Fabrication of a novel magnetic topological heterostructure and temperature evolution of its massive Dirac cone

Materials that possess nontrivial topology and magnetism is known to exhibit exotic quantum phenomena such as the quantum anomalous Hall effect. Here, we fabricate a novel magnetic topological heterostructure Mn4Bi2Te7/Bi2Te3 where multiple magnetic layers are inserted into the topmost quintuple layer of the original topological insulator Bi2Te3. A massive Dirac cone (DC) with a gap of 40–75 meV at 16 K is observed. By tracing the temperature evolution, this gap is shown to gradually decrease with increasing temperature and a blunt transition from a massive to a massless DC occurs around 200–250 K. Structural analysis shows that the samples also contain MnBi2Te4/Bi2Te3. Magnetic measurements show that there are two distinct Mn components in the system that corresponds to the two heterostructures; MnBi2Te4/Bi2Te3 is paramagnetic at 6 K while Mn4Bi2Te7/Bi2Te3 is ferromagnetic with a negative hysteresis (critical temperature ~20 K). This novel heterostructure is potentially important for future device applications.


Supplementary Note 1: Photon energy dependence
Supplementary Figure 1a shows the band dispersion image of the Mn, Te/Bi 2 Te 3 heterostructure sample 2 taken with different photon energies (hν =7. 5, 8, 8.5, 9, 9.5, 10, 14, 15, 17, and 21 eV). The Dirac cone is hν independent and can be safely regarded as surface 2 states. Supplementary Figure 1b shows the EDC spectra at theΓ point of the images shown in a and one can notice that the two peaks of the massive Dirac cone can be recognized for all the photon energies as indicated by the black lines, although there is variation in the peak intensity. At low energy, additional features outside the gapped Dirac cone appear.
The features near E F near theΓ point become strong at 21 eV and it is extremely difficult to see the gap in the EDC curve at theΓ point.

Supplementary Note 2: Spin-resolved ARPES
Supplementary Figure 2 shows the spin-resolved ARPES results for the Mn, Te/Bi 2 Te 3 heterostructure sample 3. The gap size in this sample is ∼ 40 meV (see the EDC in Supplementary Fig. 2a. The measurements were performed along theΓ −M (x in Supplementary Fig. 2b) direction and the spin-orientation along the y direction was measured as shown in Supplementary Fig. 2c for θ = −4 to +4 • . Supplementary Figure 2 d shows the results obtained in c overlapped on the band dispersion image. Red and blue markers represent the spin-up and down states that are antisymmetric with respect to theΓ point, respectively. On the other hand, brown and purple markers represent spin-up and down states that are not antisymmetric and likely show spin-polarization due to spin-dependent photoemission dipole matrix element effect [1,2]. The Dirac cone shows the expected antisymmetric spin-split structure. Supplementary Figure 3 shows the details of the quantitative analyses for peak determination to deduce the Dirac-cone gap size. First, a Shirley-type background was estimated from the raw spectra as shown in Supplementary Fig. 3a for the data taken at 16 K. After subtracting the background, the spectra were fitted multiple Gaussian peaks ( Supplementary Fig. 3b). We first fitted the experimental data with 3 peaks (including the 6 strong peak α just below E F ) for all the spectra (Supplementary Fig. 3c). By deducing the peak width of the fitted results, we found a jump between 140 and 200 K. The width of peak α increased significantly, whereas that for peak B decreased although it was monotonously increasing below 140 K (Supplementary Fig. 3e). Since the spectra features of the Dirac cone shown in Fig. 1f for temperatures higher than 200 K could also be assumed as a single peak, we also tried fitting using only 2 components, as shown in Supplementary Fig. 3d.
In this case, the peak width was larger than in the case when 3 components were used.
Especially the data at 200 K showed a single peak width of 64 meV, much larger than the other fitted results. On the other hand, for the data of 250 and 290 K, the results were fairly reasonable considering the temperature broadening. Since the temperature broadening effect alone cannot explain the fact that the peak width for the spectrum at 200 K is larger than those for higher temperature, we believe that a gradual phase transition is occurring at 200-250 K. The raw data in Fig. 1f shows that the peak becomes sharper at 250 and 290 K than those at 200 and 140 K. Such feature is also observed in sample 4 (Fig. 1g), reinforcing our statement. Taking all these facts into account, we assigned 3 peaks to the spectra below 200 K and deduced the gap size from the analysis, whereas only 2 peaks were considered for the spectra at 250 and 290 K, meaning the closing of the Dirac cone gap.
Although the transition at 200-250 K is dull, we believe that the gap of the massive Dirac cone at low temperature has closed at 290 K. Supplementary Figure 3f  Namely, we used two, one, and one peak(s) in the fitting for the data at 200 K, 250 K, and 290 K in the left panel respectively, while it was changed to one, two and two in the right panel.

Supplementary Note 4: Band dispersion image for heterostructure sample 4
Binding energy (eV)   to the theoretical prediction in Ref. [3]. As discussed in the main text, this may be due to the fact that the measurement temperature of 16 K is above the Curie temperature of the system as reported in Ref. [4]. In fact, Ref. [5] predicts that the Curie temperature of MnBi 2 Te 4 is 12 K from Monte Carlo simulations. For bulk MnBi 2 Te 4 , the Neel temperature is reported as 25 K [6]. However, this system is paramagnetic down to 6 K as shown in Fig. 4   Mn 4 Bi 2 Te 7 is 1:1). Sample N1 was determined as the single MnBi 2 Te 4 phase from LEED