Piezomagnetoelectric effects in a candidate Kitaev magnet Authors: Vilmos Kocsis, Sven Luther, Nicolás Pérez, Weiliang Yao, Hannes Kühne, Anja U B Wolter, Yuan Li, Bernd Büchner Published: 2026 ABSTRACT AbstractThe exactly solvable Kitaev model with its frustrated bond-dependent interactions has attracted enormous attention due to its exotic physics hosting fractional spin excitations as well as its promising prospects for quantum information technology. However, there is no pristine realization of the Kitaev model due to the significant Heisenberg and off-diagonal exchange interactions. While these additional exchange interactions are considered as obstacles on the route towards the desired Kitaev quantum spin liquids, the interplay between these magnetic anisotropies and the Kitaev interaction has lead to numerous intriguing phenomena. Here we demonstrate a new phenomenon, the coexistence of the Kitaev interaction with the piezomagnetoelectric effect (simultaneous magnetoelastic and magnetoelectric responses), which can offer electric field driven manipulation of the ground state and the fractional spin excitations. Our study reports the direct observation of the magnetoelectric (ME) effect in a Kitaev-Heisenberg, the quantum spin liquid candidate Na2Co2TeO6, and highlights the magnetoelastic response as a sensitive gauge of phase transitions. We discuss that the ME effect originates from the pd-hybridization mechanism, which allows local polarization independently from any magnetic order. This mechanism can transfer the frustrated magnetic interactions onto the polarization system, potentially creating a new exotic electronic state, a polarization liquid.Subject terms: Quantum fluids and solids, Ferroelectrics and multiferroics, Magnetic properties and materials Introduction Quantum spin liquids (QSL)1–5, particularly those described by the exactly solvable Kitaev model6 have attracted huge interest7–12, due to their promising features for quantum computing, such as decoherence-protected quantum information computation, storage, and transmittance13. Unlike to geometrically frustrated models, the Kitaev model draws the frustration required for the QSL state from the bond-dependent interaction and not from the lattice geometry. The pristine Kitaev model was first sought after in Mott insulators with pseudospin Jeff=1/2, such as in materials with Ir4+ and Ru3+ ions on a honeycomb lattice14–16, and recently materials with Co2+ ions are also suggested as possible hosts11. However, as the long-range ordered magnetic ground states in all Kitaev candidates show the presence of significant additional exchange interactions17,18, the realization of a pristine Kitaev model stays elusive, and so-far the long-range order could be suppressed only by the application of large magnetic field15,16,19–22. Still, these extended Heisenberg-Kitaev models have interesting consequences. The additional couplings were found to be important in the explanation of thermal transport phenomena23,24, Raman scattering25,26, and they lead to magnetic anisotropies and strong magnetoelastic effects27,28. It is an interesting question, if these unwanted (but seemingly unavoidable) interactions can couple to external fields and can be further exploited in a completely different way to help the suppression of long-range order. Thus, recently, the application of uniaxial pressure28 and electric (E) field29–32 instead of magnetic field were suggested, as new control parameters for the stabilization of QSL phases, or for controlling the Majorana-Fermi surface33, as theoretical possibilities. Moreover, the importance of magnetoelectric interactions was also pointed out as a possible explanation29,30 for the strong electric dipole activity of THz excitations in d5 transition metal Mott insulators34 and in the Tunneling Spectroscopy of QSL materials35. While polar QSL materials have been reported, such as local polarization exists in α − RuCl336 and PbCuTe2O6 shows ferroelectricity37, for long, magnetoelectric (ME) coupling was not considered possible in these Jeff=1/2 systems. In agreement with this expectation, so far no ME coupling has been found in any of the QSL candidates. However, only ME materials offer strong cross coupling between the electric and magnetic degrees of freedoms38, and readily demonstrate E control over magnetic ground states and magnetic excitations39,40. Here we provide the first direct evidence for a sizable ME and magnetoelastic coupling, and consequently, piezomagnetoelectric effects in a QSL candidate material. In order to find a QSL candidate material with ME coupling, we do a general survey among Kitaev magnets. We directly investigate both the ME and magnetoelastic properties of the QSL candidate Na2Co2TeO641 with Jeff=1/2, in order to provide evidence for the origin of the ME effect. We find large magnetostriction and a sizable ME coupling, which greatly enriches the magnetic phase diagram and highlights the complex and anisotropic magnetic interactions. We argue that the dominant source of the ME coupling in Na2Co2TeO6 is the pd-hybridization mechanism42,43, and therefore it can even host local P independent from long-range magnetic order44, while the spin-current mechanism can be also relevant. Our observation of simultaneous magnetoelasticity and magnetoelectricity demonstrates the presence of the piezomagnetoelectric effect. As an interesting consequence of the ME effect coexisting with the Kitaev interaction, we argue that an E field could emerge as a new control parameter for the ground state and fractional spin excitations. The piezomagnetoelectric effect can offer an interesting experimental and theoretical ground for the investigation of coupled frustrated (magnetic) and frustration free (electric) systems endowed with a tuneable coupling parameter. Results Due to the seemingly mutually excluding requirements of the ME effect and the Kitaev interaction, as discussed in the Supplementary Information, Na2Co2TeO6 is unique among the QSL candidates. The chiral lattice (s.g. P6322) of Na2Co2TeO6 is built up by co-aligned, edge-sharing CoO6 octahedra pairs, which form a honeycomb layer with the Te6+ ions filling in the cavities, as shown in Fig. 1(a). The honeycomb layers have alternating stacking, and are separated by Na+ ions. The octahedral environment around the two non-equivalent Co2+ ions has trigonal distortion45, while the two neighboring CoO6 are connected by a mirror symmetry (ma*, which is not a symmetry of the three-dimensional lattice). Each octahedra has a C3 and three C2 local symmetries (different from the lattice symmetries) with axes parallel and perpendicular to the c axis, respectively. When the spin of the magnetic Co2+ ion is aligned parallel to the a axis (perpendicular to the Co-Co bonds), the local C3 and two of the C2 symmetries of the CoO6 octahedron are broken. This allows the emergence of a local polarization (P) at both Co sites perpendicular to the spin and parallel to the axis of the preserved C2 symmetry which is combined by the time reversal symmetry. In the paramagnetic phase, the absence of long-range order cancels the effect of the spin, effectively re-establishes the C3 and C2 symmetries, and precludes the emergence of macroscopic polarization, that is, Na2Co2TeO6 is paraelectric. In the presence of an in-plane spin order, the effect of the spin on the local symmetries is not canceled, but even in this case the ma* symmetry connecting the CoO6 pairs renders these local P antiparallel, i. e. the structure is antiferroelectric, as shown in Fig. 1(b). Structural and thermodynamic properties of the quantum spin-liquid candidate Na2Co2TeO6. a Single honeycomb layer of CoO6 octahedra with Te6+ ions filling in the cavities. The red arrows (spins) illustrate the triple-q order with finite T ~ ∑iPi × Si toroidal moment. b Local symmetries of a CoO6 octahedra; the emergence of finite spin eliminates the C3 and two of the C2 symmetries, leaving only the C2 symmetry perpendicular to S combined with the t time reversal symmetry. The broken symmetries allow the emergence of a local polarization (P) parallel to the Co-Co bond, which are anti-parallel to each other on the edge sharing CoO6 octahedra, as these are connected by the ma* mirror symmetry. c Temperature dependence of the zero-field αc thermal expansion coefficient and the Cp heat capacity measured in H=0 kOe. At low temperatures, Na2Co2TeO6 has an easy-plane antiferromagnetic (AFM) order with low anisotropy within the honeycomb plane45–49. The long-range AFM order develops at TN1 = 26.8 K, which is followed by two further transitions at TN2 = 16.5 K, and TN3 = 7.1 K, which give additional complexity to the magnetic structure and indicate a canted planar AFM structure with weak-ferromagnetism49. At first, the magnetic ground state was described as a zigzag-type AFM order45–48, however, recently a triple-q AFM state with a vortex-like spin order and possible phase co-existence is suggested, as shown in Fig. 1(a)50–52. Most recent theoretical models also suggest the emergence of composite multilinear order parameters, the so-called spin vestigial order above TN153. We emphasize, that only the triple-q AFM order allows the emergence of the ME effect due to the finite local polarization and the finite T ~ ∑iPi × Si total toroidal moment of the unit cell54. Both our thermal expansion (αc-T) and heat capacity (Cp-T) measurements shown in Fig. 1(c) confirm these phase transitions. At TN1, the thermal expansion and the heat capacity show clear anomalies, while at TN2, αc has a minimum and Cp has a very slight anomaly. Finally, at TN3, the thermal expansion measurement reveals a broad peak which is accompanied by only a weak anomaly in Cp. As the peaks in the thermal expansion are positive, according to the Ehrenfest relation55 their uniaxial pressure dependence is also positive ∂TN∂pc>0. Therefore, unlike to α-RuCl327, which has a negative peak in αc*, in Na2Co2TeO6 uniaxial pressure further stabilizes the AFM phase. Figure 2 summarizes the field-dependent magnetic, magnetoelectric, and magnetoelastic properties of Na2Co2TeO6 measured at T=8.0 K for H∥a*. First, we consider the magnetic and ME measurements shown in Fig. 2(a, b), respectively. In the low-field region, Na2Co2TeO6 has a weak ferromagnetic moment, which is reversed at H=0 kOe and has an additional step-like increase at H=15 kOe. In the high-field region, we find a strong peak in dM/dH at H=55 kOe in Fig. 2(a), which indicates the partial spin-flop like phase transition48,49, while an additional weak peak emerges at H=75 kOe. Figure 2(b) shows the measured P-H curves for all the four combinations of the poling fields ( ± E0, ± H0), the ME poling procedure is described in the “Methods” section. The measured P-H is the highest for the same sign combination of the (E0, H0) poling fields, and zero if H0=0 kOe or H0=90 kOe; as discussed in the Supplementary Information, a finite ME response is only observed if the sample is cooled in the presence of a moderate H0 field. Na2Co2TeO6 shows a nearly linear ME effect (P = χMEH) with χME = ∓ 4 ps/m maximum susceptibility, which is comparable to other S=1/2 systems, such as LiCu2O256 and SrCuTe2O657, but much smaller than in CoSe2O558. At around H=55 kOe, the P shows a sudden decrease and χME drops to nearly zero, and the magnitude of the ΔP=5 μC/m2 change is comparable to that observed in PbCuTe2O637. This drop in the P-H curves coincides with the partial spin-flop like phase transition48,49 as indicated by the peak-like anomaly in the dM/dH, as shown in Fig. 2(a). Magnetic, magnetoelectric, and magnetoelastic properties of Na2Co2TeO6. a Magnetic field dependence of the M magnetization (solid lines) and the dM/dH field derivative (dashed lines) for H∥a* at T=8.0 K. b Poling-field dependence of the H-field induced in-plane polarization (P∥a) measured at T=8 K. The sample was cooled to the ordered phase in the presence of the four combinations of the poling fields ( ± E0, ± H0), E∥a, H∥a*. The measurements were started from the H0= ± 50 kOe poling field in the absence of E field. Solid and dashed curves indicate measurements where the magnitude of the H field is decreased or increased after poling, respectively. The P shows mostly linear H-field dependence. c Magnetic field dependence of the measured ΔLc/Lc relative length change (ΔLc∥c), and (d) the calculated λc linear magnetostriction coefficient at T=8.0 K. The ΔLc/Lc magnetostriction measured along the c axis (ΔLc∥c) at T=8.0 K for H∥a*, and the calculated λc linear magnetostriction coefficient are shown in Fig. 2(c, d), respectively. Though in this configuration the magnetostriction measurements do not account for changes of lattice symmetry, λc stays a sensitive indicator of magnetic and elastic phase transitions. The magnetostriction measurements reveal numerous anomalies for H∥a*, suggesting a more complex phase diagram than that of α-RuCl327,59. We note, that complementary magnetostriction measurements for both ΔLa∥a and ΔLa*∥a* have been reported in ref. 60. The anomalies in the λc-H measurements coincide with features in the dM/dH-H magnetization measurements. However, it is important to note, that the abundance of features in the λc-H measurements is not reflected in the P-H measurements, which implies that the origin of the ME effect is not related to magnetoelasticity. The λc-H is neither symmetric nor anti-symmetric in H. Remarkably, the field increasing (↑, red curve) and decreasing (↓, blue curve) runs are connected by the λc(H, ↑)= − λc( − H, ↓) symmetry. The λc-H magnetostriction measurements show a wide hysteresis even up to the maximum H=90 kOe field with no corresponding hysteresis in the magnetization measurements. Figure 3 shows the field induced polarization at selected temperatures for a complete cycle of the H field. The ME poling procedure was the same as before with E0=+1.5 kV/cm, H0=+50 kOe fields and E∥a, H∥a*. Measurements at higher temperatures show similar P-H curves up to TN1 as those shown in Fig. 2(b). When the H field is swept through the partial spin-flop like phase transition48,49 at H ≈ ± 55 kOe, P shows a sudden drop (ΔP) and χME is strongly reduced. When the field is decreased again below H=55 kOe (see (1) and (2) in Fig. 3), the change in P is significantly reduced, which suggests the emergence of a multi-domain ME state. In the paramagnetic phase, the linear ME effect and the drop in P at H=+55 kOe disappear. Instead, P is quadratic in field (P ~ H2), which confirms the magnetic origin. Similarly to the λc magnetostriction and to the ϵa dielectric constant measurements for H∥a* in ref. 60, the P-H loops also show a hysteresis up to the highest H field. Figure 3(b) shows the magnitude of the temperature dependence of the ΔP polarization drop measured at the spin-flop transition. The magnitude of ΔP gradually decreases with increasing temperatures, up to TN1, where the spin-flop transition and ΔP disappear. The magnetic field dependence of the out-of-plane polarization (P∥c) at T=8 K is shown in Fig. 3(c). As expected due to lattice symmetry, the out-of-plane P is almost zero at all fields and temperatures. However, in the field region of the spin-flop transition, the out-of-plane P shows a small peak-like anomaly with 20 times smaller magnitude than that of the in-plane ΔP. This peak is not related to misalignment, as unlike the in-plane P, it is mainly symmetric to the reversal of the magnetization. Therefore, it is likely related to a non-collinear and non-planar spin structure transiently appearing in the spin-flop phase. Temperature dependence of the ME effect in Na2Co2TeO6. a Prior to the measurements the sample was cooled to the ordered phase in the presence of poling (E0=+1.5 kV/cm, H0=+50 kOe) fields, E∥a, H∥a*, while measurements were started from H=+50 kOe in the absence of E field. Dashed curves indicate measurements where the H field was first swept to +90 kOe then reversed to -90 kOe, while in case of the solid curves the H field was immediately swept towards -90 kOe. The P-H loops were stacked along the vertical axis by 15 μC/m2 for better visibility. The linear ME effect and the drop in P at H=+55 kOe disappears above TN1. b Temperature dependence of the ΔP polarization change at the spin-flop phase transition. c In-plane magnetic field dependence of the out-of-plane polarization measured at T=8 K (H∥a*, P∥c). Discussion Using the combination of all heat capacity, magnetization, dilatometry, and polarization measurement data, we derive the magnetoelastic and ME phase diagram of Na2Co2TeO6 for H∥a* field shown in Fig. 4(a). Further temperature and magnetic field dependent measurements are shown in the Supplementary Information. For H∥a*, the phase diagram is significantly more complex than those of earlier reports24,49,60. Presumably, many of these new H-field-induced elastic transitions are related to changes in the interlayer stacking arrangement of the canted AFM or triple-q order along the c axis (meaning the spin order within the aa* plane is shifted relative to that in adjacent layers) or to the rearrangement of the Na+ ions due to shearing deformations, as illustrated in the supplementary information. To clarify these details, further neutron diffraction measurements are needed. We find evidence for a linear ME coupling in the field region ∣H∣ ⩽ 55 kOe (indicated by the purple area), while the decrease in P for higher fields coincides with the appearance of the suggested and debated field-induced quantum spin liquid phase41,61, which is turned into the field-induced ferromagnetic phase only at H=95 kOe. As we have directly observed both magnetoelasticity and magnetoelectricity, and the magnetic point group of the triple-q (P632’2’) order allows for the piezomagnetoelectric effect (P=π^Hσ^, where π^ is the piezomagnetoelectric tensor and σ^ is the stress tensor)62, we conclude that Na2Co2TeO6 in the triple-q state is piezomagnetoelectric; i.e., uniaxial stress can enhance the ME effect, or magnetic (electric) field can allow the piezoelectric (piezomagnetic) effect. The magnetoelectric and magnetoelastic H-T phase diagram of Na2Co2TeO6. a The phase diagram for H∥a* field is based on all magnetization, heat capacity, dilatometry and ME measurements. Gray symbols correspond to magnetic, cyan symbols to heat capacity, purple symbols to P-H, and red symbols to λc measurements. Colored regions indicate the zigzag/triple-q (zz/tq), the non-colinear spin-flopped (nc-sf), and the field-induced ferromagnetic (fm) phases. b In the triple-q (tq-a) phase the Ptot total polarization of the unit cell is reversed for the reversal of the H field (linear ME effect). The χME ME susceptibility is proportional to the T toroidic moment. c Sign of the ME effect is reversed when the other AFM state (tq-b) is selected. d Illustration of a free energy landscape with two stable ME states and a metastable QSL state, which is barely affected by the fields. The initial ME2 state is destabilized by the application of E and H fields. While the ME1 state is more stable, the system is trapped in the metastable QSL phase even when the fields are removed. e The pd-hybridization can raise a local polarization (P) around the magnetic ion even in the absence of long-range magnetic order. The ME coupling transfers the effect of the frustrated bond-dependent interactions of the spin system on the frustration-free polarization system to create a polarization liquid state. We now turn to the origin of the ME effect. It is straightforward to exclude the exchange striction mechanism, the lattice is built up by uniformly charged Co2+ ions, and thus there is no reason to expect dimerization within the ab plane, needed for finite P, despite the sizable magnetostriction. Besides, P of exchange striction origin should be quadratic in H field at all temperatures and not linear as we observe here. If the exchange striction mechanism is significant, then the numerous sharp peaks in the λc-H should also show up in P-H. In case of Na2Co2TeO6, the spin-current mechanism can in principle explain the linear ME effect via either the off-diagonal exchange interaction (Dzyaloshinskii-Moriya) or via the ring-exchange interaction53. The Dzyaloshinskii-Moriya interaction leads to the canting of the spins, which can also explain the weak-ferromagnetic moment. However, upon the reversal of the weak ferromagnetic moment, χME should also reverse, i.e., the P-H curves should be again symmetric and above TN1 the P would be zero, and therefore we regard the spin-current mechanism less relevant. As discussed in Fig. 1(b), the highly distorted O6 octahedra and the local symmetries of the crystal allows for the emergence of a local P with an antiferroelectric order both in case of the orthorhombic zigzag and the trigonal triple-q phases. This is consistent with the pd-hybridization mechanism42,43, where the hybridization between the magnetic ion d and the ligand p orbitals is affected by the orientation of the spin, leading to broken inversion symmetry, allowing for local polarization at the magnetic ion even independently from long-range magnetic order44. The distortion of the ligand cage around the Co2+ ion allows finite P=∑ℓ(S⋅eℓ)2eℓ to appear in strong coupling to the spin, which gives a finite ME effect only in case of the triple-q order. As illustrated in Fig. 4(b, c), the triple-q order can show linear ME effect on the macroscopic scale due to the simultaneous canting of the S and the local P, while the sign of the ME effect depends on the selection of the AFM domain. Besides, pd-hybridization is known to permit quadratic H-field dependence above TN1, similarly to that of Ca2CoSi2O744. Therefore we identify the pd-hybridization mechanism and the emergence of local P to be the dominant source of the ME effect in Na2Co2TeO6, while the spin-current mechanism could in principle be relevant as well. In summary, our direct measurements prove the simultaneous presence of magnetoelectric (ME) and magnetoelastic effects in Na2Co2TeO6, which we conclude as the first example for a Kitaev QSL candidate material with piezomagnetoelectricity. Although our observations provide direct evidence for piezomagnetoelectricity, the nature and origin of the local polarization should be further tested by site-selective experimental methods, such as Second Harmonic Generation, Nuclear Quadrupole Resonance, Resonant Inelastic X-ray Scattering. Using the thermodynamic, magnetic, ME, and magnetoelastic measurement data, we map out the phase diagram for H∥a* featuring the newly discovered ME and magnetoelastic phase transitions. Based on the symmetries of the lattice and the H field dependence of the ferroelectric polarization, we point out the pd-hybridization mechanism as the dominant origin for the ME effect. The coexistence of the ME effect with the Kitaev interaction can have interesting and important consequences motivating further experimental and theoretical research, as illustrated by two examples in Fig. 4(d, e). Inspired by earlier research on ME materials39,63,64, the combination of E and H field could be used as new control parameters to stabilize a QSL ground state, as shown in Fig. 4(d). In a Kitaev magnet with coexisting stable ME and metastable QSL phases the application of E and H field can destabilize a preset ME state, while leaving the QSL state unchanged. During the switching from the instable to the stable ME states, the sample can get trapped in the metastable QSL phase. A similar interaction between ME and non-ME phases has been observed in Y-type hexaferrites64. On the other hand, in ME materials with the pd-hybridization mechanism, a local polarization (P) at the magnetic ion can emerge even in the absence of long-range magnetic order, which in principle may lead to a polarization liquid state to accompany the QSL state, as shown in Fig. 4(e). Methods The platelet-shaped single crystals of Na2Co2TeO6 were grown with a flux method described in refs. 48,49. The field-dependent magnetization and heat capacity data up to H=140 kOe were measured with a vibrating sample magnetometer and with the heat capacity option of the Physical Property Measurement System (PPMS, Quantum Design). Pulsed-field magnetization up to H=500 kOe was measured at the Hochfeld-Magnetlabor Dresden (HLD), using a compensated pickup-coil magnetometer in a 4He flow cryostat65. Thermal expansion and magnetostriction were measured using the PPMS-compatible Kuechler-mini dilatometer probe, which employs the capacitance measurement technique (AH2700A, Andeen-Hagerling)66,67. The length change was measured along the c axis (ΔLc∥c), on a 190 μm-thick sample with 2.8 mm2 area. During the thermal expansion measurement the temperature of the sample was swept at a constant 0.25 K/min rate, while in case of the magnetostriction measurement the magnetic field was swept with 10 Oe/s rate at constant temperatures. In case of the capacitance measurement technique a quasi-constant ~ 4 N force is applied on the sample by a spring, which equals to 1.4 MPa uniaxial pressure. We highlight that the application of uniaxial pressure cannot introduce changes in the lattice symmetry. The linear thermal expansion coefficient along the c axis (αc) was calculated as the temperature derivative of the relative length change: where the Lc(300 K, 0 kOe) is the total thickness of the sample at room temperature. The linear magnetostriction coefficient (λc) was calculated as the H-field derivative of the relative length change: The ferroelectric polarization was measured in the PPMS equipped with an electrometer (6517A, Keithley). The silver-paint electrodes were placed along the 190 μm-thick sides of the sample parallel to the a* axis, i.e., P∥a, which provided a S=0.4085 mm2 cross section for the measurement the distance of the electrodes was 1.6 mm. For the P∥c measurement, we used the same sample with silver-paint electrodes of S=0.9 mm2 cross section. In case of the ME measurements, we applied the following ME poling procedure. The sample was warmed to T=50 K prior to every P-H cycle measurement, then cooled to the ordered phase in the presence of E0= ± 1.5 kV/cm and H0= ± 50 kOe fields (E0∥a, P∥a, H0∥a*), which is the so-called ME poling procedure. Following the poling, the E field was removed and the leads were shorted for 10-minute-long intervals for a dozen times for electrostatic discharge, then P was measured in sweeping H field starting from H0. This long discharging time was highly important, as due to the in-plane Na+ ion conduction the sample acted as a discharging battery68. The ME measurements were done on the same crystal piece, which we earlier used for dilatometry measurements. The applied 1.4 MPa uniaxial pressure can therefore introduce substantial residual strain along the c axis. As the πaa*cc component of the piezomagnetoelectric tensor is finite, the σcc uniaxial stress can introduce or enhance the χaa* = πaa*ccσcc ME tensor element. Here we note that earlier measurements did not find ME effects in Na2Co2TeO660. The application of uniaxial stress prior to our ME experiments can explain why earlier investigations and measurements on samples without uniaxial pressure did not show ME effect. The calculated χ^ME and π^ tensors were cross-checked using the MTENSOR application69, while magnetic ground states of various QSL candidates were taken from MAGNDATA70 of the Bilbao Crystallographic Server. Supplementary information Supplementary Information Transparent Peer Review file Acknowledgements The authors are grateful to fruitful discussions with Izabella Lovas, Luis Elcoro, Satoshi Nishimoto, Xiaochen Hong, Christian Hemker-Heß, Lukas Janssen, and Matthias Vojta. Illustration of the structural unit cell was created using the software VESTA71. Author contributions The thermal expansion, magnetostriction, heat capacity and magnetization measurements were done by V.K. with the help of N.P. and A.U.B.W. The magnetoelectric measurements were performed by V.K., W.Y., and Y.L grew the single crystals. S.L. and H.K performed the pulsed-field magnetization measurements. A.U.B.W., V.K., and B.B conceived the project. V.K. wrote the manuscript with the help of Y.L., A.U.B.W., and B.B.; all authors contributed to the discussion of the results. Peer review Peer review information Nature Communications thanks the anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available. Funding We acknowledge financial support from the German Research Foundation (DFG) through the Individual Research Grant (project-id 540912241), the Collaborative Research Center SFB 1143 (project-id 247310070), and the Würzburg-Dresden Cluster of Excellence on Complexity, Topology and Dynamics in Quantum Matter ctd.qmat (EXC 2147, project-id 390858490). The work at Chinese Academy of Sciences and Peking University was supported by the National Basic Research Program of China (Grant No. 2025YFA1411501) and by the National Natural Science Foundation of China (Grant No. 12474138). We acknowledge the support of HLD at HZDR, member of the European Magnetic Field Laboratory (EMFL). V. K. was supported by the Alexander von Humboldt Foundation. Open Access funding enabled and organized by Projekt DEAL. Data availability The data presented in the current study are available from the corresponding authors on request and shared on public data repository ZENODO (10.5281/zenodo.19397244). 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. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-73747-0. References Associated Data Supplementary Materials Supplementary Information Transparent Peer Review file Data Availability Statement The data presented in the current study are available from the corresponding authors on request and shared on public data repository ZENODO (10.5281/zenodo.19397244).