Manipulation of magnetization using current-induced torque is crucial for magnetic recording devices. Recently, the spin-orbit torque (SOT) that emerges in a ferromagnetic thin film on a heavy metal is focused as a new scheme for magnetization switching in perpendicularly magnetized systems. Since the SOT provides a perpendicular effective field to the system, the formation of a magnetic multiple domain state because of Joule heating is supressed in the magnetization reversal process. This means that high reliable switching is possible using the SOT. Here, by utilizing the SOT induced domain stability, we show that an electrical current directly injected to a perpendicularly magnetized Pt/Co/Pd system can magnetize itself, that is, current-induced magnetization process from multi to single domain state. A quantitative determination of the SOT is performed using the current-induced magnetization curve. The present results are of great importance as another approach to evaluate the SOT effect, as well as a demonstration of domain state switching caused by the SOT.
In magnetic recording devices, the information of a bit is retained as a magnetization direction. Spin-torque-induced magnetization switching1,2,3,4,5,6 in magnetic tunnel junctions (MTJs) and current-induced domain wall (DW) displacement7,8,9,10,11,12,13 in magnetic wires have been widely investigated as information writing methods in magnetic memory. Recently, a current-induced spin-orbit torque (SOT) emerging in a thin ferromagnetic film deposited on a heavy metal layer has been recognized as a new scheme for magnetization switching in perpendicularly magnetized materials14,15,16,17,18,19. Moreover, it is well known that current-induced DW dynamics is strongly affected by the SOT20,21,22,23,24.
The SOT is known to act on magnetizations as an effective field. This contributes to the magnetization switching and DW motion. More importantly, the SOT effective field also plays a role in stabilizing the domain state against current-induced Joule heating25, 26, which is often a severe problem in devices based on current-induced torque27,28,29. Owing to this characteristic of the SOT, random multiple domain formation is suppressed even under high current density as much as 1011–1012 A/m2 during the magnetization reversal. In this letter, we show that in a perpendicularly magnetized Pt/Co/Pd structure with a multi domain (MD) state in thermal equilibrium (having no net magnetization), the magnetization of the system increases with injected current, and finally, a single domain (SD) state is created (see Fig. 1a). This “current-induced” magnetization process is observed only when an external in-plane magnetic field parallel to the current exists, indicating that the SOT is responsible for the effect. A quantitative determination of the SOT effective field from the current-induced magnetization curve is performed.
Figure 1b shows an optical microscope image of our device. The Hall bar component consists of asymmetrical Pt/Co/Pd layers deposited on a SiO2/intrinsic Si substrate (see also Methods). The width of the channel is 5 μm. Owing to the interface magnetic anisotropy at the Pt/Co and Co/Pd interfaces, the system exhibits perpendicular magnetic anisotropy PMA. The high PMA realizes an MD state with perpendicular magnetization near the Curie temperature (~370 K in our device). Four Cr/Cu electrodes are formed to apply the current and to detect the Hall resistance RHall. The definitions of each external magnetic field (μ0H x , μ0H y , and μ0H z ) and the current flow along the x-axis are indicated in Fig. 1b. The temperature of the stage Ts, which is in thermal contact with the fabricated device, is controlled using a heater (see Methods).
Figure 2a shows the results of the RHall measurement when a dc current density Jdc of +2.8 × 109 A/m2 is injected at room temperature (Ts = 304 K). μ0H z is swept to obtain the curves. Jdc is determined by simply dividing the current by the cross-sectional area of the metallic layers. RHall is proportional to the perpendicular component of the magnetization because of the anomalous Hall effect. A clear hysteresis loop with a coercivity μ0Hc of 2.4 mT is observed. The remanent value of RHall (RHallr) is almost equal to the saturation value, indicating that the SD state is stable at fields near zero.
Next, RHall measurements with Jdc sweeping are performed under constant in-plane magnetic fields. The procedure is as follows. First, the magnetization direction of the entire device is set upward by applying a μ0H z of +30 mT. After returning to ~0 T, a constant in-plane magnetic field (μ0H x or μ0H y ) of +38 mT is applied. Then, RHall is monitored with a sweeping Jdc. The positive (negative) sweep corresponds to a Jdc change from 0 to +1.7 (−1.7) × 1011 A/m2. Figure 2b and c show RHall as a function of Jdc measured under the application of μ0H y and μ0H x , respectively. As shown in the positive sweep (PS) in Fig. 2b, RHall abruptly decreases when a Jdc of +0.8 × 1011 A/m2 is applied, and drops toward zero in the region of Jdc > +0.8 × 1011 A/m2, indicating that the MD state is formed at Jdc ≥ 0.8 × 1011 A/m2 due to the Joule heating. A rapid decrease in RHall also appears symmetrically in the negative sweep (NS) case (see NS curve in Fig. 2b), suggesting that the Joule heating effect is independent of the Jdc direction.
By contrast, the situation is completely different when μ0H x , which is parallel to Jdc, is applied. For the PS curve in Fig. 2c, a full RHall switching from positive to negative, which corresponds to the abrupt reversal of the magnetization direction from up to down, is observed at Jdc = +0.5 × 1011 A/m2. In the NS curve, no switching occurs up to Jdc = −1.7 × 1011 A/m2. When the sign of μ0H x is reversed, switching is observed only for a negative Jdc (not shown). The switching direction in the present configuration is consistent with the SOT induced switching previously reported in the Pt/Co structure2, 26. Although the sign of the spin Hall angle of the top Pd is the same as that of the Pt, the SOT in the present device is dominated by the spin current injection from the bottom Pt because the magnitude of the spin Hall angle of Pd is one order smaller than that of Pt30, 31. We also checked that in a similar device structure, the top Pd effect on the SOT is negligibly small32. In order to evaluate the magnitude of SOT effective fields, i.e. the Slonczewski-like torque μ0HSL and field-like torque μ0HFL, the harmonic Hall measurement, which is widely used for the quantitative determination of the SOT33, was conducted for a similar device structure. The procedure is shown in the Supplementary Information. μ0HSL and μ0HFL are determined to be 6.55 ± 0.20 mT/1011 A/m2 and 1.57 ± 0.13 mT/1011 A/m2, respectively. These values are close to those previously reported for Pt/ferromagnet bilayer structure22. Another important point here is that under μ0H x application, |RHall| always represents a value close to saturation even at |Jdc| ≥ 0.8 × 1011 A/m2, indicating that the MD formation is completely suppressed against Joule heating. In the present case, the magnetization experiences a finite perpendicular effective field μ0Heff derived from μ0HSL because the magnetization tilts slightly toward the x-direction. When the current direction is the same as the μ0H x direction, the sign of μ0Heff becomes negative in the Pt/Co system. As a result, the SD state becomes stable because of the gain by the Zeeman energy reduction. Thus, the SOT plays an important role in stabilizing the SD state.
In the above experiments, Td was increased by injecting a current. In the following, the MD state at thermal equilibrium is prepared by simply increasing Td using a heater. Figure 3a shows the results of the Hall measurements performed at Ts = 343 K. Jdc for this measurement is 2.8 × 109 A/m2, and the Joule heating effect is negligibly small. In this case, an anhysteric RHall curve with the almost zero remanent is observed. This indicates that the MD state is realized and the system is demagnetized. Figure 3b shows RHall as a function of Jdc obtained under μ0H z of ~0 T. RHall is almost independent of Jdc in the range of ±0.5 × 1011 A/m2. Similarly, no change in RHall is observed for a μ0H y of +38 mT, as shown in Fig. 3c. The slight RHall deviation from zero is probably due to the small z-component of the field. The slight Jdc dependence on RHall shown in Fig. 3b and c might be a result of the Oersted field. Based on these results, it can be concluded that in both cases, the MD state is kept under current injection. The result obtained under μ0H x = +38 mT is completely different, as shown in Fig. 3d. For a small Jdc, RHall showed an intermediate value, indicating that the MD state remains. However, a clear increase and decrease in RHall with increasing current toward the negative and positive magnetization directions, respectively, are observed. For both current directions, RHall saturates above |Jdc| = 0.1–0.2 × 1011 A/m2 and the RHall saturation values are consistent with those for the SD state. This result indicates that magnetization process of the system is caused by the SOT.
In order to check whether the device is fully in the SD state, a cooling experiment under current is carried out. The procedure is as follows. After making RHall saturation using current under μ0H x = +38 mT, Td is gradually decreased from 343 K to room temperature with remaining the current and x-field application. Jdc of −0.25 × 1011 A/m2 is continuously applied during the cooling. During the Td decrease, RHall is monitored. The result is shown in Fig. 3e. One can see that RHall gradually increases with decreasing Td and at room temperature it shows ~0.4 Ω, which is the value for the SD state. This indicates that the SD state is created by the current at 343 K and is maintained during cooling. On the other hand, when the above experiment is performed with an injection of Jdc = −2.8 × 109 A/m2, RHall shows a small value of ~0.1 Ω even at room temperature, suggesting that the MD state formed at 343 K is retained even at room temperature. Therefore, the RHall saturation in Fig. 3(d) corresponds to the complete SD state, i.e. a situation where no domain with an opposite magnetization exists in the device is realized against thermal agitation. The result presented here demonstrates that the magnetization curve of Pt/Co/Pd system can be obtained by only sweeping electrical current owing to the stability caused by the SOT. In addition, the magnetization direction can be reversibly controlled by simply changing the current polarity.
The magnetization processes shown in Fig. 3(a) and (d) are caused by μ0H z and μ0Heff, which is proportional to Jdc, respectively. The current-induced magnetization process is expected to develop with the DW motion in the current direction, while the magnetic domains isotropically expands when the magnetization process is caused by the external field. Here, we focus on the gains of the Zeeman energy for each case and determine μ0Heff, and consequently μ0HSL, by comparing them. First, EZ obtained by μ0H z (EZ_ H ) is calculated for the up magnetization case. Figure 4(a) shows the normalized RHall (RHalln) as a function of μ0H z in the range from 0 to +4.0 mT. EZ_ H can be calculated using the following equation:
where Ms is the saturation magnetization of the system. The integral term of (1) is defined by the coloured area of Fig. 4(a), and EZ_ H /Ms is determined to be 0.180 mT. Subsequently, using the current-induced magnetization curve shown in Fig. 3(d), EZ resulting from μ0Heff (EZ_ J ) is calculated. RHalln as a function of Jdc in the negative Jdc sweep, where the positive μ0Heff is applied to the system, is shown in Fig. 4(b). The small Jdc offset is corrected in this figure. Since μ0Heff is expected to be proportional to Jdc, EZ_ J can be determined from the following:
where α is the constant value defined as μ0Heff/Jdc. The calculation of EZ_ J /Ms is done in the same manner and is determined to be 0.662α × 1010 A/m2. The same calculations are performed for the down magnetization case. Assuming EZ_ H /Ms = EZ_ J /MS, α = 2.51 ± 0.01 mT/1011 A/m2 is finally obtained.
Next, we calculated μ0HSL (per 1011 A/m2) using μ0Heff determined above. Figure 4(c) shows a schematic where the direction of μ0HSL under the x-field is denoted. μ0Heff is the perpendicular component of μ0HSL and θ is the tilting angle of the magnetization from the z-axis. Thus, μ0HSL is expressed as μ0Heff/sinθ. From the x-field dependence of the anomalous Hall resistance, θ at μ0H x = +38 mT is approximately 19° in our Pt/Co/Pd structure. Therefore, μ0HSL of 7.67 ± 0.04 mT/1011 A/m2 is obtained. This value shows good agreement with that obtained by the harmonic measurement in our Pt/Co/Pd system. μ0Heff was also calculated using a Langevin fit and the result is consistent with the direct Zeeman energy calculation (see Supplementary Information). The calculation results presented here indicate that the SOT effective field can be quantitatively determined from the current-induced magnetization curve.
Finally, we demonstrate current-induced alternate switching between the MD and SD states. Figure 5(a) shows the sequence of Jdc injection into the device. A pulsed Jdc with three values of +0.4 × 1011, +2.7 × 109, and −0.4 × 1011 A/m2 is injected in series to create the SD state with up and down magnetization. The duration of each Jdc pulse is 1.0 s, and a μ0H x of +38 mT is applied during the measurement. The RHall value is measured during the pulse injection, and the result is shown in Fig. 5(b). Maximum and minimum RHall values corresponding to the up and down SD states appear alternately with the injections of +0.4 and −0.4 × 1011 A/m2. At Jdc = 2.7 × 109 A/m2, RHall always exhibits an intermediate value, that is, the MD state is restored. We checked that the current-induced SD state returns to the MD state within 1 ms after pulse off. This result indicates that an arbitrary switching of domain states between MD and SD can be achieved by injecting a current. Although in the present case, 1-ns-long pulses were used for the convenience of the measurement, sub-ns domain state switching is expected to be possible because the SOT induced magnetization switching occurs in this time scale34.
When the SOT effectively acts on the magnetization, the electrical current flowing in the system enhances the stability of the single domain state. In this study, owing to this domain stability, we show that the magnetization curve of the nonmagnetic/ferromagnetic metal structure can be obtained by sweeping electrical current, while it is conventionally obtained by sweeping external magnetic field. This effect would be marked in smaller size applicable to current IT devices because the SOT field is proportional to the current density flowing in the heavy metal layer. Anisotropy-wedged film16, an interlayer exchange coupled system17, and an antiferromagnet/ferromagnet layered structure18, where in-plane field free magnetization switching by the SOT was achieved, may also enable domain state switching without an in-plane field. In addition, this work offers a novel method to determine the SOT effective field from the current-induced magnetization curve.
Film deposition and device fabrication
Multilayered Ta (2.7 nm)/Pt (3.0)/Co(0.36)/Pd (0.8) film was deposited on a thermally oxidized Si substrate using rf sputtering, and a 0.5-nm-thick Ta layer was formed on the film as a cap. The base pressure of the sputter chamber was below 1.0 × 10−6 Pa, and Xe process gas was used for the deposition. The X-ray diffraction profile indicates that the Pt layer has an fcc (111) texture. The film was patterned into a Hall bar structure by photolithography and Ar ion milling. Cr (1.0)/Cu (100) electrodes were deposited by thermal evaporation and defined by a lift-off process using photolithography.
Measurements were performed using a prober system in which a vector magnetic field can be applied. The device temperature was controlled by a plate-shaped heater placed under the device and monitored using a Pt thermometer (Pt-100). A current source (Yokogawa 7651) and nano-voltmeter (Keithley 2182A) were used for anomalous Hall measurements.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The authors thank S. Ono, A. Tsukazaki, F. Matsukura, and H. Ohno for their technical help. This work was partly supported by the Grants-in-aid for Young Scientists (A) (No. 15H05419), Scientific Research (S) (No. 25220604), Specially promoted Research (No. 15H05702) from JSPS, the Murata Science Foundation, and Spintronics Research Network of Japan.
Electronic supplementary material
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.