betaFeedback

WorksPhysics and Astronomy

Unveiling the impact of anti-site defects in magnetic transitions of few-layer MnBi2Te4 by operando heating

Abstract

As the first experimentally discovered intrinsic magnetic topological insulator, MnBi2Te4 has attracted widespread attentions, providing a unique platform for the exploration of topological quantum phases, such as quantum anomalous Hall effect and axion insulator state. Despite the increasing number of potential factors affecting samples being identified, obtaining the high-quality device performance with desired topological quantum phases remains a challenge. In this work, by comparing the reflective magnetic circular dichroism (RMCD) of crystals with different defect densities that are characterized by atomically resolved scanning tunneling microscopy, we demonstrate that anti-site defects play an essential role in achieving ideal magnetic states. By measuring RMCD hysteresis loops with operando heating, we find that MnBi2Te4 few-layer samples are highly susceptible to thermal impact, even at temperature as low as 45°C. The magnetic behavior of heating-treated samples is akin to that of samples fabricated into devices, revealing the thermal impact on devices as well. Starting from few-layers with ideal layer-dependent magnetic order, thermal heating leads to the convergence of magnetization and transition fields between odd- and even-layers. The observed heating-induced magnetic evolution can serve as a valuable reference for assessing the sample quality or the density of anti-site defects. Our findings not only point out the long-standing hidden factor that arose controversies in MnBi2Te4, but also pave the way for controllably engineering the topological quantum phenomena.

† These authors equally contributed to this work.

Introduction

The interplay between magnetism and topology allows for the exploration of exotic quantum phenomena including the quantum anomalous Hall (QAH) effect[1, 2, 3, 4, 5, 6, 7] and the axion insulator state[8, 9, 10, 11, 12, 13], which have been investigated in magnetically doped topological insulator and intrinsic magnetic topological insulator. MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4}, as the first intrinsic magnetic topological insulator with better magnetic homogeneity, sparks extensive research interest[5, 6, 7, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24]. As shown in Figure 1a, its crystal structure consists of septuple layers (SLs). Figure 1b depicts its A-type antiferromagnetic structure, giving rise to layer-dependent topological states: Odd-number-SL MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} with uncompensated magnetization leads to the QAH effect, while even-number-SL MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} with zero net magnetization is predicted to achieve the axion insulator phase[16, 17, 18].

Appearance of spin-flip transitions in both odd- and even-layer $\mathrm{MnBi}_{2}\mathrm{Te}_{4}$

Figure 1. Appearance of spin-flip transitions in both odd- and even-layer MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4}. a, Crystallographic structure of MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4}. b, Schematic of the theoretically proposed layered antiferromagnetic ground state of MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4}. The spins of the Mn2+\mathrm{Mn}^{2+} ions ferromagnetically couple to each other within one septuple layer, while the adjacent layers are antiferromagnetically coupled. c–d, Variable magnetic field transverse resistance (RyxR_{yx}) of 5SL and 4SL MBT samples at the temperature of 6 K and 2 K, respectively. The back-gate voltage was tuned near the charge neutral point of the sample. Inset figure in d is the zoomed-in section that highlights the hysteresis loop.

Despite tremendous progress in recent years, further exploration of the topological quantum states in MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} remains a challenge. For odd-number-SL MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4}, the anomalous Hall effect often exhibits varying degrees of quantized Hall conductance across different samples[6, 20, 24, 25, 26, 27]. The deviation from quantization is clearly exemplified on a 5SL MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} device (Figure 1c). For even-number-SL MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4}, odd-layer-like uncompensated magnetic hysteresis loops are often observed both in magneto-optical spectroscopy[26, 28] and transport[5, 26, 27, 29] in previous studies, hindering the realization of axion insulator state. We also observed the unexpected loop in the transport measurement of a 4SL device in Figure 1d. Various external factors are identified to hamper the emergence of ideal topological states, such as surface degradation by air exposure[30] and polymer coatings[27]. However, even maintaining fabrication process isolated from air and chemicals, these devices still exhibit nonideal states both in odd- and even-layers (Figure 1c and Figure 1d). This dilemma requires further identification of crucial factors that suppress the observation of ideal topological states.

In this work, by comparing RMCD hysteresis loops from crystals with varying defect densities that are characterized by atomically resolved scanning tunneling microscopy, we unambiguously identify density of anti-site defects as a crucial factor in determining magnetic ground state and therefore the topological state. Beginning with ideal layer-dependent magnetism in MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} few-layers with low defect density, upon operando heating, the net magnetization and transition fields of the odd- and even-layers progressively converge, diminishing the contrasting odd-even-layer dependence. As the magnitude of spin-flip transition fields observed in heating-treated samples reproduces the magnetic behaviors found in fabricated devices, we point out that the nonideal behaviors in devices result from the effective heating during the electrode evaporation process, which is a commonly-used but long-overlooked factor. Our results reveal the thermal impact on the behaviors of magnetic topological states, shedding light on the manipulation of magnetism-topology interactions in this intriguing material.

Results

All MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} few-layers used in this work were mechanically exfoliated onto SiO2\mathrm{SiO}_{2}/Si substrate using adhesive tape (Methods). A typical microscopic image of MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} thin flake with different thickness is shown in Supplementary Fig. 1a. The number of layers was precisely determined by their optical contrast combined with atomic force microscopy[31] (Supplementary Fig. 1b). The sample preparation process was conducted in an argon-filled glove box, including the device fabrication for MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} samples with electrodes described later in this work. The samples were sealed and transferred to our magneto-optical cryostat in high vacuum for the RMCD and transport measurements.

We performed RMCD measurements on MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} few-layers exfoliated from two types of crystals. Type-A crystals were grown by an optimized flux method, whereas Type-B crystals were grown under normal flux method (Methods). Figure 2a and Figure 2b show the RMCD hysteresis of freshly exfoliated flakes obtained from type-A and type-B crystals, with the layer numbers ranging from 4SL to 7SL. For the spin-flop transition between the collinear AFM and canted AFM states, samples from both crystal types exhibit the even-odd-layer oscillation of transition fields. As labeled by Hc2H_{\mathrm{c2}} with green arrows in Figure 2a and Figure 2b, the transition field is about 2 T for even-layer flakes and 3.5 T for odd-layer flakes. For the spin-flip transition between the collinear AFM states (labeled by Hc1H_{\mathrm{c1}} with orange arrows in Figure 2a and Figure 2b), however, flakes from different crystal types exhibit distinct features: even-layer samples from Type-B crystals show hysteresis loops, whereas samples from Type-A crystals do not. In samples from Type-A crystals, as the magnetic field sweeping upward or downward, the odd-layer MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} samples exhibit a sudden jump, while no spin-flip transition can be identified in the even-layer ones. This finding is consistent with the theoretically proposed A-type antiferromagnetic structure[16, 17, 18]. In stark contrast to even-odd-layer dependent magnetic states, for flakes from Type-B crystals, both even- and odd-layers display prominent hysteresis loops, which is similar to previously reported optical[26, 28] or electrical transport results[5, 26, 27, 29].

Image not extracted

Figure 2. Distinct RMCD results in MnBi2_2Te4_4 with different defect density. a, b, RMCD loops of the freshly exfoliated 4SL to 7SL MnBi2_2Te4_4 samples exfoliated from the crystals grown by optimized flux method (a) and normal flux method (b). The measuring temperature is 6 K. Curves are vertically offset for comparison. c, d, corresponding STM topographies of the MnBi2_2Te4_4 samples exfoliated from crystals grown by optimized flux method (c) and normal flux method (d), respectively. The concentration of the MnBi_\mathrm{Bi} anti-site defect is about 7% in samples grown by normal flux, which is almost four times larger than that in samples grown by the optimized flux (1.9%). The density of BiTe_\mathrm{Te} defect is 0.15% and 0.79% in the optimized and normal flux method grown samples, respectively. Bias: 1 V, setpoint: 10 pA (c) and 100 pA (d). The scale bar is 10 nm.

To uncover the underlying mechanisms behind the distinct magnetic ground states observed in different types of crystals, we employed scanning tunneling microscopy (STM) for detailed characterizations. As shown in Figure 2c and 2d, both samples exhibit two types of defects, including the dark triangular MnBi\mathrm{Mn}_{\mathrm{Bi}} anti-site defect and the bright circular BiTe\mathrm{Bi}_{\mathrm{Te}} anti-site defect[32, 33, 34, 35] (see Supplementary Fig. 2 for detailed defect structures), but the defect density is much lower in type-A crystals compared to type-B.

In the type-A crystals grown by the optimized synthesis, the concentration of MnBi\mathrm{Mn}_{\mathrm{Bi}} anti-site defects is about 1.9%, which is almost four times sparser than that in crystals grown by the normal synthesis (7%). Moreover, the density of BiTe\mathrm{Bi}_{\mathrm{Te}} defects is 0.15% and 0.79% in the optimized and normal flux method grown crystals, respectively. The correlation between defect density and hysteretic response in MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} few-layers suggests that the magnetic order is modified by the defects. As previously proposed[33, 36, 37], Mn atoms in the Bi layer can introduce local magnetic moment with its orientation antiparallel to that of the Mn layer, leading to the uncompensated hysteresis loop. Therefore, controlling the defect density during growth is crucial for high-quality crystals with ideal magnetic structure[38, 39, 40].

Utilizing optimized crystals with low defect density and ideal magnetic states, we fabricated MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} devices and performed simultaneous transport and RMCD measurements to investigate the magnetic topological states. Figure 3a and 3b show the longitudinal (RxxR_{xx}) and transverse (RyxR_{yx}) resistances as a function of magnetic field in a typical 5SL MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} device at various back-gate voltages (VgV_g). The sample was measured at 6 K, well below its Néel temperature of about 23 K (Supplementary Fig. 3a). The quantized Hall resistance Ryx=±h/e2R_{yx} = \pm h/e^2 and nearly vanishing longitudinal resistance RxxR_{xx} in high magnetic field regions denote a Chern insulator state with Chern number (CC) of ±1\pm1 at Vg=−30V_g = -30 V, where the Fermi level is close to the charge neutral point (CNP) within the surface gap (Supplementary Fig. 3b-d). The reversal of anomalous Hall effect in Figure 3b may come from the competition between intrinsic Berry curvature and Dirac-gap enhanced extrinsic skew scattering in this material, because the skew scattering contributes a positive anomalous Hall effect and the intrinsic contribution from Berry curvature is negative[41]. In such electrical transport measurements, both spin-flip and spin-flop transitions are captured by the jumps at about ±0.5\pm0.5 T (Hc1H_{c1}) and ±3\pm3 T (Hc2H_{c2}), respectively. These transitions also occur in the simultaneously measured magnetic-field dependent RMCD at the exactly same magnetic fields (Figure 3c). Note that, although the transport data vary with the Femi level tuned by gate voltage, the RMCD results remain independent of the gate voltage. This observation demonstrates that the RMCD retrieves exclusive magnetization information, regardless of the specific electronic properties of materials.

Image not extracted

Figure 3. Comparison between electrical transport and the corresponding RMCD results in 5SL and 4SL MnBi₂Te₄ devices. a-c, Magnetic-field dependence of RxxR_{xx} (a), RyxR_{yx} (b), and RMCD signal (c) for a 5SL device at varying gate voltages. d, Optical microscopic image of the 5SL device measured in a-c. Metal (Au) contacts to the sample were e-beam deposited through a stencil mask in a Hall bar geometry. e, f, RMCD images of the 5SL device taken at zero magnetic field, sweeping from -6.5 T to 0 T (e) or from 6.5 T to 0 T (f). Both of the images show a single domain structure throughout the sample. g-i, Magnetic-field dependence of RxxR_{xx} (g), RyxR_{yx} (h), and RMCD signal (i) for the 4SL device at varying gate voltages. j-l, Optical microscopy (j) and RMCD images at 0 T (k, from -6.5 T and l, from 6.5 T) of the 4SL device measured in g-i. The distinct magnetizations at zero magnetic field between upward and downward sweeping are captured by the RMCD images.

The RMCD also enables the examination of possible domain structures through microscopic imaging that could not be obtained by transport measurements. Figure 3d and 3e-f show the optical microscopy of the 5SL MnBi2Te4\mathrm{MnBi_2Te_4} device and the corresponding RMCD images at zero field, respectively. When the magnetic field was swept upward from -6.5 T (Figure 3e) or downward from 6.5 T (Figure 3f), both the RMCD images exhibit a homogeneous intensity across the device, but with the opposite magnetization. The RMCD microscopy suggests that, unlike what was reported for bulk MnBi2Te4\mathrm{MnBi_2Te_4}[42, 43], the odd-layer device contains a single magnetic domain with non-zero switchable magnetization.

The RMCD and transport measurements were also conducted in a 4SL MnBi2Te4\mathrm{MnBi_2Te_4} device (Figure 3g-Figure 3i). Similar to the odd-layer devices, a high-field Chern insulator state is also achieved at gate voltage ranging from -25 V to 15 V, centered around the CNP of -5 V (Supplementary Fig. 4). Strikingly, a spin-flip transition arises in the magneticfield dependent RMCD loop, distinct from that in the freshly exfoliated 4SL sample presented in Figure 2a. Such a transition can also be distinguished in the magnetic-field dependent RxxR_{xx} and RyxR_{yx}, despite the large noise in RyxR_{yx} at CNP due to the giant longitudinal resistance. The corresponding optical microscopy and RMCD images of the device are shown in Figure 3j-Figure 3l, again demonstrating that the emergent magnetic state uniformly exists in the entire device.

By further comparing the details of magnetic transitions between the freshly exfoliated 4SL sample (Figure 2a) and the fabricated 4SL device (Figure 3l), two key distinctions can be observed. The most evident one, as mentioned above, is the appearance of spin-flip transitions in the device, suggesting the formation of an additional magnetic order. Moreover, the transition field of the spin-flop transition has expanded from about 1.7 T to 2.0 T. These distinctions suggest a variation in the magnetic ground state of MnBi2_2Te4_4 few-layers after being fabricated into devices. Because we directly deposit electrodes onto samples covered by a stencil mask[5, 29], the samples remain free from any polymers or chemical solutions. Such a variation likely originates from the electrode deposition process, which prompts us to investigate the key factor that drives this modification.

To investigate this factor, we performed a series of RMCD measurements at 6 K after the as-exfoliated MnBi2_2Te4_4 few-layers were consecutively heated up to 90 °C inside the magneto-optical cryostat (see Methods for details of the thermal treatment). Figure 4a-Figure 4d show the RMCD results before and after the consecutive heating processes for 3-6SL, respectively. Variations in both spin-flip and spin-flop transitions can be identified in the odd-layers upon heating. A spin-flip transition emerges in the 4SL and 6SL samples after heating up to 60 °C and 75 °C, respectively. The transition field of their spin-flop transition also changes with heating temperature. These observations suggest that the thermal treatment alters the magnetic structure of MnBi2_2Te4_4 few-layers.

Magnetic evolution upon thermal treatment in MnBi$_2$Te$_4$ few-layers

Figure 4. Magnetic evolution upon thermal treatment in MnBi2_2Te4_4 few-layers. a-d, Magnetic-field dependent RMCD signals acquired in 3-6 SL MnBi2_2Te4_4 before and after heating with temperatures ranging from 45 °C to 90 °C. The samples were exfoliated at room temperature and marked as 21 °C. All of the RMCD data were taken at the temperature of 6 K. e, f, The extracted spin-flip (Hc1H_{c1}, e) and spin-flop (Hc2H_{c2}, f) transition field as functions of heating temperature. The transition fields are extracted from the positive-field side of the data, as the difference between positive and negative transition fields is less than 0.1 T. g, Remnant RMCD intensity at zero magnetic field as a function of heating temperature. All of the data shown in e-g are taken from multiple few-layers located in the same substrate. Several samples with the same number of layers are measured, and denoted by various of symbols. This allows for a direct comparison of the magnetic-field dependent RMCD between samples with the same or different numbers of layers.

The characteristics of RMCD hysteresis loops in MnBi2_2Te4_4 few-layers are extracted and plotted in Figure 4e-Figure 4g. For a direct and fair comparison, all the measured samples are on the same SiO2_2/Si substrate, and can contain multiple regions with the same thickness. Figure 4e and 4f show the transition fields of the spin-flip (Hc1H_{\mathrm{c1}}) and spin-flop (Hc2H_{\mathrm{c2}}) transitions, respectively, as functions of heating temperature. Here we only extract the positive critical fields, because of the absence of a noticeable exchange bias effect under a symmetric, high magnetic sweep and thus the negligible difference between two sides. For the odd-layers, Hc1H_{\mathrm{c1}} increases with heating temperature, whereas Hc2H_{\mathrm{c2}} decreases. By contrast, the even-layers show the opposite trends in both Hc1H_{\mathrm{c1}} and Hc2H_{\mathrm{c2}}. The same contrast is observed in the remnant RMCD intensity at zero magnetic field (Figure 4g), which increases monotonically with heating temperature in the even-layers, while it decreases in the odd-layers. Eventually around the heating temperature of 90 °C or higher, the hysteresis loops become alike between the odd- and even-layers, making the identification of layer parity more difficult.

Such magnetic evolution upon heating is further reproduced in multiple samples located on three SiO2_2/Si substrates (Supplementary Fig. 5). For the transition fields Hc1H_{\mathrm{c1}} and Hc2H_{\mathrm{c2}}, minor fluctuations among different samples are observed, and can be attributed to the inhomogeneous defects[44, 45] or undesirable strains[46, 47] in the samples, which result in the field-pinning of domain wall and consequently modify the magnetic transition fields. Meanwhile, the remnant RMCD (Figure 4g), which is proportional to the out-of-plane magnetization, has better consistency between different samples with the same thickness. Nevertheless, the RMCD results demonstrate the vulnerability of the MnBi2_2Te4_4 few-layers subject to heating, with both the magnetization and transition fields exhibiting significant changes at a heating temperature as low as 45 °C.

Magnetic behavior of the MnBi2_2Te4_4 devices (Figure 3) is therefore reproduced by heating the as-exfoliated few-layers (Figure 4), revealing the thermal impact of electrode deposition for device fabrication as the key factor. Moreover, the transition fields (Hc1H_{\mathrm{c1}} and Hc2H_{\mathrm{c2}}), along with the remnant RMCD intensity, as a function of heating temperature can serve as a reference guide to evaluate the thermal impact of electrode deposition in MnBi2_2Te4_4 devices. Examining the hysteresis loops for the devices shown in Figure 3, the equivalent heating temperature is estimated to about 90 °C for Au electrodes. When the even-layer 4SL device was made with additional Ti deposition as a wetting layer for Au electrodes, as shown in Supplementary Fig. 6, the hysteresis loops seem more like those of an odd-layer. The equivalent heating temperature thus becomes even higher than 90 ∘C90\,^\circ\mathrm{C}. This temperature increase is understandable because the Ti deposition imposes more severe thermal impact to the samples than the Au evaporation.

Because the magnetic property of MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} few-layers, particularly the appearance of spin-flip transitions in the even-layers, is directly correlated with the higher density of anti-site defects, as illustrated in Figure 2, the heating effect might be closely related to the accumulation of such defects. As the top surface layer of MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} is more susceptible to thermal impact, here we note that the defects are likely unevenly distributed between different layers, with the top surface having relatively higher defect density. Our experimental findings would inspire the future studies on how the defect distribution, or more generally the defect engineering, actively affects the exotic quantum phenomena and performance in MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4}[38, 39, 40].

Conclusions

In summary, we find that MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} crystals with different defect densities exhibit distinct magnetic transition behaviors under an out-of-plane magnetic field, highlighting the crucial role of anti-site defects in shaping their magnetic properties. By simultaneously conducting transport and RMCD measurements, we establish the connection between electrical transport properties and magnetism in MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} few-layers that conforms to the layer parity. Utilizing RMCD measurements with operando heating, we find the magnetic property of MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4} is highly sensitive to heating, even at temperature as low as 45∘C45^\circ\mathrm{C}. Moreover, the remnant magnetization can serve as a reference for evaluating the sample quality or the density of anti-site defects. These insights are crucial for advancing the development of MnBi2Te4\mathrm{MnBi}_{2}\mathrm{Te}_{4}-based quantum devices, and gaining a deeper understanding of its magnetic and topological properties.

Methods

MnBi2_2Te4_4 crystal growth and selection

MnBi2_2Te4_4 single crystals were synthesized through the self-flux growth method.

High-purity Mn (99.95%), Bi (99.999%), and Te (99.999%) powders with a molar ratio of 1:2:4 were thoroughly mixed and loaded into a silica crucible. The crucible was subsequently sealed within a silica ampoule under vacuum conditions (<10−4 mbar<10^{-4}\ \mathit{mbar}).

In the case of type-A crystals, we implemented an optimized growth protocol involving extended isothermal holding at the crystallization temperature combined with a slow annealing process. This modified approach significantly reduced the defect density in the crystals, enabling the production of high-quality MnBi2_2Te4_4 samples. Detailed growth parameters and optimization procedures are described in ref. [48]. For the type-B crystals, the mixture was initially heated to 900∘C900^\circ\mathrm{C} and maintained at this temperature for 24 hours to ensure complete homogenization. It was then subjected to a controlled cooling process, gradually decreasing the temperature to 595∘C595^\circ\mathrm{C} over 150 hours, followed by immediate centrifugation at this temperature to separate the crystals from the flux.

Sample preparation and device fabrication

MnBi2_2Te4_4 few-layer samples were obtained by standard mechanical exfoliation onto plasma cleaned Si substrates with 285-nm-thick SiO2_2 coating. The flake thickness was determined by optical contrast and atomic force microscopy (AFM). Proceeding to the device fabrication, a Hall bar configuration was achieved by initially placing a pre-patterned SiN membrane window over the selected thin flake, serving as a precise stencil mask. We then utilized an electron-beam evaporator to deposit electrode films (60-nm Au or 5-nm Ti/60-nm Au) through the stencil mask. The whole process was carried out in a glove box with argon/nitrogen atmosphere (H2_2O < 0.1 ppm, O2_2 < 0.5 ppm). To prevent from air exposure, all the samples were sealed with a cover glass by vacuum grease in custom-built sample holders and transferred from glove box to the magneto-optical cryostat swiftly.

RMCD and transport measurements with in situ heating

All the RMCD and transport measurements were conducted in a home-built variable-temperature magneto-optical cryostat down to 6 K and up to 7 T. The thermal treatment was performed in situ inside the same cryostat, involving: heating the substrate to specific temperatures, maintaining at each temperature for a duration of 2 hours, and cooling down to the base low temperature for RMCD and transport measurements. Most of the measurements were obtained at the base temperature of 6 K, unless otherwise noted. Each sample was sequentially heated to 45°C, 60°C, 75°C, and 90°C because of the sample irreversibility. The entire process was meticulously conducted under high vacuum conditions, with a pressure maintained below 1×10−81 \times10^{-8} Torr inside the magneto-optical cryostat.

The RMCD measurements were in the backscattering configuration. The magnetic field was applied perpendicular to the sample plane. A linearly polarized 633 nm HeNe laser (Thorlabs), was used. After passing through an optical chopper (Thorlabs) at 197 Hz, its polarization was modulated by a photoelastic modulator (PEM) at 50.052 kHz. Using a 50× objective (Nikon, NA 0.55), the laser beam was focused onto the sample with the spot diameter of ∼2\sim2 μm. The excitation power was 1 μW. The reflected light was detected by an Avalanche Photodiode (Thorlabs), and analyzed by a lock-in amplifier (Zurich Instruments). The magnitude of RMCD was determined by the ratio of first-harmonic components at 50.052 kHz and 197 Hz.

Transport measurements were performed by using a standard lock-in technique with an excitation current of 10∼8010 \sim80 nA (Signal recovery 7265). The bottom gate voltage VgV_g was controlled by a source meter (Keithley 2400). The RxxR_{xx} (RyxR_{yx}) data shown in this paper were symmetrized (anti-symmetrized).

Data availability

The data supporting the findings of this study are available within the paper. Source data are provided with this paper.

Acknowledgements

The work at Fudan University was supported by National Key Research and Development Program of China (Grant Nos. 2022YFA1403300, 2024YFA1409800), National Natural Science Foundation of China (Grant Nos. 12034003, 12427807, 12204115, 1235000130, 12350404), Science and Technology Commission of Shanghai Municipality (Grant Nos. 20JC1415900, 23JC1400400, 23JC1400600, 24LZ1400100, 2019SHZDZX01), Shanghai Municipal Education Commission (Grant No. 2021KJKC-03-61), and China Postdoctoral Science Foundation (Grant No. 2022M720812).

Author contributions

S.W.W. and Y.Z. conceived and supervised the project. X.Y.C. and S.W. conducted the transport and RMCD experiments with the assistance of J.J.G., Z.X.G., C.Y.H. and R.H.C. J.J.G., Z.X.G. and M.Y.L. grew the MnBi2_2Te4_4 crystals and prepared the few-layer samples/devices. Z.W.H. conducted the STM measurements. X.Y.C., J.J.G., S.W., Y.B.Z. and S.W.W. analyzed the data and wrote the paper with contributions from all authors.

Competing interests

The authors declare no competing interests.

Figures and Captions

References

  1. [1]Chang, C. Z. et al. Experimental observation of the quantum anomalous Hall effect in a magnetic topological insulator. Science 340, 167-170 (2013).DOI
  2. [2]Checkelsky, J. G. et al. Trajectory of the anomalous Hall effect towards the quantized state in a ferromagnetic topological insulator. Nat Phys 10, 731-736 (2014).DOI
  3. [3]Kou, X. et al. Scale-invariant quantum anomalous Hall effect in magnetic topological insulators beyond the two-dimensional limit. Phys Rev Lett 113, 137201 (2014).pubmed.ncbi.nlm.nih.gov/25302915
  4. [4]Chang, C.-Z. et al. High-precision realization of robust quantum anomalous Hall state in a hard ferromagnetic topological insulator. Nat Mater 14, 473-477 (2015).
  5. [5]Deng, Y. et al. Quantum anomalous Hall effect in intrinsic magnetic topological insulator MnBi₂Te₄. Science 367, 895-900 (2020).DOI
  6. [6]Lian, Z. et al. Antiferromagnetic quantum anomalous Hall effect under spin flips and flops. Nature 641, 70-75 (2025).DOI
  7. [7]Li, Y. et al. Reentrant quantum anomalous Hall effect in molecular beam epitaxy-grown MnBi₂Te₄ thin films. arXiv preprint arXiv:2401.11450 (2024).arxiv.org/abs/2401.11450
  8. [8]Mogi, M. et al. A magnetic heterostructure of topological insulators as a candidate for an axion insulator. Nat Mater 16, 516-521 (2017).DOI
  9. [9]Mogi, M. et al. Tailoring tricolor structure of magnetic topological insulator for robust axion insulator. Sci Adv 3, eaao1669 (2017).DOI
  10. [10]Xiao, D. et al. Realization of the Axion Insulator State in Quantum Anomalous Hall Sandwich Heterostructures. Phys Rev Lett 120, 056801 (2018).DOI
  11. [11]Liu, C. et al. Robust axion insulator and Chern insulator phases in a two-dimensional antiferromagnetic topological insulator. Nat Mater 19, 522-527 (2020).DOI
  12. [12]Gao, A. et al. Layer Hall effect in a 2D topological axion antiferromagnet. Nature 595, 521-525 (2021).
  13. [13]Qiu, J.-X. et al. Axion optical induction of antiferromagnetic order. Nat Mater 22, 583-590 (2023).DOI
  14. [14]Gong, Y. et al. Experimental Realization of an Intrinsic Magnetic Topological Insulator*. Chin Phys Lett 36, 076801 (2019).DOI
  15. [15]Otrokov, M. M. et al. Prediction and observation of an antiferromagnetic topological insulator. Nature 576, 416-422 (2019).DOI
  16. [16]Zhang, D. et al. Topological axion states in the magnetic insulator MnBi₂Te₄ with the quantized magnetoelectric effect. Phys Rev Lett 122, 206401 (2019).DOI
  17. [17]Li, J. et al. Intrinsic magnetic topological insulators in van der Waals layered MnBi₂Te₄-family materials. Sci Adv 5, eaaw5685 (2019).DOI
  18. [18]Otrokov, M. et al. Unique thickness-dependent properties of the van der Waals interlayer antiferromagnet MnBi₂Te₄ films. Phys Rev Lett 122, 107202 (2019).DOI
  19. [19]Ge, J. et al. High-Chern-number and high-temperature quantum Hall effect without Landau levels. Natl Sci Rev 7, 1280-1287 (2020).DOI
  20. [20]Cai, J. et al. Electric control of a canted-antiferromagnetic Chern insulator. Nat Commun 13, 1668 (2022).DOI
  21. [21]Zhang, Z. et al. Controlled large non-reciprocal charge transport in an intrinsic magnetic topological insulator MnBi₂Te₄. Nat Commun 13, 6191 (2022).DOI
  22. [22]Gao, A. et al. Quantum metric nonlinear Hall effect in a topological antiferromagnetic heterostructure. Science 381, 181-186 (2023).
  23. [23]Wang, N. et al. Quantum metric-induced nonlinear transport in a topological antiferromagnet. Nature 621, 487-492 (2023).DOI
  24. [24]Wang, Y. et al. Towards the quantized anomalous Hall effect in AlOₓ-capped MnBi₂Te₄. Nat Commun 16, 1727 (2025).DOI
  25. [25]Liu, C. et al. Magnetic-field-induced robust zero Hall plateau state in MnBi₂Te₄ Chern insulator. Nat Commun 12, 4647 (2021).
  26. [26]Ovchinnikov, D. et al. Intertwined Topological and Magnetic Orders in Atomically Thin Chern Insulator MnBi₂Te₄. Nano Lett 21, 2544-2550 (2021).DOI
  27. [27]Li, Y. et al. Fabrication-induced even-odd discrepancy of magnetotransport in few-layer MnBi₂Te₄. Nat Commun 15, 3399 (2024).DOI
  28. [28]Yang, S. et al. Odd-Even Layer-Number Effect and Layer-Dependent Magnetic Phase Diagrams in MnBi₂Te₄. Phys Rev X 11, 011003 (2021).DOI
  29. [29]Chen, B. et al. Even-Odd Layer-Dependent Exchange Bias Effect in MnBi₂Te₄ Chern Insulator Devices. Nano Lett 24, 8320-8326 (2024).DOI
  30. [30]Li, G. M. et al. Observation of Ultrastrong Coupling between Substrate and the Magnetic Topological Insulator MnBi₂Te₄. Nano Lett 22, 3856-3864 (2022).DOI
  31. [31]Chen, X. et al. Comparative Raman study of well-protected atomically thin MnBi₂Te₄ and Bi₂Te₃. Physical Review B 112, 085408 (2025).
  32. [32]Huang, Z., Du, M.-H., Yan, J. & Wu, W. Native defects in antiferromagnetic topological insulator MnBi₂Te₄. Phys Rev Mater 4, 121202 (2020).
  33. [33]Garnica, M. et al. Native point defects and their implications for the Dirac point gap at MnBi₂Te₄(0001). npj Quantum Mater 7, 7 (2022).arxiv.org/abs/2109.01615
  34. [34]Liu, M. et al. Visualizing the interplay of Dirac mass gap and magnetism at nanoscale in intrinsic magnetic topological insulators. Proceedings of the National Academy of Sciences 119, e2207681119 (2022).DOI
  35. [35]Lüpke, F. et al. Defect-induced displacement of topological surface state in quantum magnet MnBi₂Te₄. arXiv preprint arXiv:2507.05161 (2025).DOI
  36. [36]Islam, F. et al. Role of Magnetic Defects in Tuning Ground States of Magnetic Topological Insulators. Adv Mater 35, 2209951 (2023).DOI
  37. [37]Lai, Y., Ke, L., Yan, J., McDonald, R. D. & McQueeney, R. J. Defect-driven ferrimagnetism and hidden magnetization in MnBi₂Te₄ Physical Review B 103, 184429 (2021).
  38. [38]Yan, J. Q. et al. Crystal growth and magnetic structure of MnBi₂Te₄. Phys Rev Mater 3, 064202 (2019).DOI
  39. [39]Yang, S. & Ye, Y. Intrinsic and defect-related magnetism of MnBi₂Te₄(Bi₂Te₃)ₙ family from the Bulk to two-dimensional limit. 2D Mater 12, 012003 (2025).DOI
  40. [40]Hu, C., Qian, T. & Ni, N. Recent progress in MnBi₂ₙTe₃ₙ₊₁ intrinsic magnetic topological insulators: crystal growth, magnetism and chemical disorder. Natl Sci Rev 11 (2023).DOI
  41. [41]Zhang, S. et al. Experimental Observation of the Gate-Controlled Reversal of the Anomalous Hall Effect in the Intrinsic Magnetic Topological Insulator MnBi₂Te₄ Device. Nano Lett 20, 709-714 (2020).DOI
  42. [42]Sass, P. M., Kim, J., Vanderbilt, D., Yan, J. & Wu, W. Robust A-Type Order and Spin-Flop Transition on the Surface of the Antiferromagnetic Topological Insulator MnBi₂Te₄. Phys Rev Lett 125, 037201 (2020).DOI
  43. [43]Sass, P. M. et al. Magnetic Imaging of Domain Walls in the Antiferromagnetic Topological Insulator MnBi₂Te₄. Nano Lett 20, 2609-2614 (2020).DOI
  44. [44]Yang, S. et al. Defect-Assisted Domain Nucleation Drives Unique Exchange-Bias Phenomena in MnBi₂Te₄. Phys Rev X 14, 041024 (2024).arxiv.org/abs/2405.02549
  45. [45]Sun, Q.-C. et al. Magnetic domains and domain wall pinning in atomically thin CrBr₃ revealed by nanoscale imaging. Nat Commun 12, 1989 (2021).
  46. [46]Wang, Y. et al. Strain-Sensitive Magnetization Reversal of a van der Waals Magnet. Adv Mater 32, 2004533 (2020).DOI
  47. [47]Cenker, J. et al. Reversible strain-induced magnetic phase transition in a van der Waals magnet. Nat Nanotechnol 17, 256-261 (2022).DOI
  48. [48]Guo, Z. et al. Quantized Topological States and Parity Anomaly in Intrinsic Quantum Anomalous Hall Insulator MnBi₂Te₄. arXiv preprint arXiv:2507.03342 (2025).arxiv.org/abs/2507.03342

Paper details

vlnc:W_dwm7u3cy
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents