Electrical current switching of the noncollinear antiferromagnet Mn$_3$GaN

TL;DR

This study demonstrates room-temperature electrical current switching of noncollinear antiferromagnetic Mn$_3$GaN/Pt bilayers, showing efficient manipulation of Hall resistance with significantly lower current densities than in single films, advancing AFM spintronics.

Contribution

It reveals that Pt layers enable electrical switching of noncollinear AFM Mn$_3$GaN at lower currents, offering a new platform for AFM spintronics applications.

Findings

Hall resistance can be manipulated by current pulses in bilayers

Lower critical current density compared to single films

Pt layer is essential for switching behavior

Abstract

We report electrical current switching of noncollinear antiferromagnetic (AFM) Mn$_3$GaN/Pt bilayers at room temperature. The Hall resistance of these bilayers can be manipulated by applying a pulse current of $1.5\times10^6$~A/cm$^2$, whereas no significant change is observed up to $\sim10^8$~A/cm$^2$ in Mn$_3$GaN single films, indicating that the Pt layer plays an important role. In comparison with ferrimagnetic Mn$_3$GaN/Pt bilayers, a lower electrical current switching of noncollinear AFM Mn$_3$GaN is demonstrated, with a critical current density two orders of magnitude smaller. Our results highlight that a combination of a noncollinear AFM antiperovskite nitride and a spin-torque technique is a good platform of AFM spintronics.

Full text

††preprint: AIP/123-QED

Antiferromagnets (AFMs) are promising materials for spintronics application owing to a number of properties, such as the absence of stray magnetic fields, terahertz spin dynamics, and stability against external perturbations, and, in this context, electrical manipulation of the Néel vector is an important emerging technique.AFM_spintronics1 ; AFM_spintronics2 ; AFM_spintronics3 ; Romain_Nature Recent experiments have demonstrated electrical current switching of the AFM Néel vector by Néel spin–orbit torque (SOT) in CuMnAsCuMnAs_SOT ; CuMnAs_SOT2 and Mn2As,Mn2Au_SOT ; Mn2Au_SOT2 and by spin torque in NiO/Pt bilayersPRL_NiO_SOT ; Lorenzo_NiO_SOT and Pt/NiO/Pt trilayers.Moriyama_NiO_SOT These experiments show that the AFM Néel vector can be manipulated by electrical current densities of the order of $10^{7}$–$10^{8}$A/cm2, which are similar to the current densities required with SOT in typical ferromagnet (FM)/heavy metal (HM) bilayers. On the other hand, according to an early study of AFM spintronics,AFM_STT_theory1 a current density two orders of magnitude smaller may theoretically be capable of manipulating the AFM Néel vector. Indeed, current-induced switching of exchange bias at such lower current densities has been realized in AFM/FM bilayers.STT_exchange_bias1 ; STT_exchange_bias2 ; STT_exchange_bias3 These results suggest the possibility of switching of the AFM Néel vector using lower electrical currents.

Recently, noncollinear AFMs become one of the most fascinating classes among AFMs, because they exhibit a large magneto-transport and a novel spin transport phenomena.Hua_Chen_PRL ; orbital_AHE1 ; orbital_AHE2 ; noncollinear_spin However, the spin-torque switching of the AFM Néel vector is demonstrated in only collinear AFM, while the theoretical study shows that the domain of noncollinear AFM can be manipulated by spin current with realistic current density of $\sim 10^{7}$A/cm2.Yamane_noncollinear Thus, noncollinear AFMs call for the spin current injection thorough spin-Hall effect. The antiperovskite manganese nitrides Mn3GaN (MGN) is a good platform to investigate the electrical current manipulation of noncollinear AFMs, as well as for comparisons of the critical current density between FM and AFM materials, because MGN is a noncollinear AFM while a perpendicularly magnetized ferrimagnet can be obtained by applying tetragonal distortion.NIMS_MGN1 ; NIMS_MGN2 ; MGN_growth Both the Néel and Curie temperatures are reported to be above 380 KNeel_MGN1 and 660–740 K,NIMS_MGN1 ; NIMS_MGN2 respectively, which are relatively high for this class of materials.

In the antiperovskite manganese nitrides Mn${}_{3}A$N (where $A=$ Ni, Ga, Sn, etc.), the Mn atoms form a kagome lattice in the (111) plane. These kagome lattices with nonzero Berry curvature may provide an interesting magneto-transport phenomena such as a large anomalous Hall effect (AHE) even with quite small canted magnetization of the order of $10^{-3}\mu_{\mathrm{B}}$ per atom, Hua_Chen_PRL ; orbital_AHE1 ; orbital_AHE2 although the AHE is empirically proportional to net magnetization in conventional FMs.AHE1 The AHE has been realized in the noncollinear AFMs Mn${}_{3}X$ ($X$ = Sn, Ge, Pt) where the Mn atoms form a kagome lattice along different crystalline planes.Mn3Sn ; Mn3Ge ; Mn3Pt Since noncollinear AFM Mn${}_{3}A$N has nonvanishing Berry curvature, the AHE can be realized.MGN_AHE ; Cu-MNN_AHE ; Mn3NiN_AHE This is an other motivation to study antiperovskite manganese nitrides.

In this letter, we investigate the spin-torque switching of noncollinear AFM MGN. We observe significant changes in the Hall resistance when current pulses are applied to AFM-MGN/Pt bilayers at room temperature, while no clear change is observed in the longitudinal resistance. By comparing AFM-MGN single films and FM-MGN/Pt bilayers, we show that the spin torque plays an important role in AFM-MGN/Pt bilayers, the critical current density for which is two orders of magnitude smaller than that for FM-MGN/Pt bilayers. The possible origin of changes of the Hall resistance is discussed.

High-quality epitaxial AFM-MGN and FM-MGN films were prepared on MgO (001) substrates by reactive magnetron sputtering using a Mn70Ga30 target under an Ar/N2 atmosphere. As reported in previous work,MGN_growth AFM- and FM-MGN can be controllable by both the deposition rate and N2 gass partial pressure. The AFM-MGN films were grown at 450 *∘*C substrate temperature using 9 % N2 + 91 % Ar gas mixtures with the deposition rate of 1.2 nm/min. On the other hand, the FM-MGN films were grown at 450 *∘*C substrate temperature using 0.25 % N2 + 99.75 % Ar gas mixtures with the deposition rate of 3.1 nm/min. After the film growth was complete, we deposited a Pt layer (3 nm) by magnetron sputtering at room temperature. The crystal structure was analyzed using out-of-plane X-ray diffraction (XRD) measurements with Cu $K\alpha$ radiation. The magnetic properties of the MGN/Pt bilayers were characterized using SQUID magnetometry. Transport properties were measured using Hall bars ($20~{}\upmu\mathrm{m}\times 100~{}\upmu\mathrm{m}$) prepared by a conventional photolithographic process. The current is injected to MGN(bottom)/Pt(top) bilayers through the contact pads.

Figure 1(a) shows the out-of-plane XRD patterns of 10 nm thick AFM-MGN and FM-MGN films on MgO substrates. Only the (002) AFM- and FM-MGN peaks exhibit Bragg peaks. The out-of-plane lattice constants of AFM- and FM-MGN are 0.3885 nm and 0.3680  nm, respectively. The obtained out-of-plane lattice constants are consistent with those found for AFM- and FM-MGN in previous studies.NIMS_MGN1 ; NIMS_MGN2 ; MGN_growth ; MGN_growth ; Neel_MGN1 The AHE loops of AFM- and FM-MGN/Pt bilayers are shown in Figs. 1(b) and 1(c), respectively. A hysteresis loop is clearly observed for the FM-MGN/Pt bilayers. For AFM-MGN/Pt bilayers, a nonlinear AHE curve is obtained, while no hysteresis loop is visible. To determine the magnetic properties of AFM-MGN/Pt bilayers, the temperature dependences of the resistivity and magnetization were measured. In the resistivity curve, there is a clear anomaly at 200 K and a kink at 380 K, as shown in Fig. 1(d). It is well known that the antiperovskite nitrides Mn${}_{3}A$N exhibit a kink at the AFM transition.Tashiro_MGN ; Cu-MNN_AHE ; antipero_RT In the bulk materials, several values of the Néel temperature have been reported, ranging from 280 K to more than 380 K.Neel_MGN1 ; Neel_MGN2 Thus, the Néel temperature of our films can be determined to be around 380 K. On the other hand, in the temperature dependence of the magnetization shown in Fig. 1(e), an FM-like transition is observed at 200 K. Since AFM-MGN/CoFe bilayers exhibit an exchange bias at low temperatures (not shown), we conclude that AFM and weak FM coexist below 200 K.

To test the possibility of electrical current switching of noncollinear AFMs, sequential write–read operations were performed. A sequence of write–read operations is shown in Fig. 2(b). After injection of a write current $I_{\mathrm{Pulse}}$ with 20 ms pulse width, the Hall resistance $R_{xy}$ was measured by a probe current $I_{\mathrm{Probe}}$ of approximately $(0.5$–$1)\times 10^{5}$ A/cm2. Between $I_{\mathrm{Pulse}}$ and $I_{\mathrm{Probe}}$, we set a delay of 4 s. All measurements presented here were performed without an external magnetic field. The change in Hall resistance $\Delta R_{xy}$ as a function of current density for AFM-MGN(10 nm)/Pt(3 nm) bilayers is shown in Fig. 2(c). $R_{xy}$ changes sharply and a clear hysteresis loop is observed at room temperature (294 K), indicating success of the electrical write–read operation. On the other hand, no change of $R_{xy}$ is observed at 4 K, suggesting that the appearance of weak FM may obstruct the electrical write operation in applied current density range. The other possibility is that AHE signal discussed below disappears by the appearance of weak FM, so that no change of $R_{xy}$ is observed. Figure 2(d) presents the current density dependence of $\Delta R_{xy}$ in FM-MGN(10 nm)/Pt(3 nm) bilayers in the absence of an external magnetic field. It is well known that, in principle, an external magnetic field parallel to the current is required for SOT in the case of a perpendicularly magnetized FM, so no switching behavior is observed. Thus, we also performed with an external magnetic field range of $-0.5\sim 0.5$ T. However, no significant switching was observed up to $\sim 9.5\times 10^{7}$ A/cm2 in FM-MGN(10 nm)/Pt(3 nm) bilayers, indicating that a larger electrical current is needed. Indeed, the critical current density is reported to be $(1.1$–$1.5)\times 10^{8}$ A/cm2 in ferrimagnetic MnGa/HM bilayers.MnGa_SOT1 ; MnGa_SOT2 To clarify the role of the Pt layer of AFM-MGN/Pt bilayers, we also performed the same operations on AFM-MGN(10 nm) single films. As can be seen in Fig. 2(e), no current density dependence is observed in these single films. This indicates that the Pt layer plays an important role as a pure spin current generator through a spin Hall effect like the SOT in FM/HM bilayers. From these results, we conclude that spin-torque switching occurs in AFM-MGN/Pt bilayers. From Fig. 2(c), the critical current density is estimated to be approximately $1.5\times 10^{6}$ A/cm2, which is one to two orders of magnitude smaller than that for typical FM/HM bilayersCoFeB_SOT ; Pt/Co_SOT and the threshold current for NiO/Pt spin-torque switching.Lorenzo_NiO_SOT ; Moriyama_NiO_SOT ; PRL_NiO_SOT In the present study, Pt thickness is fixed to be 3 nm. It is known that the thickness of the heavy metal layer can affect to the SOT in FM case,HM_thick_depe so that it is worth to investigate the Pt layer thickness dependence in the future study.

Next, we would like to discuss the possible cause of changes in $R_{xy}$. In general, the Hall signals contain the ordinary Hall effect, the planar Hall effect, and the AHE. The planar Hall effect is known to be one of anisotropic magnetoresistance and is used in electrical detection of the AFM Néel order of CuMnAs,CuMnAs_SOT ; CuMnAs_SOT2 Mn2Au,Mn2Au_SOT ; Mn2Au_SOT2 and IrMn.MnIr_AMR As for FM SOT,SOT_domain we assume that there are two types of AFM domains, which have high and low $R_{xy}$, respectively, and they can propagate under the influence of an electrical current, as shown in Fig. 3(a). In fact, this domain propagation is expected theoretically even in noncollinear AFMs.Yamane_noncollinear If the planar Hall effect is dominant in our bilayers, then the $V_{xx,\mathrm{II/III}}$ signal will not be equal to $V_{xx,\mathrm{I}}$ and it will change when the $V_{xy}$ signals change. On the other hand, if the AHE is dominant, then the $V_{xx,\mathrm{II/III}}$ signal will be equal to $V_{xx,\mathrm{I}}$,Cu-MNN_AHE while the $V_{xy,\mathrm{II/III}}$ signal will not be equal to $V_{xy,\mathrm{I}}$. Therefore, we record $R_{xx}$ and $R_{xy}$ simultaneously to distinguish these two possibilities.

Figure 3(b) shows the pulse count dependence of $R_{xy}$ (top) and $R_{xx}$ (bottom) for AFM-MGN/Pt bilayers. Here, $I_{\mathrm{Pulse}}=\pm 4.3\times 10^{6}$ A/cm2 with 20 ms pulse width was injected at each count. After injection of $I_{\mathrm{Pulse}}$, $R_{xy}$ was measured by a probe current $I_{\mathrm{Probe}}$ of about $3\times 10^{5}$ A/cm2 with a delay of 4 s from $I_{\mathrm{Pulse}}$. This procedure is the same as in Fig. 2, although a different bilayer was measured, so $\Delta R_{xy}$ is different. We note that almost the same critical current density was obtained for the several different AFM-MGN/Pt bilayers we studied, but $\Delta R_{xy}$ shows a large sample dependence, probably because of local variations in the quality of the AFM-MGN film. $R_{xy}$ jumps between $178.05\pm 0.06$ m$\Omega$ and $171.34\pm 0.04$ m$\Omega$ on $\pm I_{\mathrm{Pulse}}$ injection, except for a single error at the 21st injection. The obtained $\Delta R_{xy}$ and $\Delta R_{xy}$ ratio are 6.71 m$\Omega$ and 3.84 $\%$, respectively, which are enough larger than the experimental errors. At the same time, $R_{xx}$ shows small oscillations but no clear jumps on $\pm I_{\mathrm{Pulse}}$ injection. If the domains propagate from state I to state II, then $R_{xx}$ will increase/decrease monotonically with respect to the domain wall position. On the other hand, if the domains propagate from state I to state III, then $R_{xx}$ will jump between two values like $R_{xy}$. Since no such increasing/decreasing or jump behavior of $R_{xx}$ is observed on $\pm I_{\mathrm{Pulse}}$ injection, the planar Hall effect can be excluded as a cause of $\Delta R_{xy}$.

The AHE in noncollinear AFMs is theoretically expected and experimentally realized in antiperovskite Mn3NiN systems.Cu-MNN_AHE ; Mn3NiN_AHE Although AFM-MGN is also theoretically expected to exhibit the AHE,MGN_AHE there have been no reports of this to date. As shown in Fig. 1(b), a nonlinear AHE curve is obtained for AFM-MGN/Pt bilayers, showing that the AHE possibly occurs for these. These results may suggest that the magnetic anisotropy of AFM-MGN is too large to manipulate by the application of an external magnetic field, which deserves future study.

After several hundred cycles of $\pm I_{\mathrm{Pulse}}$ injection, we finally performed continuous $+I_{\mathrm{Pulse}}$ or $-I_{\mathrm{Pulse}}$ write operations. The results for AFM-MGN/Pt bilayers are shown in Fig. 3(c). The $R_{xy}$ switching still survives even after several hundred cycles of write–read operations. $R_{xy}$ increases (decreases) on continuous $+I_{\mathrm{Pulse}}$ ($-I_{\mathrm{Pulse}}$) injection and then exhibits saturation. As discussed for SOT of synthetic AFMs,synthetic_AFM_SOT the observed asymptotic $R_{xy}$ behavior can be fitted by an exponential decay function $y=y_{0}+Ae^{-(x+x_{0})/\tau}$. The time constants $\tau$ for $\pm I_{\mathrm{Pulse}}$ injection are different: $\tau=21.7$ for $+I_{\mathrm{Pulse}}$ and 35.6 for $-I_{\mathrm{Pulse}}$. This behavior has not been reported for other AFM/HM bilayers.Lorenzo_NiO_SOT ; synthetic_AFM_SOT $R_{xx}$ was also recorded at the same time, as shown at the bottom of Fig. 3(c). However, since the room temperature decreased monotonically during the continuous $\pm I_{\mathrm{Pulse}}$ injection measurements, so did $R_{xx}$. On the other hand, there is no clear relationship between $R_{xy}$ and decreasing room temperature, showing the stability against temperature perturbations. Together with the stability against write operations, these results highlight the potential application of this approach.

In summary, we have shown the possibility of spin-torque switching of noncollinear AFM-MGN at room temperature. A significant change in Hall resistance is observed on the application of current pulses to AFM-MGN/Pt bilayers, while no clear change is observed in AFM-MGN single films, indicating that spin torque plays an important role in manipulating the Hall resistance. We have demonstrated that the critical current density for AFM-MGN is two orders of magnitude smaller than that for FM-MGN and that $R_{xy}$ switching can survive even after several hundred cycles of write–read operations. By simultaneous recording of $R_{xx}$ and $R_{xy}$ as functions of applied pulse counts, possible evidence of the AHE in AFM-MGN has been found. These results open the pathway to the efficient control of noncollinear AFM order.