Large magneto-optical Kerr effect induced by collinear antiferromagnetic order Authors: H Yoshimochi, K Yoshida, R Oiwa, T Nomoto, N D Khanh, A Kitaori, R Takagi, R Arita, S Seki Published: 2026 ABSTRACT AbstractIn modern technology, optical readout of magnetic information is conventionally achieved by the magneto-optical Kerr effect, i.e., polarization rotation of reflected light. The Kerr rotation is sensitive to time-reversal symmetry breaking and generally proportional to magnetization, enabling optical readout of the ↑ and ↓ spin states in ferromagnets. By contrast, antiferromagnets with a collinear antiparallel spin arrangement have long been considered inactive to such magneto-optical responses, because of (time-reversal followed by translation t) symmetry and lack of macroscopic magnetization. Here, we report the identification of a large magneto-optical Kerr effect induced by collinear antiferromagnetic order, through detailed measurements of a room-temperature antiferromagnetic insulator α . Our first-principles calculations successfully reproduce both the absolute magnitude and spectral shape of the Kerr rotation and ellipticity with remarkable accuracy, which unambiguously proves that it originates from a -symmetry-broken collinear antiferromagnetic order, rather than magnetization. This compound hosts temperature-dependent transition between easy-plane and easy-axis antiferromagnetic states, and their contrasting behaviors suggest that the selection rule is governed by the detail of magnetic symmetry. The present results demonstrate that even a simple collinear antiferromagnetic order can induce a large magneto-optical Kerr effect, and highlight -symmetry-broken antiferromagnets as a promising material platform for highly sensitive optical detection of and spin states.Subject terms: Materials science, Physics Introduction Currently, magnetic information is mainly stored by the ↑ and ↓ spin states of ferromagnets. These two states are related by the time-reversal operation , and their distinguishability originates from the breaking of time-reversal symmetry in ferromagnets (Left panel of Fig. 1a). Such a time-reversal symmetry breaking gives rise to various functional off-diagonal responses, such as spontaneous Hall effect1 and the spontaneous magneto-optical Kerr effect2, both of which are frequently used for electrical and optical readout of magnetic information. Classification of various types of collinear magnets and spontaneous magneto-optical Kerr effect. (a), Comparison of ferromagnet (FM), conventional antiferromagnet (AFM), and TtS-broken AFM. Red arrows represent local magnetic moments. Blue and red circles indicate magnetic and non-magnetic ions, respectively. Magnetic domains A and B are converted into each other by time-reversal operation . These two domains (i.e., and ) are identical under a translation operation t in conventional AFMs, but not identical in TtS-broken AFMs with appropriately located non-magnetic ions. (b), Comparison of spontaneous magneto-optical Kerr effect in FMs and TtS-broken AFMs. In the latter case, the collinear spin order induces a fictitious magnetic field, which leads to the emergence of Kerr rotation even without magnetization M. Gray and blue arrows represent the propagation vector and polarization direction of the light, respectively. The sign of Kerr rotation angle is opposite between the domains A and B. Note that the spontaneous Kerr effect is not allowed in conventional antiferromagnet, because of its TtS. Here, the spontaneous Hall effect represents the generation of transverse electric field (∥ y) by the application of electric current (∥ x), which originates from the off-diagonal component () of the electrical conductivity tensor defined by . In ferromagnets, is generally proportional to magnetization M along the z-direction ()1. On the other hand, the magneto-optical Kerr effect refers to the change in the polarization state of reflected light, when linearly polarized light is incident on the surface of a magnetic material (Left panel of Fig. 1b)2. This effect is characterized by the Kerr rotation angle , describing the rotation of the polarization axis, and the Kerr ellipticity , describing the induced ellipticity. The magneto-optical Kerr effect can be regarded as an optical-frequency extension of the Hall effect. For example, in the polar Kerr geometry, where incident direction of light (z) is normal to the sample surface, the relation holds. Consequently, and are also expected to be proportional to in ferromagnets. These phenomena provide electrical and optical probes of the ↑ and ↓ spin states in ferromagnets. In contrast, it has long been considered that such functional responses cannot be expected in antiferromagnets (AFMs) with collinear antiparallel spin alignment, because of their vanishing net magnetization. This can be intuitively understood from symmetry considerations. In conventional collinear antiferromagnets, one can consider the and spin configurations, which are converted into each other by the time-reversal operation . As illustrated in the middle panel of Fig. 1a, however, these two configurations are identical under a lattice translation t and are therefore indistinguishable. As a result, the combined symmetry (TtS) is preserved, prohibiting the appearance of aforementioned off-diagonal responses. Very recently, however, it has been theoretically pointed out that even a collinear antiferromagnetic order can break TtS and induce spontaneous Hall/Kerr effect when combined with a crystal lattice of appropriate symmetry3,4. For example, when non-magnetic ions are arranged in a staggered manner, as shown in the right panel of Fig. 1a, the and states are no longer equivalent under a translation. In this case, TtS is broken, and the emergence of a spontaneous Hall/Kerr effect is often allowed even in the absence of net magnetization (Right panel of Fig. 1b). This new class of collinear antiferromagnets with symmetry breaking, often termed “altermagnets”, is predicted to exhibit similar functional responses as ferromagnets5–11, with various unique advantages such as negligibly small stray fields and ultrafast spin dynamics12,13. According to the latest experimental studies, such a spontaneous Hall effect driven by antiferromagnetic order (rather than M) has indeed been identified in a few TtS-broken collinear antiferromagnets such as FeS14 and MnTe15. In this case, the Berry curvature of the electronic band structure acts as a fictitious magnetic field for conduction electrons, resulting in the spontaneous Hall response without the necessity of M. These findings suggest that TtS-broken antiferromagnets may also host a spontaneous magneto-optical Kerr effect independently of magnetization, since it follows the same symmetry-based selection rules as the Hall effect. (Note that TtS-breaking alone is not a sufficient condition for the emergence of fictitious magnetic field, and some additional symmetry breaking is necessary as discussed later.) Importantly, while the Hall effect can only be observed in metals, the magneto-optical Kerr effect is accessible even in insulating materials. Moreover, its spectral information will provide a better means to understand the microscopic origin of the observed response. In this study, we report the identification of a large magneto-optical Kerr effect induced by collinear antiferromagnetic order, through detailed measurements of a room-temperature antiferromagnetic insulator α . In this compound, the and spin states induce the opposite sign of spontaneous Kerr effect. Our first-principles calculations successfully reproduce both the absolute magnitude and spectral shape of the Kerr rotation and ellipticity with remarkable accuracy, which unambiguously proves that it originates from a -symmetry-broken collinear antiferromagnetic order, rather than magnetization. This compound hosts temperature-dependent transition between easy-plane and easy-axis antiferromagnetic states, and their contrasting behaviors are also investigated in detail. The present results demonstrate that even a simple collinear antiferromagnetic order can induce a large magneto-optical Kerr effect, and highlight TtS-broken antiferromagnets as a promising material platform for highly sensitive optical detection of and spin states. Our target material α crystallizes in the corundum structure with the trigonal space group , as shown in Fig. 2a. The crystal lattice consists of alternating stacking of Fe and O ion layers along the c-axis, and a crystallographic unit cell contains four magnetic Fe (S=5/2) sites. Each Fe ion is sandwiched by a pair of oppositely oriented triangular plaquettes of O ions, which realizes the staggered manner of non-magnetic ion distribution as considered in the right panel of Fig. 1a. The magnetism is dominated by the third- and fourth-nearest neighbor interactions between Fe sites, which are both antiferromagnetic16–19. As a result, the easy-plane collinear antiferromagnetic order as shown in Fig. 2a and the middle panel of Fig. 3g is stabilized below the magnetic ordering temperature K, where antiparallel spins are oriented along the y-axis (i.e., the direction normal to both a- and c-axes)20. Spontaneous magneto-optical Kerr effect in a collinear antiferromagnet α at room temperature. (a,d) Schematic illustration of the spin arrangements in the easy-plane AFM phase of α . Magnetic domains A (a) and B (d) (i.e., the and spin states) are converted into each other by time-reversal operation, which are characterized by the opposite sign of fictitious magnetic field and tiny spontaneous magnetization along the a-axis. (b,c) Magnetic field dependence of magnetization M and polar magneto-optical Kerr rotation angle measured with the normal incidence to the a-plane surface for B ∥ a at 300 K in the easy-plane AFM phase. The incident light energy hν is set as 2.25 eV. Here, the domain A (domain B) is selected by applying the negative (positive) sign of external magnetic field. The definition of and are indicated in b and c, respectively. The presence (absence) of spontaneous magneto-optical Kerr effect in the easy-plane (easy-axis) AFM state. (a–d) Magnetic field dependence of magnetization M and polar magneto-optical Kerr rotation angle under B ∥ a at 50 K (a,b) and 245 K (c,d). e, B-T magnetic phase diagram for B ∥ a, where the background color represents the amplitude of . f, Temperature dependence of spontaneous Kerr rotation angle , obtained by linearly extrapolating the -B curve to B=0 (as shown in Fig. 2c). (g) Symmetry analysis of various collinear spin arrangements on the α crystal lattice. Magnetic point group, magnetic structure and corresponding symmetry-imposed shape of optical conductivity tensor are presented for easy-axis AFM, easy-plane AFM and ferromagnetic (FM) spin states. In α , the space-inversion symmetry is broken at the middle point between two neighboring Fe sites (i and j), which leads to the appearance of Dzyaloshinskii-Moriya (DM) interaction 21. Here, and are spins at the sites i and j, and is a DM vector parallel to the c-axis. To obtain the energy gain by DM interaction, spins are slightly tilted from the y-axis toward the a-axis, leading to the emergence of tiny spontaneous magnetization along the a-axis. Notably, this compound is also known to host the Morin transition at K, where the sign of magnetic anisotropy changes22. Below , easy-axis antiferromagnetic state is realized, in which antiparallel spins are oriented along the c-axis as shown in the left panel of Fig. 3g. In the latter phase, the aforementioned DM-induced spin tilting and associated spontaneous magnetization are absent. Previously, the presence of finite magneto-optical Kerr responses in α has been reported23–25. Nevertheless, the clarification of its detailed microscopic mechanism (i.e., theoretical explanation of absolute magnitude and spectral shape of and ), as well as the relationship between two distinctive AFM orders, remains an essential challenge. In the following, we focus on the polar Kerr geometry, where the incident and reflected lights are normal to the a-plane surface of the sample. Figure 2b and c display the magnetic field dependence of magnetization M and Kerr rotation angle , measured at 300 K (i.e., the easy-plane antiferromagnetic state) with B ∥ a. In both M and profiles, a clear step-like anomaly is observed around B=0. Despite very small amplitude of spontaneous magnetization /Fe, the observed spontaneous Kerr rotation angle ( mdeg) is exceptionally large, which is comparable or even larger than typical ferromagnets. Importantly, while M increases linearly with a magnetic field, remains nearly constant up to 5 T. These features imply that magnetization M is not the main source of the observed signal. (We also confirmed that Kerr ellipticity shows a similar B-dependence as as detailed in Supplementary Note I and Supplementary Fig. 1.) Such a spontaneous Kerr signal in the easy-plane antiferromagnetic phase can be naturally understood by considering its magnetic symmetry. According to the Neumann’s principle, when the magnetic point group of the target system is specified, the symmetry-imposed shape of optical conductivity tensor (i.e., the presence/absence of each tensor component) is uniquely identified26,27. Figure 3g summarizes the magnetic point group and the corresponding symmetry-imposed shape of the conductivity tensor for various collinear spin configurations on a α crystal lattice. Importantly, the magnetic point group of the easy-plane antiferromagnetic state is 2/m, which is the same as the one for the ferromagnetic state with M ∥ a. As a result, these two states are characterized by the common form of the optical conductivity tensor with non-zero off-diagonal component 26,27. This means that reflected light in the easy-plane antiferromagnetic state should behave as if there is fictitious magnetic field (or fictitious magnetization) along the x (||a) axis, which leads to the appearance of a spontaneous polar Kerr effect for the a-plane surface. In such an environment, DM interaction is allowed to cause additional canting of antiferromagnetically aligned moments toward the a-direction, which explains the appearance of a weak spontaneous magnetization 21. Note that the aforementioned spontaneous Kerr effect is allowed even without this additional spin canting, since the easy-plane collinear antiferromagnetic order itself is characterized by TtS-broken magnetic point group 2/m allowing non-zero . There are two degenerated time-reversal domains, i.e., the domains A and B characterized by the and spin arrangements, respectively (Figs. 2a and d). They are converted into each other by the time-reversal operation, and therefore should be characterized by opposite signs of the fictitious magnetic field, spontaneous Kerr rotation and . By applying the positive (negative) sign of external magnetic field along the a-axis, their degeneracy is lifted through and the domain A (domain B) can be selected. This suggests that the step-like anomaly observed in the profile should represent the B-induced switching between the domains A and B. Below the Morin transition temperature , the sign of magnetic anisotropy is changed and the easy-axis antiferromagnetic state is realized. As shown in the left panel of Fig. 3g, the corresponding magnetic point group is , which indicates the absence of any off-diagonal component of the conductivity tensor and associated spontaneous Kerr effect. To confirm this symmetry-based prediction, magnetic-field-dependence of M and are measured at 50 K in the easy-axis antiferromagnetic phase (Figs. 3a and b). Unlike the easy-plane antiferromagnetic phase, the step-like anomalies around B=0 associated with spontaneous and are absent, which is consistent with the aforementioned symmetry analysis. For better understanding of the contrasting behaviors between the easy-plane and easy-axis AFM states, similar measurements are also performed at 245 K (Figs. 3c and d) close to the Morin transition temperature K. Here, the application of B ∥ a induces a transition from easy-axis AFM state to easy-plane AFM state at 2.5 T, which is identified as a small anomaly in the magnetization profile. In the easy-plane AFM state, remains nearly constant against B. It indicates that the observed Kerr signal is governed by the antiferromagnetic order parameter, and the M-linear component of is negligibly small. In the easy-axis AFM state, on the other hand, the linear slope of -B curve is much larger than that of the easy-plane AFM phase, despite similar amplitude of linear slopes in M-B profiles between these two phases. It suggests that B-linear increase of in the easy-axis AFM phase is probably not caused by M, but rather by the B-induced tilting of Néel vector (defined by , with and representing the spins at two distinctive sublattices) away from the c-axis towards the y-axis. Since this -tilting induces the same spin component as the easy-plane AFM state, it leads to the appearance of finite scaling with the B-induced L-tilting angle. This direction of -tilting should be favored under , because the combination of and component leads to the additional DM energy gain (which is the source of weak spontaneous in the easy-plane AFM state with ). In this context, the observed B-linear increase of in the easy-axis AFM state can be also ascribed to the B-induced component of easy-plane-type antiferromagnetic order parameter, rather than small net magnetization. Figure 3e summarizes the experimentally measured amplitude of plotted on the B-T phase diagram. The easy-plane antiferromagnetic state is always characterized by large amplitude of , which remains almost constant against B. In the easy-axis antiferromagnetic state, on the other hand, linearly increases as a function of B, while its amplitude is smaller. Figure 3f indicates the temperature dependence of spontaneous Kerr rotation angle , obtained by linearly extrapolating the -B curve to B=0. It clearly demonstrates that large spontaneous Kerr effect is observable only in the easy-plane antiferromagnetic state, in accord with the symmetry-based prediction in Fig. 3g. To understand the microscopic origin of the observed large Kerr response, we examined the spectroscopic features of the magneto-optical signals. Figure 4b displays the experimental energy spectra of the Kerr rotation and ellipticity measured at 300 K (easy-plane AFM phase). The spectrum exhibits a broad peak at ∼ 2.3 eV, accompanied by a sign change in the spectrum. This dispersive line shape in is the Kramers-Kronig counterpart to the absorptive peak structure observed in the spectrum, which validates the resonance nature of the electronic transitions responsible for the large Kerr signal. Microscopic origin of spontaneous magneto-optical Kerr effect in the easy-plane AFM phase of α . (a) Electronic band structure and density of state (DOS) of α in the easy-plane AFM state, theoretically calculated based on DFT calculation with on-site Hubbard correlation U = 4.0 eV. The letter symbols represent the positions in Brillouin zone, schematically illustrated in the right panel. (b) Experimental energy spectra of Kerr rotation angle and Kerr ellipticity , measured at 300 K (easy-plane AFM state) with B ∥ a at 1.0 T. (c) Theoretical energy spectra of and in the easy-plane AFM state with M=0, calculated based on the electronic structure in a. d, Full logarithmic plot of magnetization dependence of the polar magneto-optical Kerr rotation for various magnetic materials. Conventional ferromagnets are located in the yellow shadowed region. To theoretically account for these features, we performed first-principles calculations based on density functional theory (DFT). Figure 4a shows the calculated electronic band structure and density of states (DOS) in the easy-plane AFM state. The DOS analysis reveals that the Fe 3d orbitals are split into occupied and unoccupied bands, separated by the O 2p bands. This electronic configuration characterizes α as a charge-transfer insulator, where the energy gap of around 2 eV corresponds to the optical transition from the occupied O 2p orbitals to the unoccupied Fe 3d orbitals. Figure 4c presents the Kerr spectra theoretically calculated for the easy-plane AFM state with M=0, based on the electronic structure obtained by DFT calculation. The calculated spectra exhibit a broad peak structure in at ∼ 2.5 eV, accompanied by a sign reversal in . These features are in good agreement with the experimental data in Fig. 4b. In both experimental and theoretical data, the maximum amplitude of Kerr signal is mdeg. The above results indicate that the observed spectral shape and amplitude of Kerr signals are quantitatively well reproduced by the DFT calculation with M=0. In Supplementary Fig. 3 and Supplementary Note III, we also performed similar DFT calculations for the easy-plane AFM state with and without tiny additional spin canting towards the a-axis, which confirmed that the calculated Kerr rotation and Kerr ellipticity are rarely affected by the small net magnetization . The above results unambiguously prove that in the present compound is dominantly induced by TtS-broken antiferromagnetic order itself, rather than net magnetization. In Supplementary Note III and Supplementary Fig. 4, we also provided the theoretically calculated Kerr spectra for easy-axis AFM phase. It predicts the absence of spontaneous and , consistent with the experimental results (Fig. 3a) and symmetry-based analyses (Left panel of Fig. 3g). In Fig. 4d, the Kerr rotation angle is plotted against the spontaneous magnetization for α , as well as the other previously reported materials28–43, following Ref.44. Here, the conventional ferromagnets are located in the yellow-shaded region, where almost linearly scales with M. In this case, the typical value of ratio is 0.2 ∼ 2.0 deg T. On the other hand, the present antiferromagnetic α clearly deviates from this trend, accompanied by a large Kerr rotation angle with a small net magnetization ( /Fe). Consequently, the Kerr rotation angle per unit magnetization reaches deg T, which is the largest value ever reported and one or two orders of magnitude larger than that of typical FMs. This further supports the nontrivial origin of spontaneous Kerr effect in α , and demonstrates that the TtS-broken collinear or weakly canted antiferromagnets with the appropriate symmetry can be a promising material platform to realize large magneto-optical effect with a magnitude comparable to ferromagnets despite vanishingly small net magnetization. In this study, we report the identification of a large magneto-optical Kerr effect induced by collinear antiferromagnetic order, through detailed measurements of a room-temperature antiferromagnetic insulator α . Our first-principles calculations successfully reproduce both the spectral shape and absolute magnitude of the Kerr rotation and ellipticity with remarkable accuracy. It unambiguously establishes that this effect originates from the intrinsic antiferromagnetic order itself, rather than net magnetization. Previously, spontaneous Hall and Kerr effects with vanishingly small magnetization have been explored mainly in a limited class of TtS-broken non-collinear helimagnets, such as Mn (X= Sn, Ge)44–46 and CoS (M= Ta, Nb)47–49. Our results demonstrate that even the simplest collinear antiferromagnetic order can induce a large spontaneous Kerr effect at room temperature, when combined with an appropriate crystal lattice symmetry. The observed contrasting behaviors between the easy-plane and easy-axis AFM phases, i.e., the presence and absence of spontaneous Kerr response, respectively, suggest that the breaking of symmetry is a necessary but not sufficient condition for the emergence of a fictitious magnetic field, highlighting the importance of additional symmetry breaking. Notably, even in the latter easy-axis AFM phase, the application of an external magnetic field induces a magneto-optical response that is significantly larger than what would be expected from a simple -driven mechanism. This enhancement can be attributed to the field-induced -component that is equivalent to that of the easy-plane AFM state. These observations indicate that magneto-optical responses in TtS-broken antiferromagnets can be significantly amplified beyond conventional -based mechanisms, which provides a powerful means for highly sensitive optical readout of and spin states. In such systems, efficient optical writing of and spin states via inverse effects (such as the inverse Faraday effect) may be also expected. Our findings suggest that TtS-broken antiferromagnets potentially serve as a viable alternative to ferromagnets for optical information technologies, and further exploration of this new class of materials and their unique properties associated with fictitious magnetic field would be interesting subjects for future study. Note that very recent independent works50,51 reach the similar conclusions regarding the large magneto-optical Kerr effect in the easy-plane AFM phase of α . While their measurements were limited to room temperature (easy-plane AFM phase), our study provides comprehensive temperature dependence of MOKE signal spanning across the Morin transition. It reveals the distinct symmetry-driven magneto-optical responses between the easy-plane and easy-axis AFM phases, which contributes to better understanding for the origin of observed MOKE signal. Methods Sample preparation Plate-shaped single crystals of α , with flat surfaces normal to the a-axis polished to an optical grade, were purchased from Crystal Base Co., Ltd. The purity and single crystalline character of the sample were confirmed by the magnetization measurements and X-ray Laue back-reflection method, the latter of which ensures that the flat surfaces indeed correspond to the a-plane. Magnetization measurement. Magnetization measurements were performed using a Magnetic Properties Measurement System (MPMS, Quantum Design) with a superconducting quantum interference device (SQUID) magnetometer on a rectangular-shaped sample. Magneto-optical Kerr measurement. Polar Kerr measurements were performed by a standard method based on photoelastic optical phase modulator (PEM), where the incident and reflected lights are normal to the a-plane surface of the sample. A supercontinuum laser (SC-OEM, YSL Photonics) was used as a light source. The beam with the selective wavelength was extracted by using the wavelength selection module based on the acousto-optic crystal (AOTF-PRO, YSL Photonics). The beam was linearly polarized by using a Glan-laser prism and modulated at ∼ 50 kHz and ∼ 227 Hz with PEM and a mechanical chopper, respectively. The reflected beam then passed through a Glan-laser prism and detected with a Silicon photodetector. We performed synchronous detection of the reflected beam at the first and second harmonic frequency of the PEM modulation by using a lock-in amplifier. The static reflectivity was simultaneously measured by using a lock-in amplifier with the frequency of the mechanical chopper. External magnetic field up to 5 T was applied perpendicular to the sample surface, by using a superconducting magnet. Since the Kerr signal is an odd function of magnetic field B, the obtained and profiles, as well as the energy spectra of the and , are anti-symmetrized with respect to B as and , respectively, with and representing the raw experimental data. This anti-symmetrization eliminates contributions from the natural optical rotation, Voigt effect, and other B-odd extrinsic artifacts arising from the optical setup. The measurement at 300 K is performed without an optical window. To eliminate the extrinsic contribution, a silver thin film was deposited on the half of sample surface by using the radio-frequency sputtering system, and the reference data of the silver surface was subtracted from the data of the α surface. For the measurement at low temperature, B-induced Faraday rotation by an optical window was evaluated in advance, and subtracted from the data. First-principles calculations The electronic band structure and Kerr spectra of α were calculated within spin DFT based on the projector augmented wave method 52, as implemented in the Vienna Ab initio Simulation Package (VASP) 53,54. The calculations have been performed for the rhombohedral structure, using the lattice constants and 55. In the easy-plane (easy-axis) antiferromagnetic state, a collinear antiferromagnetic order was assumed, with spins lying in the c-plane (c-axis). The exchange-correlation functional was treated within the generalized gradient approximation (GGA) in the form proposed by Perdew-Burke-Ernzerhof 56. A plane-wave energy cutoff of 520 eV was employed, and the Brillouin zone was sampled using a k grid. To account for strong electron correlation effects in the Fe 3d orbitals, the GGA+U method was used with an effective on-site Hubbard interaction of U = 4.0 eV. Maximally localized Wannier functions were constructed using the Wannier90 package 57,58 based on the spin DFT calculations, from which a 76-orbital Wannier tight-binding model was built using 256 Bloch bands on the k grid, with the Fe-d and O-p orbitals taken as initial projections. The Kerr rotation angle and Kerr ellipticity shown in Fig. 4c and Supplementary Fig. 3b–e were calculated within linear response theory using the Kubo formula 59,60, where the optical conductivity tensor was evaluated using the Wannier tight-binding Hamiltonian on a k-point grid. Supplementary Information Supplementary Information. Acknowledgements The authors thank N. Ogawa and R. Hirakida for enlightening discussions and experimental helps. The illustrations of crystal structure were drawn by VESTA61. Author contributions S.S. and R.T. conceived the project. H.Y., K.Y. and R.T. characterized the magnetic properties with the assistance of N.D.K, A.K., and S.S. Magneto-optical Kerr measurements were carried out by H.Y. and K.Y. First-principles calculation was performed by R.O., T.N., and R.A. The manuscript was written by H.Y. and S.S. with the assistance of R.O. and R.A. All the authors discussed the results and commented on the manuscript. Funding This work was partly supported by Grants-In-Aid for Scientific Research (grant nos 21H04440, 21H04990, 21K13876, 21K18595, 22KJ1061, 22H04965, 23K19029, 24K00579, 24H02235, 24K00581, 25K21684, 25H00420, 25H00611, 25H01252, 25H01246, 25K00956, 26K17082, 26H01285) from JSPS, PRESTO (grant nos JPMJPR18L5, JPMJPR20B4, JPMJPR23Q3) and CREST (grant no. JPMJCR23O4) from JST, Katsu Research Encouragement Award, UTEC-UTokyo FSI Research Grant Program of the University of Tokyo, Special Postdoctoral Researcher Program at RIKEN, Asahi Glass Foundation, Murata Science Foundation, Ozawa-Yoshikawa Memorial Electronics Research Fund, Research Foundation for the Electrotechnology of Chubu, Tokuyama Science Foundation, and The Mitsubishi Foundation. H.Y. was supported by JSPS through the research fellowship for young scientists and Iwadare Scholarship Foundation. Data availability The data presented in the current study are available from the corresponding authors on reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. H. Yoshimochi and K. Yoshida have contributed equally. Supplementary Information The online version contains supplementary material available at 10.1038/s41598-026-56512-7. References Associated Data Supplementary Materials Supplementary Information. Data Availability Statement The data presented in the current study are available from the corresponding authors on reasonable request.