WorksPhysics and Astronomy
Néel vector switching by magnetization-dependent layer-resolved spin-orbit torque
Abstract
Layer-resolved spin-orbit torques offer a route to controlling antiferromagnetic order through local spin-charge conversion. First-principles calculations show that the layer-resolved spin Hall conductivity of the antiferromagnetic topological insulator consists of a magnetization-independent spin Hall term plus two spin anomalous Hall terms that are even in the Néel vector yet evolve with the local magnetic order, while local inversion asymmetry differentiates surface and inner layers. This response enables reversible, field-free switching of the Néel vector in finite-septuple-layer films. Although the magnetization-independent spin Hall term alone can reverse the order, the spin anomalous Hall terms provide magnetization-dependent feedback that redistributes the layer-resolved torques and reshapes the switching trajectory, offering a route to tailoring current-driven switching through layer and interface engineering.
Manipulation of the magnetization by spin-orbit torque (SOT) stands as a crucial advancement in spintronics due to its merits in realization of ultrafast switching as well as low-power writing operation [4] [1, 2, 3]. In general, SOT originates from the spin current (or spin-polarized current) applied to the magnet, and is given by
where is the unit vector for the local moment and is the secondary spin-torque coefficient. The spin current vector is composed by the product of the magnitude of the spin current and the spin polarization direction. It is intriguing that the spin current direction is not manifestly present in Eq. (1). In conventional SOT devices based on heavy-metal/ferromagnet (HM/FM) heterostructures, the spin current is generated by the spin Hall effect (SHE) in the HM layer [5, 6, 7, 8], so that the direction is manifested in the spin-injected direction to the FM layer, as well as in its decay with thickness. In magnets that spontaneously generate SOT [9, 10, 11], the direction of the spin current produced by the intrinsic SHE becomes irrelevant. Instead, the magnitude and polarization direction of the spin current determine the resulting SOT of which the spin current vector given by
where is the spin Hall conductivity (SHC), is the anomalous Hall conductivity (AHC), and is the applied electric field along the longitudinal direction . In particular, the characterization of the spin polarization axis of SHE in inherently inhomogeneous magnets such as antiferromagnets (AFM) is both crucial and challenging [12, 13]. The low conductivities and ultrafast spin dynamics of many AFM render them not only outstanding SOT sources, but also ideal tunable media [10, 14, 15, 16, 17].
In this Letter, we investigated the layer-resolved SHE in the AFM topological insulator MnBiTe [21, 22, 23], in which the topological surface states can provide extremely high SOT with small electric current due to the bulk insulating inside the topological band gap [9]. We discovered the magnetization-dependent SHE where the layer-resolved SHC (LSHC) is given by
where is the normalized local moment of AFM with sublattice and , operator represents Hadamard product. As illustrated in Figure 1, the first term in Eq. (3) represents the conventional SHE contribution, completely independent of the local magnetic state[24, 25]. The second and the third term, corresponding to the spin anomalous Hall effect (SAHE)[26, 27, 28, 29], are both even functions of , indicating both AFM sublattices with opposite sign of magnetization yield the same SHC. By quantifying the coefficient vectors in Eq. (3) as well as the AHC for several-layered systems, we then elucidated the distinct behaviors of LSHE-induced SOT (LSOT) in the surface and interior layers of MnBiTe, and further demonstrated its field-free robust control via LSOT over the AFM order through spin dynamics simulations.
Figure 1. Schematic of the Néel vector orientation-dependent spin response in MnBiTe. (a) Response of spin polarization in Eq. (3). is orientation independent; changes in magnitude and vanishes for in-plane order, whereas changes in both magnitude and direction. Red and blue arrows denote antiparallel local moments, and orange arrows denote response vectors. (b) An electric current along generates a transverse spin current along , with opposite -polarized carriers shown in red and blue.
The electronic band structures of the finite septuple layers (SL) MnBiTe were obtained by employing density-functional theory (DFT) calculations with projector-augmented wave pseudopotentials implemented in the Vienna ab initio simulation (VASP) package[30, 31]. The generalized gradient approximation (GGA) in Perdew, Burke, and Ernzerhof (PBE) formation is used as the exchange-correlation energy[32, 33], and the Hubbard U method ( eV, eV) is applied on Mn () orbitals to include strong-correlation effects[34]. An energy cutoff of 600 eV was used for the plane-wave expansion throughout the calculations. The -centered -points meshes were adopted for the calculations of few-layer hexagonal lattices with a vacuum space of Å. For structural relaxations we both relaxed the atoms until the Hellmann-Feynman forces are less than 1 meV/Å. The DFT-D3 method is employed to treat the van der Waals interaction[35]. The tight-binding Hamiltonian in a Wannier-function (WF) basis is constructed by performing a unitary transformation implemented in WANNIER90[36]. The Kubo formula implemented in Wannier90 code is used to calculate the -resolved spin Berry curvature, which is given as[37, 38]
The intrinsic LSHC is calculated using following formulas[39]
where represents the Fermi-Dirac distribution function, , is the smearing parameter and are the band energy for each , is the primitive cell volume, and is the number of points in the BZ. For a layer or a group of layers , the projection operator is defined as , where denotes a localized Wannier orbital assigned to [40]. The corresponding layer-resolved Hermitian spin-current operator is .
Further computational details are provided in the Supplemental Material.
Bulk MnBiTe is a van der Waals AFM with a rhombohedral structure (space group ) and spatial inversion symmetry. Each SL follows the Te–Bi–Te–Mn–Te–Bi–Te stacking sequence. The Mn moments order ferromagnetically within each SL and couple antiferromagnetically between adjacent SLs along the axis[41, 42]. This out-of-plane -type AFM order breaks the mirror symmetry and time-reversal symmetry while preserving the combined symmetry , where is a half-lattice translation. Each SL also retains a threefold rotation and a mirror plane .
For the geometry considered here, an electric field applied along the direction generates transverse Hall currents along , as shown in Figure 1(b). To characterize the spin polarization carried by these currents, we obtained the spin polarization based on layer-resolved AHC (LAHC) and LSHC for each SL via [43, 44], where is the layer-resolved spin-polarized current in the spin up/down channels, and and are the LSHC and LAHC of layer , respectively. Figure 2 shows the resulting LSHC and layer-resolved polarization profile in 4 SLs MnBiTe with the Néel vector oriented along the axis. The component is negligible in all four layers, whereas alternates in sign between adjacent SLs, consistent with the staggered AFM order. By contrast, has the same sign in each surface layer and its adjacent subsurface layer, reverses between the two inner layers, and has opposite signs on the top and bottom surfaces. Even in the high symmetric configuration with the Néel vector oriented along the axis, the layer-resolved polarization of SHC is tilted in plane because of the lack of the mirror plane along . It is expected that when deviates from the axis, the lower magnetic symmetry allows richer spin polarization from the magnetization-dependent LSHC.
Figure 2. In 4 SLs MnBiTe with the Néel vector oriented along the axis, (a) schematic of the layer-resolved spin polarization in 4 SLs MnBiTe. Yellow spheres with arrows represent the local spin polarization directions. (b) LSHC components , , and evaluated within the band gap, (c) the corresponding layer-resolved spin polarization components, and (d) layer-resolved magnetic moment .
LSHC inside the topological band gap dependent on the Néel vector direction is then investigated in both 2 SLs and 4 SLs MnBi₂Te₄. As shown in Figure 3(a)–(Figure 3c), in the 2 SLs case, the component of SHC remains zero for all Néel vector directions. and exhibit the same symmetric angular dependence with the rotation of the Néel vector in the or plane. Specifically, varies from approximately for to for in-plane orientations, whereas changes from approximately for to for in-plane orientations. This indicates that the SHC in the 2 SLs case is almost independent of the in-plane Néel vector components, which is also verified in Figure 3(c). Considering that the SHC is periodic in the zenith angle and independent of the azimuthal angle, we can thus obtain the magnetization-dependent spin anomalous Hall conductivity (SAHC) as:
Figure 3. Magnetization dependence of the in-gap SHC for 2 SLs and 4 SLs MnBi₂Te₄. (a)–(c) Total SHC of 2 SLs MnBi₂Te₄; (d)–(f) and (g)–(i) detail the separated SHC contributions from the surface (layers 1 and 4) and inner (layers 2 and 3) layers of 4 SLs MnBi₂Te₄, respectively. Columns correspond to the Néel vector rotated within the , , and planes. Blue triangles, orange diamonds, and red circles denote , respectively. Symbols denote the first-principles results, while lines show the fits.
This term is an even function of . The corresponding coefficient vector is given in Table 1.
| System / Layer | Tilt angle | |||
| 2 SLs (Total) | ||||
| 4 SLs surface | ||||
| 4 SLs inner |
Table 1. Fitted coefficient vectors (in units of ) with local response axis tilt angles for the total response of a 2 SLs MnBiTe and the layer-resolved contributions of a 4 SLs MnBiTe (surface and inner layers).
For 4 SLs MnBi₂Te₄, the LSHC contributions from the surface layers (1 and 4) [Figure 3(d)–(Figure 3f)] and the inner layers (2 and 3) [Figure 3(g)–(Figure 3i)] are investigated separately. In contrast to 2 SLs, 4 SLs MnBi₂Te₄ develops a finite component in the rotation plane. is an odd function with respect to the direction of in the plane. In the rotation plane in which is zero, the symmetry point of and shows a clear offset from the principal crystallographic directions. It is due to the in-equivalence of the and plane in MnBi₂Te₄. With the Néel vector lying in the plane, the in-plane magnetic moment component is invariant under the mirror symmetry operation . In contrast, the in-plane component of the Néel vector lying in the plane changes the sign under mirror symmetry . Consequently, the symmetry of the total SHC is broken in the rotation plane. Therefore, the offset is described by a local response axis tilted within the plane and the corresponding SAHC is given by:
The term takes a form similar to the SAHE found in certain ferromagnetic materials[26, 45]. Fits to Eq. (6) yield tilt angles of approximately for the surface-layer contribution and for the inner-layer contribution (Table I). The results of coefficient vectors in Table I show that the inner layers exhibit a larger magnetization-dependent spin Hall response than the surface layers, including 2 SLs cases. This also indicates that the topological surface states themselves are not only insensitive to magnetic impurities, but the SHC arising from them is also sensitive to the magnetic direction or Néel vector direction. This is similar to the magnetic insensitive SHC in the magnetic topological material MnSb₂Te₄[9].
To examine the dynamical consequences of the layer-resolved spin response, we performed spin dynamics simulations of 4 SLs MnBi₂Te₄ using GPU-supported micromagnetic packages MicroMagnetic.jl[46]. Figure 4(a) shows the layer-resolved spin dynamics of 4 SLs MnBi₂Te₄ under periodic bipolar spin current pulses. The spin current generated by the LSHE for each SL is denoted by , is the Hall current density and is the cross-sectional area perpendicular to the current direction. A positive current pulse switches the Néel vector of each layer along the spin-polarization direction, converting the initial configuration into the reversed state. Once the pulse is turned off, the interlayer antiferromagnetic exchange coupling drives further relaxation and stabilizes this switched configuration. A subsequent negative current pulse then flips the layer magnetization into the opposite direction, eventually restoring the initial state. The transient oscillations during the switching process reflect the dynamic competition between current-driven reorientation and interlayer exchange relaxation [47, 48]. Spin dynamic simulations of 6 SLs MnBiTe provide similar results [see details in the Supplemental Material].
Figure 4. (a) Layer-resolved Néel vector switching driven by LSOT in 4 SLs MnBiTe. The switching is induced by the complete spin response tensor, comprising , and . The upper panels show the time evolution of the out-of-plane magnetization component (–) in each SL under periodic bipolar current pulses with amplitudes of A m, whereas the lower panels show the corresponding current density. (b) The spin-flip frequency as a function of the spin current density under the situations that , and is included, only , is included, and only is included, respectively.
We then employ the spin-flip frequency as , where is the half spin-flip time to quantify the rate of magnetic domain switching. As shown in Figure 4(b), the increases as the driving current increases, indicating that the magnetic domains switch faster under a larger SOT.
SOT. However, it saturates at about A m, because this frequency is largely determined by the relaxation of the domains after the SOT is removed. The complete spin response provides the highest and the lowest critical current density for domain flipping. Further investigation on the LSOT exhibits a higher critical current density. It is due to the pronounced oscillations developing in the layer-resolved Néel vector during the current pulse, whereas the trajectory is much smoother when is omitted. The detailed studies are shown in the Supplemental Material. Overall, the switching efficiency is governed by layer dependence of the spin response, with the complete response enabling the fastest reversal.
It is worth noting that the simulations above utilized the spin current generated by LSHE, rather than the charge current. Given that the charge conductivity approaches zero within the topological band gap, the magnitude of the actual charge current employed would be one to two orders of magnitude smaller than that of the spin Hall current. Although precise experimental observation of the quantum anomalous Hall effect in two-dimensional magnetic topological materials remains challenging, the large spin Hall angle can persist even in the presence of neutral defects, owing to the robustness of topological edge states [9]. In other words, research on magnetic-related LSHE not only provides new insights for achieving low-power, high-efficiency spintronic control using two-dimensional magnetic topological materials, but also serves as an effective way to identify the topological edge states of magnetic topological materials.
Acknowledgments
This work was financially supported by the National Natural Science Foundation of China (12274309).
References
- [4]S. Reimers, Y. Lytvynenko, Y. R. Niu, E. Golias, B. Sarpi, L. S. I. Veiga, T. Denneulin, A. Kovács, R. E. Dunin-Borkowski, J. Bläser, M. Kläui, and M. Jourdan, Nature Communications 14, 1861 (2023).
- [5]L. Liu, O. J. Lee, T. J. Gudmundsen, D. C. Ralph, and R. A. Buhrman, Physical Review Letters 109, 096602 (2012).
- [6]L. Zhu, D. C. Ralph, and R. A. Buhrman, Physical Review Letters 122, 077201 (2019).
- [7]K. Ando, S. Takahashi, K. Harii, K. Sasage, J. Ieda, S. Maekawa, and E. Saitoh, Physical Review Letters 101, 036601 (2008).
- [8]B. F. Miao, S. Y. Huang, D. Qu, and C. L. Chien, Physical Review Letters 111, 066602 (2013).
- [9]Z. Li, S. Pan, S. Liu, Y. Zhang, A. Li, L. Ai, X. Xie, X. Cao, Z. Jia, X. Han, J. Yu, and F. Xiu, National Science Review 12, nwaf178 (2025).
- [10]H. Guo, Z. Li, J. Lu, C. Y. Guo, Y. Han, S. Sun, Y. Wu, W. Yang, D. Xiao, Z. Zhu, L. Peng, Y. Ye, Y. Hou, J. Yang, and Z. Luo, Nature Communications 16, 8911 (2025).
- [11]T. D. Skinner, K. Olejník, L. K. Cunningham, H. Kurebayashi, R. P. Campion, B. L. Gallagher, T. Jungwirth, and A. J. Ferguson, Nature Communications 6, 6730 (2015).
- [12]J. Železný, H. Gao, K. Výborný, J. Zemen, J. Mašek, A. Manchon, J. Wunderlich, J. Sinova, and T. Jungwirth, Physical Review Letters 113, 157201 (2014).
- [13]J. Železný, Y. Zhang, C. Felser, and B. Yan, Physical Review Letters 119, 187204 (2017).
- [14]P. Wadley, B. Howells, J. Železný, C. Andrews, V. Hills, R. P. Campion, V. Novák, K. Olejník, F. Maccherozzi, S. S. Dhesi, S. Y. Martin, T. Wagner, J. Wunderlich, F. Freimuth, Y. Mokrousov, J. Kuneš, J. S. Chauhan, M. J. Grzybowski, A. W. Rushforth, K. W. Edmonds, B. L. Gallagher, and T. Jungwirth, Science 351, 587 (2016).
- [15]K. Olejník, T. Seifert, Z. Kašpar, V. Novák, P. Wadley, R. P. Campion, M. Baumgartner, P. Gambardella, P. Němec, J. Wunderlich, J. Sinova, P. Kuzel, M. Müller, T. Kampfrath, and T. Jungwirth, Science Advances 4, eaar3566 (2018).
- [16]Y. Gueckstock, A. L. Chekhov, S. Y. Bodnar, O. Gueckstock, S. Reimers, Y. Lytvynenko, Y. Skourski, M. Wolf, T. S. Seifert, O. Gomonay, and others, Nature Communications 14, 6038 (2023).
- [17]J. Tang and R. Cheng, Physical Review Letters 132, 136701 (2024).
- [21]M. M. Otrokov, I. I. Klimovskikh, H. Bentmann, D. Estyunin, A. Zeugner, Z. S. Aliev, S. G. Gaf., A. Wolter, A. Koroleva, A. M. Shikin, and others, Nature 576, 416 (2019).
- [22]J. Li, Y. Li, S. Du, Z. Wang, B.-L. Gu, S.-C. Zhang, K. He, W. Duan, and Y. Xu, Science Advances 5, eaaw5685 (2019).
- [23]S. Liu, J.-X. Yu, E. Zhang, Z. Li, Q. Sun, Y. Zhang, L. Cao, L. Li, M. Zhao, P. Leng, X. Cao, A. Li, J. Zou, X. Kou, J. Zang, and F. Xiu, Nano Letters 24, 16 (2024).
- [24]S. Varotto, M. Cosset-Chéneau, C. Grézes, Y. Fu, P. Warin, A. Brenac, J.-F. Jacquot, S. Gambarelli, C. Rinaldi, V. Baltz, J.-P. Attané, L. Vila, and P. Noël, Physical Review Letters 125, 267204 (2020).
- [25]N. Soya, M. Yamada, K. Hamaya, and K. Ando, Physical Review Letters 131, 076702 (2023).
- [26]V. P. Amin, J. Li, M. D. Stiles, and P. M. Haney, Physical Review B 99, 202405 (2019).
- [27]S. Ihama, T. Taniguchi, K. Yakushiji, A. Fukushima, Y. Shiota, S. Tsunegi, R. Hiramatsu, S. Yuasa, Y. Suzuki, and H. Kubota, Nature Electronics 1, 120 (2018).
- [28]P. Li, J.-Z. Zhang, Z.-X. Guo, T. Min, and X. Wang, Sci. China Phys., Mech. Astron 66, 227511 (2022).
- [29]Y. Omori, E. Sagasta, Y. Niimi, M. Gradhand, L. E. Hueso, F. Casanova, and Y. Otani, Physical Review B 99, 014403 (2019).
- [30]G. Kresse and J. Furthmüller, Computational Materials Science 6, 15 (1996).
- [31]G. Kresse and J. Furthmüller, Physical Review B 54, 11169 (1996).
- [32]P. E. Blöchl, Physical Review B 50, 17953 (1994).
- [33]G. Kresse and D. Joubert, Physical Review B 59, 1758 (1999).
- [34]D. Zhang, M. Shi, T. Zhu, D. Xing, H. Zhang, and J. Wang, Physical Review Letters 122, 206401 (2019).
- [35]S. Grimme, J. Antony, S. Ehrlich, and H. Krieg, Journal of Chemical Physics 132, 154104 (2010).
- [36]A. A. Mostofi, J. R. Yates, G. Pizzi, Y.-S. Lee, I. Souza, D. Vanderbilt, and N. Marzari, Computer Physics Communications 185, 2309 (2014).
- [37]D. J. Thouless, M. Kohmoto, M. P. Nightingale, and M. den Nijs, Physical Review Letters 49, 405 (1982).
- [38]X. Wang, J. R. Yates, I. Souza, and D. Vanderbilt, Physical Review B 74, 195118 (2006).
- [39]J. Qiao, J. Zhou, Z. Yuan, and W. Zhao, Physical Review B 98, 214402 (2018).
- [40]S. Pan, X. Chen, H.-F. Huang, and J.-X. Yu, Physical Review B 113, 224416 (2026).
- [41]M. M. Otrokov, T. V. Menshchikova, M. G. Vergniory, I. P. Rusinov, A. Yu Vyazovskaya, Y. M. Koroteev, G. Bihlmayer, A. Ernst, P. M. Echenique, A. Arnau, and E. V. Chulkov, 2D Materials 4, 025082 (2017).
- [42]Y. Gong, J. Guo, J. Li, K. Zhu, M. L. Xia, X. Liu, Q. Zhang, L. Gu, L. Tang, X. Feng, D. Zhang, W. Li, C. Song, L. Wang, P. Yu, X. Chen, Y. Wang, H. Yao, W. Duan, Y. Xu, S.-C. Zhang, M. Ma, Q.-K. Xue, and K. He, Chinese Physics Letters 36, 076801 (2019).
- [43]A. Rycerz, J. Tworzydlo, and C. W. J. Beenakker, Nature Physics 3, 172 (2007).
- [44]S.-g. Cheng, J. Zhou, H. Jiang, and Q.-F. Sun, New Journal of Physics 18, 103024 (2016).
- [45]J. Zhou and J.-C. Charlier, Physical Review Research 3, L024033 (2021).
- [46]W. Wang, B. Lyu, L. Kong, H. Fanghor, and H. Du, Chinese Physics B 33, 107508 (2024).
- [47]H. V. Gomonay and V. M. Loktev, Physical Review B 81, 144427 (2010).
- [48]R. Cheng, M. W. Daniels, J.-G. Zhu, and D. Xiao, Physical Review B 91, 064423 (2015).



