103/PhysRevB.95.104435 For the case of the ferromagnetically coupled system, on the other hand, m1 and m2 first instantaneously exhibited the tilting toward the –x and +x directions, respectively, when the same charge current as before was applied. This was followed by the antiparallel alignment along the y axis, similar to the antiferromagnetically coupled case. Note that |m1z| > |m2z| in this intermediate state (m1z ≤ 0 and m2z ≥ 0) because |σ1| ≤ |σ2| due to tm1 ≥ tm2 [Inset of the far right figure in Fig. 4(b)]. For the ferromagnetically coupled case, the fact that σ1 and σ2 are in the opposite directions may appear to be unfavorable, since the spin–orbit torque effect seems to be canceled between the two layers, as has been discussed above. However, because the interlayer exchange coupling is not very strong, the spin–orbit torque can overcome the former when jc is sufficiently large and lead temporarily to the “antiferromagnetic” configuration of m1 and m2. After turning off jc, m1 first started to switch toward the negative z direction, with m2 dragged by m1 via the ferromagnetic coupling, to complete the switching into the final ferromagnetically coupled state. Although the antiferromagnetically and the ferromagnetically coupled cases exhibit similar intermediate states during the current application, the switching current density (jsw) can be different for the two cases. Figure 4(c) shows jsw as a function of |JIEC|. For |JIEC| ≤ 1.5 × 10−3 J m−2, jsw was smaller for the ferromagnetically coupled case. This tendency was widely observed when tm2 is varied in the range of 0.3 nm ≤ tm2Fig. 4(d). In addition, for the ferromagnetically coupled system, the magnetization switching was observed even when tm2 = 0.792 nm (= 0.99 × tm1), whereas the switching did not happen when exactly tm1 = tm2, indicating that the asymmetry in the properties of magnetic layers, such as the difference in the layer thicknesses, is indispensable for the deterministic SOT switching. Based on the macrospin model, therefore, we expect that a ferromagnetically coupled system with relatively low |JIEC| is promising for efficient spin–orbit torque switching.These calculations also suggest that the formation of the complicated domain structure in Fig. 3(c) may not directly be attributed to the fact that σμ lies in the opposite directions in the top and bottom layers. A multiple domain state may be more favored for the ferromagnetically coupled case to lower the magnetostatic energy, accompanied by the domain-wall trapping after the switching. This suggests that the highly efficient switching is expected not only for an antiferromagnetically coupled film but also for a ferromagnetically coupled film if one can fabricate a nanometer-scaled device with its size smaller than the single magnetic domain size.In summary, we investigated the current-induced magnetization switching in the magnetic multilayer system with interlayer exchange coupling sandwiched by two spin Hall layers of Pt. The domain structures formed during switching varied depending on whether the layers were ferromagnetically coupled or antiferromagnetically coupled. The macrospin calculations for the spin–orbit torque switching revealed that the dual spin–orbit torque from the two Pt layers effectively acts on the two Co layers not only for the antiferromagnetically coupled case but also for the ferromagnetically coupled one. The present results give an understanding of the current-induced magnetization switching mechanism in a magnetic multilayer system and information to design a spintronic device with an efficient magnetization switching.
The authors thank S. Mitani for his technical support and T. Sasaki for her help to carry out the film deposition by ion beam sputtering. This work was supported by JSPS KAKENHI Grant-in-Aid for JSPS Fellows (No. JP22J13517), Grant-in-Aid for Scientific Research (S) (No. JP18H05246), Grant-in-Aid for Scientific Research (A) (No. JP20H00299), and CREST “Creation of Innovative Core Technologies for Nano-enabled Thermal Management” (No. JPMJCR17I1) from JST. R.M. was supported by JSPS through the “JSPS Postdoctoral Fellowship for Research in Japan (Standard)” (No. P21064). The device fabrication was partly carried out at the Cooperative Research and Development Center for Advanced Materials, IMR, Tohoku University. The group in Mainz acknowledges the support by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) 268565370 (SFB TRR173 Project Nos. A01 and B02) as well as TopDyn, the Horizon Europe Project No. 101070290 (NIMFEIA), and the DAAD (Project Nos. 57664531 and 57663728).
Conflict of Interest
The authors have no conflicts to disclose.
Author Contributions
Hiroto Masuda: Data curation (equal); Formal analysis (equal); Funding acquisition (equal); Investigation (lead); Writing – original draft (lead). Koki Takanashi: Conceptualization (equal); Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal). Yuta Yamane: Data curation (equal); Formal analysis (equal); Writing – original draft (supporting). Takeshi Seki: Conceptualization (equal); Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal). Klaus Raab: Investigation (supporting); Writing – review & editing (equal). Takaaki Dohi: Investigation (supporting); Writing – review & editing (equal). Rajkumar Modak: Investigation (supporting); Writing – review & editing (equal). Ken-ichi Uchida: Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal). Jun'ichi Ieda: Supervision (equal); Writing – review & editing (equal). Mathias Kläui: Conceptualization (equal); Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal).
The data that support the findings of this study are available from the corresponding authors upon reasonable request.
REFERENCES
1. A. S. Núñez, R. A. Duine, P. Haney, and A. H. MacDonald, Phys. Rev. B 73, 214426 (2006). https://doi.org/10.1103/PhysRevB.73.214426, Google ScholarCrossref, ISI2. T. Jungwirth, X. Marti, P. Wadley, and J. Wunderlich, Nat. Nanotechnol. 11, 231 (2016). https://doi.org/10.1038/nnano.2016.18, Google ScholarCrossref, ISI3. V. Baltz, A. Manchon, M. Tsoi, T. Moriyama, T. Ono, and Y. Tserkovnyak, Rev. Mod. Phys. 90, 015005 (2018). https://doi.org/10.1103/RevModPhys.90.015005, Google ScholarCrossref, ISI4. L. Néel, Science 174(4013), 985–992 (1971). https://doi.org/10.1126/science.174.4013.985, Google ScholarCrossref5. T. Kampfrath, A. Sell, G. Klatt, A. Pashkin, S. Mährlein, T. Dekorsy, M. Wolf, M. Fiebig, A. Leitenstorfer, and R. Huber, Nat. Photonics 5, 31–34 (2011). https://doi.org/10.1038/nphoton.2010.259, Google Scholar
留言 (0)