This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Spin-Orbit Torque Switching of Noncollinear Antiferromagnetic Antiperovskite Manganese Nitride Mn3GaN

T. Hajiri [email protected] Department of Materials Physics, Nagoya University, Nagoya 464-8603, Japan    K. Matsuura Department of Materials Physics, Nagoya University, Nagoya 464-8603, Japan    K. Sonoda Department of Materials Physics, Nagoya University, Nagoya 464-8603, Japan    E. Tanaka Department of Materials Physics, Nagoya University, Nagoya 464-8603, Japan    K. Ueda Department of Materials Physics, Nagoya University, Nagoya 464-8603, Japan    H. Asano Department of Materials Physics, Nagoya University, Nagoya 464-8603, Japan
Abstract

Noncollinear antiferromagnets have promising potential to replace ferromagnets in the field of spintronics as high-density devices with ultrafast operation. To take full advantage of noncollinear antiferromagnets in spintronics applications, it is important to achieve efficient manipulation of noncollinear antiferromagnetic spin. Here, using the anomalous Hall effect as an electrical signal of the triangular magnetic configuration, spin–orbit torque switching with no external magnetic field is demonstrated in noncollinear antiferromagnetic antiperovskite manganese nitride Mn3GaN at room temperature. The pulse-width dependence and subsequent relaxation of Hall signal behavior indicate that the spin–orbit torque plays a more important role than the thermal contribution due to pulse injection. In addition, multistate memristive switching with respect to pulse current density was observed. The findings advance the effective control of noncollinear antiferromagnetic spin, facilitating the use of such materials in antiferromagnetic spintronics and neuromorphic computing applications.

I INTRODUCTION

Noncollinear antiferromagnetic (AFM) materials have attracted significant attention in basic and applied science because in addition to having the excellent properties of collinear AFM—such as fast dynamics, suitability for high-density integration, and stability against external perturbations—noncollinear AFM materials can overcome the weakness of collinear AFM materials—namely the small electrical signal—by the anomalous Hall effect (AHE) [1, 2, 3, 4]. Thus, efficient control of noncollinear AFM spin is essential for the application of such materials in AFM spintronics. In the past decade, electrical manipulation via spin–transfer torque (STT) and spin–orbit torques (SOTs) has become one of the most promising techniques in the field of ferromagnet (FM)–based spintronics, not only to take the place of dynamic random access memory in the current computer memory hierarchy [5] but also to instantiate multistate magnetoresistive random access memories for neuromorphic computing [6, 7].

Recent studies demonstrate that like collinear AFM materials, noncollinear AFM materials can be controlled via SOT in the same way as FM materials. Collinear AFM materials consisting of NiO/Pt bilayers and Pt/NiO/Pt trilayers show a critical current density (JcJ_{\mathrm{c}}) on the order of 107–-108 A/cm2 [8, 9, 10], similar to that in typical FM/heavy metal (HM) bilayers. In contrast, JcJ_{\mathrm{c}} values one to two orders of magnitude smaller JcJ_{\mathrm{c}} have been observed in noncollinear AFM Mn3GaN (MGN)/Pt bilayers (1.5×1061.5\times 10^{6} A/cm2[12] and Mn3Sn/W bilayers (5×1065\times 10^{6} A/cm)2{}^{2}) [13]. In the case of a collinear AFM system, 90 switching of the Ne´\rm{\acute{e}}el vector is required because the electrical signal, such as spin Hall magnetoresistance and anisotropic magnetoresistance, is maximal. To achieve 90 switching of the Ne´\rm{\acute{e}}el vector, diagonal current flow using an eight- or four-terminal device with complex electrical write/read operation is required [8, 9, 10], except for CuMnAs using two terminal writing device [11]. A noncollinear AFM system, by contrast, needs 180 switching of each spin of triangular magnetic configuration, which can be accomplished using a simple four-terminal Hall device with no external magnetic field in Mn3GaN [12], or with an external magnetic field in Mn3Sn [13]. Therefore, in addition to their large electrical signals due to non-zero Berry curvature [14], noncollinear AFM systems have a distinct advantage in spintronic applications.

In the antiperovskite manganese nitrides MnA3{}_{3}AN (where AA = Ni, Ga, Sn, etc.), the Mn atoms form a kagome lattice in the (111) plane. The noncollinear AFM MnA3{}_{3}AN with a nonzero Berry curvature has been predicted to exhibit a large anomalous Hall effect (AHE) and an anomalous Nernst effect even with a quite small canted magnetization of the order of 0.0010.0010.010.01 μB\mu_{\rm B} per atom [15, 16, 17]. The AHE has been established in Mn3Ni1-xCuxN films [18, 19], strained Mn3NiN films [20], and strained Mn3SnN films [21]. Although electrical current switching of both Hall resistance with no external magnetic field and nonlinear Hall resistance with respect to an external magnetic field have been reported in MGN/Pt bilayers, no clear evidence of AHE has yet been presented [12]. In addition, a thermal contribution due to pulse current injection such as a thermal activation effect, a joule heating effect, or an electromigration effect can change the Hall resistance [22, 23, 24]. Hence, as has been pointed out, both the low-electrical-current writing and reading operations of MGN/Pt bilayers could be of nonmagnetic thermal origin [25].

Refer to caption
Figure 1: (a) Out-of-plane and in-plane XRD profiles for MGN films around (002)/(200) peaks. (b) In-plane ϕ\phi scans of MGN films. (c) Temperature (TT) dependence of the resistivity ρxx\rho_{\mathrm{xx}} and magnetization MM of MGN films. For the magnetization measurement, an external magnetic field μ0H\mu_{0}H of 10 mT was applied parallel to the [001] direction. (d) Out-of-plane magnetic hysteresis loops and (e) anomalous Hall resistivity loops for MGN films at various temperatures. Inset of panel (e) is an enlargement of the anomalous Hall resistivity loop at 300 K. (f) Out-of-plane magnetization at 2 T and anomalous Hall resistivity ρxy,2T\rho_{\mathrm{xy},\rm{2T}} at 2 T for MGN films as a function of temperature. The inset presents the anomalous Hall resistivity at 300 K as a function of the c/ac/a ratio. c/a=0.9962c/a=0.9962 film is 35 nm thick, while other films are 50 nm thick.

In this investigation, we studied the magnetotransport properties of MGN films and performed systematic switching operations of strained MGN/HM (HM = Pt, Ta) bilayers. At room temperature, no AHE was obtained in relaxed MGN films, whereas the AHE appeared with respect to the ratio between the lattice constants cc and aa (c/ac/a ratio). The temperature dependence of the magnetization and the AHE suggests that skew scattering is dominant below 200 K, whereas another origin, probably noncollinear AFM order with nonzero Berry curvature, is dominant above 200 K. To estimate the effect of thermal contributions on write/read operations, the pulse-width dependence and relaxation after pulse injection were measured. We show that the thermal activation effect, joule heating effect, and electromigration effect play a minor role. The existence of multistate signal amplitude with respect to pulse current with no external magnetic field was demonstrated.

II EXPERIMENTAL DETAILS

MGN films were grown by reactive magnetron sputtering on MgO(001) substrates using a Mn3Ga target at 400 C. As details of the film growth were reported in our previous work, the c/ac/a ratio was controlled by precise control of N2 partial pressure during film growth [26]. The crystal structure was analyzed using both in-plane and out-of-plane X-ray diffraction (XRD) measurements with Cu KαK\alpha radiation. Magnetic properties were characterized using superconducting quantum interference device magnetometry. Transport properties were characterized by the standard DC four-terminal method. For write/read operations, a layer of Pt or Ta (3nm) was deposited by magnetron sputtering at room temperature after film growth. 20 μ\mum-width Hall bars with Ti/Cu contact pads were prepared by a conventional photolithographic process. All SOT measurements were performed with no external magnetic field at room temperature (300 K). The sequence of write/read operations was the same as described in our previous report [12].

III RESULTS AND DISCUSSIONS

III.1 Film Characteristics and Magnetotransport Properties

Typical out-of-plane 2θ2\thetaω\omega and in-plane 2θχ2\theta_{\chi}ϕ\phi X-ray diffraction (XRD) patterns for the 35-nm-thick MGN films are shown in Fig. 1(a). Only the MGN (002) and (200) planes exhibit Bragg peaks, in the out-of-plane and in-plane XRD patterns, respectively. In addition, epitaxial growth is confirmed by the results of ϕ\phi-scan measurement as shown in Fig. 1(b), showing that their epitaxial relationship is MgO(001)[100]//MGN(001)[100]. The lattice constants cc and aa are 0.38817 nm and 0.38965 nm, respectively, giving c/a=0.9962c/a=0.9962. The temperature dependencies of the resistivity ρxx\rho_{\mathrm{xx}} and magnetization MM are shown in Fig. 1(c). In the ρxx\rho_{\mathrm{xx}} curve, there is a clear anomaly at 200 K. Likewise, an FM-like transition is observed at 200 K in the magnetization curve. These temperature dependencies were also observed in our previous switching study of MGN/Pt films [12]. The ground state of MGN at room temperature is well known to exhibit Γ5g\Gamma_{5g} spin structure [27]. The coexistence of Γ5g\Gamma_{5g} and M-1 phase below the FM-like transition temperature has also been previously reported [28]. As the MGN/FM bilayers exhibit an exchange bias at 4 K [29], it is concluded that Γ5g\Gamma_{5g} order and M-1 phase coexist below 200 K.

Figure 1(d) and 1(e) show the magnetic MM hysteresis and anomalous Hall resistivity ρxy\rho_{\mathrm{xy}} loops for the MGN films measured along the 001\langle 001\rangle direction at various temperatures. In the ρxy\rho_{\mathrm{xy}} loops, linear contribution from the ordinary Hall effect has been subtracted. The ρxy\rho_{\mathrm{xy}} loops before subtract the ordinary Hall effect and a summary of the Hall coefficient values are given in Fig. 6 of Appendix A. In both loops, hysteresis is clearly exhibited not only in the coexistence phase below 200 K but also in the Γ5g\Gamma_{5g} single phase. Both the magnetization and the ρxy\rho_{\mathrm{xy}} loops of the MGN films have similar coercive field (HcH_{c}) values, indicating that the AHE is directly related to the MGN magnetic order. The temperature dependencies of the magnetization and ρxy\rho_{\mathrm{xy}} at 2 T are shown in Fig. 1(f). Above 200 K, the magnetization remains nearly constant at 0.02\sim 0.02 μB/Mn\mu_{\rm B}/\mathrm{Mn}, and below 200 K, it increases monotonically with decreasing temperature. ρxy\rho_{\mathrm{xy}}, on the other hand, increases slightly with decreasing temperature above 200 K and increases dramatically with decreasing temperature below 200 K, indicating that the increase in ρxy\rho_{\mathrm{xy}} is strongly linked to the increase in magnetization. In contrast to the magnetization, however, ρxy\rho_{\mathrm{xy}} begins to decrease below 75 K. As ρxy\rho_{\mathrm{xy}} is found to be proportional to ρxx\rho_{\mathrm{xx}} below 75 K, ρxy\rho_{\mathrm{xy}} below 200 K would be dominated by a net magnetization.

According to theoretical studies on piezomagnetism of MGN with Γ5g\Gamma_{5g} order, a net magnetization can appear by the induction of strain [30, 31], and the magnitude of the net magnetization increases linearly with respect to the strain ϵ\epsilon %, with a coefficient of 0.013 μB/Mn/%\mu_{\rm B}/\mathrm{Mn}/\% [30]. Our MGN films show ϵ\epsilon and remnant magnetization of approximately 0.3 % and 0.004 μB/Mn\mu_{\rm B}/\mathrm{Mn} at 300 K, respectively. As the net magnetization observed in our MGN films is similar to the value calculated theoretically, it is considered that the canted Γ5g\Gamma_{5g} order is realized above 200 K. The Γ5g\Gamma_{5g} order does not show the AHE because it has mirror symmetry, and the symmetry operations make the Berry curvature vanish after integration over the entire Brillouin zone [15]. In antiperovskite nitride films with Γ5g\Gamma_{5g} order, however, the anomalous Hall conductivity (AHC) tensor is reported to be highly sensitive to strain. Although strain-free Mn3NiN films show no AHE [18], strained Mn3NiN films do show AHE [20]. This is explained by the reduction of the symmetry of a space group, under which nonzero Berry curvature is induced when a finite strain is applied [20]. Strained Mn3SnN films likewise show a large AHE, but it is suggested that the biaxial strain induces Γ4g\Gamma_{4g} order from Γ5g\Gamma_{5g} order [21]. These past findings indicate that AHE cannot be accounted for by either extrinsic scattering processes or changes in magnetization; hence, nonzero Berry curvature plays an important role. In our MGN films, ρxy\rho_{\mathrm{xy}} is observed to increase with decreasing c/ac/a ratio at 300 K as shown in the inset of Fig. 1(f), highlighting the fact that appearance of AHE at 300 K in MGN films is also strongly related to the film strain and/or reduced magnetic space group. Therefore, although we cannot experimentally separate the contributions from canted net magnetization and nonzero Berry curvature due to noncollinear AFM order above 200 K, we can state that part of the AHE comes from noncollinear AFM order. From the view point of SOT, we will discuss the possible origin of AHE in the later section.

III.2 Dependence of Reversible Electrical Switching on HM

Table 1: Resistivities of Mn3GaN, Pt, and Ta single-layer films, and Mn3GaN/Pt and Mn3GaN/Ta bilayers at 300 K. The resistivities of Mn3GaN/Pt and Mn3GaN/Ta bilayers were calculated using a parallel circuit model from the corresponding single layer resistivities.
Film ρxx\rho_{\mathrm{xx}} (μΩ\mu\Omega cm)
Mn3GaN 1271
Pt 112
Ta 972
Mn3GaN/Pt 504  (exp.),  541 (calc.)
Mn3GaN/Ta 1261 (exp.),  1222 (calc.)

For the electrical write/read operations, typical Hall bar devices of bilayers consisting of strained-MGN (20 nm) with either Pt or Ta (3 nm), with 20 μ\mum width were prepared. From a parallel circuit model, the ρxx\rho_{\mathrm{xx}} values for both the MGN/Pt and MGN/Ta bilayers can be derived using ρxx\rho_{\mathrm{xx}} of each single-layer film, which enable us to estimate current density through the Pt and Ta layers (Jc,HMJ_{\mathrm{c},HM}). The ρxx\rho_{\mathrm{xx}} values at 300 K for MGN, Pt, and Ta single-layer films and MGN/Pt and MGN/Ta bilayers are given in Table. 1. Using these bilayers, sequential write/read operations were performed, with the results as shown in Fig. 2. Here, the top and bottom axes are the current densities derived from the Pt/Ta layer alone and the bilayer total thickness, respectively, and the left and right axes are the ρxy\rho_{\mathrm{xy}} values derived from the full bilayer and the MGN layer alone, respectively. In addition, the constant offset probably due to geometrical imperfections of the Hall bar and/or thermoelectric voltage was subtracted. ρxy\rho_{\mathrm{xy}} changes with respect to JJ, and a clear hysteresis loop is observed in both bilayers at 300 K with no external magnetic field, indicating success of the electrical write/read operation. The critical current densities JcJ_{\mathrm{c}} (Jc,HMJ_{\mathrm{c},HM}) for MGN/Pt and MGN/Ta were obtained as 2.72×1062.72\times 10^{6} A/cm2 (13.13×10613.13\times 10^{6} A/cm2) and 4.81×1064.81\times 10^{6} A/cm2 (6.13×1066.13\times 10^{6} A/cm2), respectively. Theoretically, JcJ_{\mathrm{c}} would be proportional to θHM1\theta_{HM}^{-1}, where θHM\theta_{HM} is the spin Hall angle of the HM layer [32]. We find that the ratio Jc,Pt:Jc,Ta=2.1:1J_{\mathrm{c},\rm{Pt}}:J_{\mathrm{c},\rm{Ta}}=2.1:1 is nearly equal to the ratio |θPt|1:|θTa|1=1.9:1|\theta_{\rm{Pt}}|^{-1}:|\theta_{\rm{Ta}}|^{-1}=1.9:1, indicating that spin current generated in the HM layer plays an important role. We note that the different ρxy\rho_{xy} switching width between MGN/Pt and MGN/Ta devices is probably due to local variations in the quality and/or strain of the thin films, which is discussed in Fig. 7 of Appendix B. Besides, the comparison of ρxy\rho_{xy} width between field-sweep and SOT measurements using the same device is presented in Fig. 8 of Appendix C.

Refer to caption
Figure 2: Sequential electrical current switching for (a) MGN/Pt and (b) MGN/Ta bilayers with no external magnetic field at room temperature. The current density JJ is derived by the total thickness of the bilayers, and the current density through HM layer JHMJ_{HM} is derived by HM thickness and the current through the HM layer as estimated using a parallel circuit model. In the same manner, ρxy\rho_{\mathrm{xy}} and ρxy,MGN\rho_{xy,\rm{MGN}} are derived by the total thickness of bilayers and by MGN thickness and the current through MGN layer.

To evaluate the effect of joule heating by applying pulse current, the pulse-width dependence was investigated. Figure 3(a) shows ρxy\rho_{\mathrm{xy}} as a function of JJ with several pulse widths for MGN/Ta bilayers. Although the switching ρxy\rho_{\mathrm{xy}} amplitudes remain nearly the same, the hysteresis loops become slightly narrower with increasing pulse width. JcJ_{\mathrm{c}} is plotted as a function of pulse width in Fig. 3(b). Here, JcJ_{\mathrm{c}} is fitted by the thermal activation model [33];

Jc=Jc0[11Δln(ττ0)],J_{\mathrm{c}}=J_{\mathrm{c}0}\left[1-\frac{1}{\Delta}\rm{ln}(\frac{\tau}{\tau_{0}})\right], (1)

where Jc0J_{\mathrm{c}0} is the critical current density at 0 K, Δ\Delta is the thermal stability factor, and τ01\tau_{0}^{-1} is the thermally activated switching frequency, for which we assume a frequency of 1/τ0=11/\tau_{0}=1 THz. It can be seen that JcJ_{\mathrm{c}} can be fitted by the thermal activation model with Jc0=7.97×106J_{\mathrm{c}0}=7.97\times 10^{6} A/cm2 and Δ=58.7\Delta=58.7. These results highlight the finding that a JcJ_{\mathrm{c}} value of the order 10610^{6} A/cm2 originates intrinsically in spin torque whereas the thermal activation, through it exists, plays a minor role.

Refer to caption
Figure 3: (a) ρxy\rho_{\mathrm{xy}} as a function of JJ observed for several pulse widths in MGN/Ta bilayers with no external magnetic field at room temperature. (b) Critical current density JcJ_{\mathrm{c}} as a function of pulse width. The bold solid line is the result of a fit to a thermal activation model.

III.3 Thermal Contribution to Write/Read Operations

Because heating and electromigration effects can affect ρxy\rho_{\mathrm{xy}} signal amplitudes [23], the relaxation behavior after switching was investigated. Figure 4 presents the continuous write/read operation using Ipulse=±13×106I_{\rm{pulse}}=\pm 13\times 10^{6} A/cm2 with a 5 μ\mus pulse width and subsequent relaxation measurements for MGN/Ta bilayers at 300 K. Figure 4(b) is an enlargement of the continuous ±Ipulse\pm I_{\rm{pulse}} write part shown in Fig. 4(a). As discussed for synthetic AFM materials [34] and in our previous MGN/Pt study [12], the observed asymptotic ρxy\rho_{\mathrm{xy}} behavior can be fitted by an exponential decay function y=y0+Aexp[(x+x0)/τ]y=y_{0}+A_{~{}}\mathrm{exp}\left[-(x+x_{0})/\tau\right] with time constants τ\tau of 14.9 pulse number (197.0 s) for +Ipulse+I_{\rm{pulse}} and 15.7 pulse number (207.6 s) for Ipulse-I_{\rm{pulse}} with continuous write/read operations in approximately 13.25 s cycles. The relaxation behavior after switching was measured after three cycles of the continuous ±Ipulse\pm I_{\rm{pulse}} write operations.

Refer to caption
Figure 4: (a) Continuous write/read operations and subsequent relaxation measurements with no external magnetic field at room temperature. (b) Continuous write/read operation part from panel (a). The dashed lines are the results of a fit to the exponential decay function. (c) Relaxation measurement parts from panel (a). The bold solid lines are the results of a fit to the double exponential function.

Figure 4(c) presents the ρxy\rho_{\mathrm{xy}} relaxation as a function of time after continuous ±Ipulse\pm I_{\rm{pulse}} write operations. The relaxation behavior is characterized by fit to a double exponential function [23],

d=d0+d1exp(tτ1)+d2exp(tτ2),d=d_{0}+d_{1}~{}\mathrm{exp}\left(-\frac{t}{\tau_{1}}\right)+d_{2}~{}\mathrm{exp}\left(-\frac{t}{\tau_{2}}\right), (2)

where d0d_{0} is a base line and is the value reached when attenuated, d1d_{1}, d2d_{2} are amplitude parameters, and τ1\tau_{1}, τ2\tau_{2} are the relaxation times. Both relaxations after ±Ipulse\pm I_{\rm{pulse}} write operations fit well with d0=0.0467d_{0}=-0.0467 μΩ\mu\Omegacm, d1=0.0219d_{1}=-0.0219 μΩ\mu\Omegacm, d2=0.0003d_{2}=-0.0003 μΩ\mu\Omegacm, τ1=8.63\tau_{1}=8.63 h, τ2=11.67\tau_{2}=11.67 h for after the +Ipulse+I_{\rm{pulse}} write operation, and d0=0.0331d_{0}=0.0331 μΩ\mu\Omegacm, d1=0.0100d_{1}=0.0100 μΩ\mu\Omegacm, d2=0.0194d_{2}=0.0194 μΩ\mu\Omegacm, τ1=0.25\tau_{1}=0.25 h, τ2=3.89\tau_{2}=3.89 h for after the Ipulse-I_{\rm{pulse}} write operation. With regard to the change in Hall resistance due to the effects of annealing and electromigration, the short and long decay times are reported to be approximately 4 min and 50 min, respectively [23]. Compared with these values, both the short and long decay times observed here are five to ten times longer. In addition, for the change in Hall resistance due to annealing and electromigration effects, the Hall resistance asymptotes to the initial value before pulse injection is a few minutes to a few hours [23, 35], in contrast with our MGN/Ta bilayers, for which d0d_{0} is a rather large value. Whereas our electrical measurements do not enable us to fully distinguish SOT and other possible contributions such as thermal activation and electromigration effects, the relaxation behavior and pulse-width behavior in the MGN/Ta bilayers highlight the fact that SOT plays an important role in the present switching behavior. In the case of collinear AFM materials such as NiO, 9090^{\circ} switching of the Ne´\rm{\acute{e}}el vector is required, which can be achieved by the flow of current in two orthogonal directions. However, current flow in orthogonal directions causes inhomogeneous current density due to the current crowding effect, which induces a change in Hall resistance due to annealing and electromigration effect [23]. In the case of noncollinear AFM materials, in contrast, 180180^{\circ} switching of each noncollinear spin is required, which can be achieved by the flow of current in a straight line; this implies that the current crowding effect in this study can be considered small.

Refer to caption
Figure 5: Dependence of ρxy\rho_{\mathrm{xy}}JJ loops on JminJ_{\rm{min}} for MGN/Ta bilayers with no external magnetic field at room temperature.

III.4 Memristive Switching with Respect to Pulse Current Density

Finally, here we discuss the memristive behavior of MGN/Ta bilayers. Figure 5 shows ρxy\rho_{\mathrm{xy}}JJ loops at 300 K, where Jmax=13.0×106J_{\rm{max}}=13.0\times 10^{6} A/cm2 was first applied and subsequently scanned from JmaxJ_{\rm{max}} to JminJ_{\rm{min}} and from JminJ_{\rm{min}} to JmaxJ_{\rm{max}}. ρxy\rho_{\mathrm{xy}} exhibits multiple stable signal amplitudes according to the magnitude of JminJ_{\rm{min}}. The same behavior has been observed in Mn3Sn/Pt bilayers [13] and AFM/FM bilayers [37], suggesting that the phenomenon of multiple stable magnitudes of the Hall resistance originates in the multi-AFM domain character, which allows us to tune the signal amplitude in an analog manner. In contrast to Mn3Sn/Pt bilayers [13], MGN/Ta bilayers show a memristive behavior with no external magnetic field, highlighting the advantage of MnA3{}_{3}AN systems for use in neuromorphic computing.

According to theoretical study of SOT of Γ4g\Gamma_{4g} order, the noncollinear spins rotate in the (111) plane of the kagome lattice, where the injected spins are directed perpendicular to the kagome lattice. Therefore, JcJ_{\mathrm{c}} is determined by an in-plane anisotropic energy of the kagome lattice and the injected spin direction. Further improvements in SOT efficiency in noncollinear MnA3{}_{3}AN systems may be attained using (110)-oriented films of low-anisotropy materials. Theoretically, Mn3Ga1-xNix[36] has a low anisotropy energy, implying that it would be worthwhile to investigate the dependence of SOT efficiency on AA atoms. On the other hand, if the AHE comes from only net magnetization (for simplicity, assume a net magnetization in the perpendicular direction like Ferrimagnet), the magnetic field parallel to the current direction is generally needed to realize the SOT switching. In contrast, SOT of noncollinear AFM theoretically satisfies even though with no external magnetic field [32, 36]. Indeed, no switching has been observed with no external magnetic field at low temperatures where the Ferrimagnetic M1 phase is dominant in the AHE [12]. Although the joule heating can affect to the Hall resistivity, the pulse-width measurement and relaxation measurement show a heating effect plays a minor role. From these results, we can conjecture that AHE is possibly related to noncollinear AFM order from the view point of SOT at 300 K.

IV Conclusion

In this study, we have shown the AHE and SOT switching of noncollinear AFM MGN at room temperature. By tuning the c/ac/a ratio, two origins of AHE were observed: one for AHE above 200 K, possibly related to noncollinear AFM order, and one for AHE below 200 K, dominated by magnetization. Using MGN/HM bilayers, we have demonstrated the SOT switching of noncollinear AFM spin in MGN at room temperature with no external magnetic field. The effects of thermal activation on JcJ_{\mathrm{c}} and the effects of heating and electromigration on ρxy\rho_{\mathrm{xy}} were excluded by pulse-width measurements and relaxation measurements after pulse injection. In addition, multistate memristive switching with respect to pulse current density was demonstrated. These results show that, efficient SOT can be attained in MGN/HM bilayers with memristive functionality with no external magnetic field, and demonstrate the potential application in AFM spintronics and neuromorphic computing.

Acknowledgements.
This work was supported by the Japan Society for the Promotion of Science (KAKENHI Grant Nos. 20H02602 and 19K15445), Tokuyama Science Foundation, the Hori Science and Arts Foundation, and Kyosho Hatta Foundation. Part of this work was carried out under the Cooperative Research Project Program of the Research Institute of Electrical Communication, Tohoku University.

V Appendix

Appendix A Ordinary Hall effect of MGN films

Figure 6(a) shows the Hall resistivity ρxy\rho_{xy} as a function of external magnetic field for MGN films (c/a=0.9962c/a=0.9962) at various temperature before subtract the ordinary Hall effect. The temperature dependence of the Hall coefficient RHR_{H} is summarized in Fig. 6(b).

Refer to caption
Figure 6: (a) Hall resistivity ρxy\rho_{xy} as a function of external magnetic field for MGN films (c/a=0.9962c/a=0.9962) before subtract the ordinary Hall effect. (b) Temperature dependence of the Hall coefficient RHR_{H}.

Appendix B Sample dependence of ρxy\rho_{xy} switching width in SOT measurements

Figure 7 shows the sample dependence of ρxy\rho_{xy} switching width in SOT measurements as a function of cc lattice constant. The ρxy\rho_{xy} obtained by the field sweep measurements of unpatterned films is also plotted. While no ρxy\rho_{xy} switching is observed for samples with cc lattice constants longer than 0.3890 nm, ρxy\rho_{xy} switching is observed in samples with shorter cc lattice constants, which is consistent with the results of AHE measurements of unpartterned films. On the other hand, although no large difference in cc lattice constant between 0.3882 and 0.3887 nm among SOT films in which ρxy\rho_{xy} switching were observed, ρxy\rho_{xy} switching width shows sample dependence. Since the amplitude of ρxy\rho_{xy} is strongly related to c/ac/a ratio as discussed in the main text, the sample dependence of ρxy\rho_{xy} switching width probably due to local variations in the quality and/or strain of the thin films.

Refer to caption
Figure 7: Sample dependence of ρxy\rho_{xy} switching width in SOT measurements for MGN(20 nm)/HM(3 bn) bilayers. MGN/Pt and MGN/Ta results presented in the main text are the devices Pt#1 and Ta#1, respectively. For Pt#2 device, ρxy,MGN\rho_{xy,\rm{MGN}} was derived by assuming that the current flows through MGN and Pt in the same proportion as in Pt#1 device.

Appendix C Comparison of ρxy\rho_{xy} width between field-sweep and SOT measurements using same device

Figure 8 shows the results of the Hall and SOT measurements using the same SOT device. Unfortunately, since our superconducting magnet has limited bore sizes, we cannot measure AHE of SOT devices. On the other hand, as shown in Fig. 8(a), the nonlinear Hall effect was observed within ±1.5\pm 1.5 T range in SOT devices [12]. Although the hysteresis was not observed probably due to insufficient magnetic field, we confirmed that the change of Hall resistance RxyR_{xy} is similar between field sweep and SOT measurements in the same sample as shown in Fig. 8(b).

Refer to caption
Figure 8: Hall and SOT measurements using the same SOT device. (a) RxyR_{xy} after subtract estimated ordinary Hall effect vs external magnetic field of MGN/Pt bilayers. (b) RxyR_{xy} vs pulse number of MGN/Pt bilayers. All measurements were performed at room temperature using the same MGN/Pt Hall device. The data are reproduced from Appl.Phys.Lett.Appl.Phys.Lett. 115, 052403 (2019) [12], with the permission of AIP Publishing.

References

  • [1] S. Nakatsuji, N. Kiyohara, T. Higo, Large anomalous Hall effect in a non-collinear antiferromagnet at room temperature, Nature 527, 212 (2015).
  • [2] T. Jungwirth, X. Martí, P. Wadley, J. Wunderlich, Antiferromagnetic spintronics, Nat. Nanotechnol. 11, 231 (2016).
  • [3] V. Baltz, A. Manchon, M. Tsoi, T. Moriyama, T. Ono, Y. Tserkovnyak, Antiferromagnetic spintronics, Rev. Mod. Phys. 90, 015005 (2018).
  • [4] H. Bai, X. Zhou, Y. Zhou, X. Chen, Y. You, F. Pan, C. Song, Functional antiferromagnets for potential applications on high-density storage and high frequency, J. Appl. Phys. 128, 210901 (2020).
  • [5] S. Bhatti, R. Sbiaa, A. Hirohata, H. Ohno, S. Fukami, S. N. Piramanayagam, Spintronics based random access memory: a review, Mater. Today 20, 530 (2017).
  • [6] J. Grollier, D. Querlioz, K. Y. Camsari, K. Everschor-Sitte, S. Fukami, M. D. Stiles, Neuromorphic spintronics, Nat. Electron. 3, 360 (2020).
  • [7] P. Rzeszut, J. Checiński, I. Brzozowski, S. Zietek, W. Skowroński, T. Stobiecki, Multi-state MRAM cells for hardware neuromorphic computing, arXiv:2102.03415, (2021).
  • [8] X. Z. Chen, R. Zarzuela, J. Zhang, C. Song, X. F. Zhou, G. Y. Shi, F. Li, H. A. Zhou, W. J. Jiang, F. Pan, Y. Tserkovnyak, Antidamping-Torque-Induced Switching in Biaxial Antiferromagnetic Insulators, Phys. Rev. Lett. 120, 207204 (2018).
  • [9] L. Baldrati, O. Gomonay, A. Ross, M. Filianina, R. Lebrun, R. Ramos, C. Leveille, F. Fuhrmann, T. R. Forrest, F. Maccherozzi, S. Valencia, F. Kronast, E. Saitoh, J. Sinova, M. Kläui, Mechanism of Néel Order Switching in Antiferromagnetic Thin Films Revealed by Magnetotransport and Direct Imaging, Phys. Rev. Lett. 123, 177201 (2019).
  • [10] T. Moriyama, K. Oda, T. Ohkochi, M. Kimata, T. Ono, Spin torque control of antiferromagnetic moments in NiO, Sci. Rep. 8, 14167 (2018).
  • [11] P. Wadley, S. Reimers, M. J. Grzybowski, C. Andrews, M. Wang, J. S. Chauhan, B. L. Gallagher, R. P. Campion, K. W. Edmonds, S. S. Dhesi, F. Maccherozzi, V. Novak, J. Wunderlich, T. Jungwirth, Current polarity-dependent manipulation of antiferromagnetic domains, Nat. Nanotechnol. 13, 362 (2018).
  • [12] T. Hajiri, S. Ishino, K. Matsuura, H. Asano, Electrical current switching of the noncollinear antiferromagnet Mn3GaN, Appl. Phys. Lett. 115, 052403 (2019).
  • [13] H. Tsai, T. Higo, K. Kondou, T. Nomoto, A. Sakai, A. Kobayashi, T. Nakano, K. Yakushiji, R. Arita, S. Miwa, Y. Otani, S. Nakatsuji, Electrical manipulation of a topological antiferromagnetic state, Nature 580, 608 (2020).
  • [14] H. Chen, Q. Niu, A. H. Macdonald, Anomalous Hall Effect Arising from Noncollinear Antiferromagnetism, Phys. Rev. Lett. 112, 017205 (2014).
  • [15] G. Gurung, D.-F. Shao, T. R. Paudel, E. Y. Tsymbal, Anomalous Hall conductivity of noncollinear magnetic antiperovskites, Phys. Rev. Mater. 3, 044409 (2019).
  • [16] V. T. N. Huyen, M.-T. Suzuki, K. Yamauchi, T. Oguchi, Topology analysis for anomalous Hall effect in the noncollinear antiferromagnetic states of MnA3{}_{3}AN (A=A= Ni, Cu, Zn, Ga, Ge, Pd, In, Sn, Ir, Pt), Phys. Rev. B 100, 094426 (2019).
  • [17] I. Samathrakis, H. Zhang, Tailoring the anomalous Hall effect in the noncollinear antiperovskite Mn3GaN, Phys. Rev. B 101, 214423 (2020).
  • [18] K. Zhao, T. Hajiri, H. Chen, R. Miki, H. Asano, P. Gegenwart, Anomalous Hall effect in the noncollinear antiferromagnetic antiperovskite Mn3Ni1-xCuxN, Phys. Rev. B 100, 045109 (2019).
  • [19] R. Miki, K. Zhao, T. Hajiri, P. Gegenwart, H. Asano, Epitaxial growth and orientation-dependent anomalous Hall effect of noncollinear antiferromagnetic Mn3Ni0.35Cu0.65N films, J. Appl. Phys. 127, 113907 (2020).
  • [20] D. Boldrin, I. Samathrakis, J. Zemen, A. Mihai, B. Zou, F. Johnson, B. D. Esser, D. W. McComb, P. K. Petrov, H. Zhang, L. F. Cohen, Anomalous Hall effect in noncollinear antiferromagnetic Mn3NiN thin films, Phys. Rev. Mater. 3, 094409 (2019).
  • [21] Y. You, H. Bai, X. Chen, Y. Zhou, X. Zhou, F. Pan, C. Song, Room temperature anomalous Hall effect in antiferromagnetic Mn3SnN films, Appl. Phys. Lett. 117, 222404 (2020).
  • [22] C. C. Chiang, S. Y. Huang, D. Qu, P. H. Wu, C. L. Chien, Absence of Evidence of Electrical Switching of the Antiferromagnetic Néel Vector, Phys. Rev. Lett. 123, 227203 (2019).
  • [23] B. J. Jacot, G. Krishnaswamy, G. Sala, C. O. Avci, S. Vélez, P. Gambardella, C.-H. Lambert, Systematic study of nonmagnetic resistance changes due to electrical pulsing in single metal layers and metal/antiferromagnet bilayers, J. Appl. Phys. 128, 173902 (2020).
  • [24] M. Meinert, D. Graulich, T. Matalla-Wagner, Electrical Switching of Antiferromagnetic Mn2Au and the Role of Thermal Activation, Phys. Rev. Appl. 9, 064040 (2018).
  • [25] H. Yan, Z. Feng, P. Qin, X. Zhou, H. Guo, X. Wang, H. Chen, X. Zhang, H. Wu, C. Jiang, Z. Liu, Electric-Field-Controlled Antiferromagnetic Spintronic Devices, Adv. Mater. 32, 1905603 (2020).
  • [26] S. Ishino, J. So, H. Goto, T. Hajiri, H. Asano, Preparation and evaluation of Mn3GaN1-x thin films with controlled N compositions, AIP Adv. 8, 056312 (2018).
  • [27] D. Fruchart, E. F. Bertaut, Magnetic Studies of the Metallic Perovskite-Type Compounds of Manganese, J. Phys. Soc. Jpn. 44, 781 (1978).
  • [28] K. Shi, Y. Sun, J. Yan, S. Deng, L. Wang, H. Wu, P. Hu, H. Lu, M. I. Malik, Q. Huang, C. Wang, Baromagnetic Effect in Antiperovskite Mn3Ga0.95N0.94 by Neutron Powder Diffraction Analysis, Adv. Mater. 28, 3761 (2016).
  • [29] T. Hajiri, H. Goto, H. Asano, Scaling the electrical current switching of exchange bias in fully epitaxial antiferromagnet/ferromagnet bilayers, Phys. Rev. B 102, 014404 (2020).
  • [30] P. Lukashev, R. F. Sabirianov, K. Belashchenko, Theory of the piezomagnetic effect in Mn-based antiperovskites, Phys. Rev. B 78, 184414 (2008).
  • [31] J. Zemen, Z. Gercsi, K. G. Sandeman, Piezomagnetism as a counterpart of the magnetovolume effect in magnetically frustrated Mn-based antiperovskite nitrides, Phys. Rev. B 96, 024451 (2017).
  • [32] Y. Yamane, O. Gomonay, J. Sinova, Dynamics of noncollinear antiferromagnetic textures driven by spin current injection, Phys. Rev. B 100, 054415 (2019).
  • [33] N. H. D. Khang, S. Nakano, T. Shirokura, Y. Miyamoto, P. N. Hai, Ultralow power spin–orbit torque magnetization switching induced by a non-epitaxial topological insulator on Si substrates, Sci. Rep. 10, 12185 (2020).
  • [34] T. Moriyama, W. Zhou, T. Seki, K. Takanashi, T. Ono, Spin-Orbit-Torque Memory Operation of Synthetic Antiferromagnets, Phys. Rev. Lett. 121, 167202 (2018).
  • [35] M. Dunz, T. Matalla-Wagner, M. Meinert, Spin-orbit torque induced electrical switching of antiferromagnetic MnN, Phys. Rev. Res. 2, 013347 (2020).
  • [36] G. Gurung, D.-Fu Shao, E. Y. Tsymbal, Spin-torque switching of noncollinear antiferromagnetic antiperovskites, Phys. Rev. B 101, 140405(R) (2020).
  • [37] S. Fukami, C. Zhang, S. DuttaGupta, A. Kurenkov, H. Ohno, Magnetization switching by spin–orbit torque in an antiferromagnet–ferromagnet bilayer system, Nature Mater. 15, 535 (2016).