Critical analysis of proximity-induced magnetism in ${\mathrm{MnTe}/\mathrm{Bi}}_2{\mathrm{Te}}_3$ heterostructures

An elegant approach to overcome the intrinsic limitations of magnetically doped topological insulators is to bring a topological insulator in direct contact with a magnetic material. The aspiration is to realize the quantum anomalous Hall effect at high temperatures where the symmetry-breaking magnetic field is provided by a proximity-induced magnetization at the interface. Hence, a detailed understanding of the interfacial magnetism in such heterostructures is crucial, yet its distinction from structural and magnetic background effects is a rather nontrivial task. Here, we combine several magnetic characterization techniques to investigate the magnetic ordering in $\mathrm{MnTe}/{\mathrm{Bi}}_2{\mathrm{Te}}_3$ heterostructures. A magnetization profile of the layer stack is obtained using depth-sensitive polarized neutron reflectometry. The magnetic constituents are characterized in more detail using element-sensitive magnetic x-ray spectroscopy. Magnetotransport measurements provide additional information about the magnetic transitions. We find that the supposedly antiferromagnetic MnTe layer does not exhibit an x-ray magnetic linear dichroic signal, raising doubt that it is in its antiferromagnetic state. Instead, Mn seems to penetrate into the surface region of the ${\mathrm{Bi}}_2{\mathrm{Te}}_3$ layer. Furthermore, the interface between MnTe and ${\mathrm{Bi}}_2{\mathrm{Te}}_3$ is not abrupt, but extending over $\sim 2.2$ nm. These conditions are the likely reason that we do not observe proximity-induced magnetization at the interface. Our findings illustrate the importance of not solely relying on one single technique as proof for proximity-induced magnetism at interfaces. We demonstrate that a holistic, multitechnique approach is essential to gain a more complete picture of the magnetic structure in which the interface is embedded.

Figure 1

Plots of the Hall resistance $R_{\mathrm{xy}}$ in different temperature regimes. The hysteresis is due to the AHE effect and obtained after subtracting a linear background due to the ordinary Hall effect. In (a), the low temperature data is shown, which is governed up to 10 K by a two-step behavior. The insert shows the geometry of the resistance measurement. The higher temperature data shown in (b) and (c) is dominated by an increasingly square hysteresis loop, which rounds off between 150 and 200 K and finally vanishes between 200 and 250 K. The inset in (b) shows the areas enclosed by the hysteresis loops as a function of temperature.

Figure 2

PNR (nuclear structural information) of the $\mathrm{CrSe}/\mathrm{MnTe}/{\mathrm{Bi}}_2{\mathrm{Te}}_3$ trilayer stack as fitted by three different models. (a), (c), (e) Fresnel reflectivity and best fits for 3 and 300 K, respectively. (b), (d), (f) Full-scale nSLD profile. The gray regions represent the 95% Bayesian uncertainty intervals. Model 1 is the simplest possible case, with magnetism only in the CrSe buffer layer. In model 2, magnetism exists both in the CrSe and MnTe layer. In model 3, all layers are assumed to be magnetic.

Figure 3

PNR of the $\mathrm{CrSe}/\mathrm{MnTe}/{\mathrm{Bi}}_2{\mathrm{Te}}_3$ trilayer stack as fitted by three different models. The full nSLD profile, the Fresnel reflectivity curves, and a short description of the models can be found in Fig. 2. Plots of the spin asymmetry (SA) and corresponding magnetic scattering length density profiles (mSLD) are shown in (a), (c), (e) and (b), (d), (f), respectively. The upper panels show the 3 K and the lower panels the 300 K data sets and mSLD. The SA data sets were cofitted with all structural/nuclear parameters shared between the two temperatures. The gray regions represent the 95% Bayesian uncertainty intervals.

Figure 4

(a), (c) XAS and (b), (d) XMCD at the Mn and Cr $L_{2,3}$ edges measured in surface-sensitive total electron yield (TEY) mode. The measurements were carried out at 1.5 and 250 K, but only the XAS at 1.5 K for both positive (pc) and negative circular (nc) helicity is shown. A magnetic field of 6 T was applied along the incident x-ray direction, perpendicular to the sample plane, to saturate the magnetization of the sample. The geometry of the setup and available detectors is shown in the inset in (c).

Figure 5

(a), (c) XAS and (b), (d) XMCD at the Mn and Cr $L_{2,3}$ edges measured in fluorescence yield (FY). The inset in (c) shows the out-of-plane XMCD hysteresis at the Cr $L_3$ edge at 1.5 and at 250 K.

Figure 6

XMLD at the Mn $L_{2,3}$ edges in FY mode. (a) XAS for linear horizontal (lh) and linear vertical (lv) polarization at 1.5 K, and (b) XMLD spectra. The incidence angle of the x rays was perpendicular to the sample surface to probe the in-plane component of the antiferromagnetic vector. No magnetic field was applied.