Control over epitaxy and the role of the InAs/Al interface in hybrid two-dimensional electron gas systems

In situ synthesized semiconductor/superconductor hybrid structures became an important material platform in condensed matter physics. Their development enabled a plethora of novel quantum transport experiments with focus on Andreev and Majorana physics. The combination of InAs and Al has become the workhorse material and has been successfully implemented in the form of one-dimensional structures and two-dimensional electron gases. In contrast to the well-developed semiconductor parts of the hybrid materials, the direct effect of the crystal nanotexture of Al films on the electron transport still remains unclear. This is mainly due to the complex epitaxial relation between Al and the semiconductor. Here, we present characterization of Al thin films grown on shallow InAs two-dimensional electron gas systems by molecular beam epitaxy. Using a growth approach based on an intentional roughening of the epitaxial interface, we demonstrate growth of grain-boundary-free Al. We show that the implemented roughening does not negatively impact either the electron mobility of the two-dimensional electron gas or the basic superconducting properties of the proximitized system. This is an important step in understanding the role of properties of the InAs/Al interface in hybrid devices. Ultimately, our results provide a growth approach to achieve a high-degree of epitaxy in lattice-mismatched materials.

Figure 1

(a) ADF-STEM image of the shallow InAs 2DEG structure. A schematic of the structural design is shown in the left panel, with the zoom-in inset showing the detailed structure of the quantum well region. The yellow dashed line highlights locations of the images used for the lattice constant evaluation. The 575 nm and 725 nm markers along the line highlight regions where a change in the ADF contrast indicates strain relaxation. (b) HAADF-STEM image of a single misfit dislocation. (c) ADF-STEM image of an extended twin defect, with a higher magnification image in the inset (scale bar is 1 nm). (d) The deviation of the measured lattice spacing as a function of the distance from the substrate (extracted from HAADF-STEM images taken in the center of each grown layer) from the bulk lattice constant for the individual layers. The lattice spacing is evaluated from the spacings of both the (001) and (110) planes. BB stands for bottom barrier, QW for quantum well, and TB for top barrier, which are plotted in the green shaded area. The local minima (green circles) correspond to locations of local strain relaxation highlighted in (a). (e) Same dependency, but for the lattice spacing of each individual layer in the structure. The blue line shows the bulk lattice constant of each layer.

Figure 2

(a) HAADF-STEM image of the interface between an Al:$A$(112) grain and the semiconductor, showing Al(112) to SE(110) lateral matching. Both the image and the model (top right panel) show an undisturbed epitaxial matching. The red line highlights coherent matching over a step on the SE surface. (b) HAADF-STEM image of a grain of the same orientation (Al:$A$), but rotated by ${90}^{\circ }$ around the [111] axis, i.e., showing Al(110) to SE(110) matching (Al:$A$(110)). The Bragg-filtered image (bottom right panel) shows that the lattice mismatch along this direction is relaxed by networks of misfit dislocations with $\mathrm{Al}\times \mathrm{SE} 5\times 3$ periodicity. Local contrast was normalized in all the STEM images.

Figure 3

(a) HAADF-STEM image of the interface between an Al:$B$(001) grain and the semiconductor, showing Al(001) to SE(110) lateral matching. The Bragg-filtered image (bottom right panel) shows that the strain along this direction is relaxed by the typical formation of misfit dislocations (highlighted by red circles). (b) HAADF-STEM image of a grain of the same orientation (Al:$B$), but rotated by ${90}^{\circ }$ around the [011] axis, i.e., showing Al(110) to SE(110) matching (Al:$B$(110)). Both the image and the model (top right panel) show significant mismatch. The Bragg-filtered image (bottom right) shows that the lattice mismatch along this direction is relaxed by networks of misfit dislocations with $\mathrm{Al}\times \mathrm{SE} 3\times 2$ periodicity. Local contrast was normalized in both STEM images.

Figure 4

(a) ADF-STEM images of high grain density (top) and a single orientation (bottom) Al thin films. The red arrows highlight regions with strong changes in contrast. (b) A higher magnification and contrast normalized HAADF image showing structure of one of the possible grain boundaries, highlighted by red arrows in (a). (c) Contrast normalized HAADF-STEM image of an Al film deposited on a roughened surface. (d) Schematic of the possible orientations of different Al films on a heavily anisotropically roughened surface. (e) HAADF-STEM images of two perpendicular projections ([110] and [$1\overline{1}0$]) of the top interface of the semiconductor, demonstrating the anisotropy and scale of the intentional roughening. Local contrast normalization was performed for images in (b) and (a).

Figure 5

(a) BF-STEM image of Al films deposited on a surface capped with 4 MLs of GaAs showing a change of contrast in the center. The HAADF image in the inset shows the orientation of the Al film. (b) ADF-STEM image, showing the origin of the contrast in (a). The fast Fourier transform in the inset shows that the Al crystal is mirrored over the boundary. (c) Contrast normalized HAADF image of the crystal around the boundary. (d) BF-STEM image of Al films deposited on a surface capped with 5 MLs of GaAs. The HAADF-STEM image in the inset shows the orientation of the Al film. (e) ADF-STEM image, showing an abrupt change in contrast in the same film as in the bottom panel of Fig. 4. (f) HAADF-STEM image of the same area. The fast Fourier transform does not show a detectable change in the crystal structure.

Figure 6

(a) Hall resistance (green) and longitudinal resistance (blue) as a function of magnetic field for the InAs 2DEG on standard (full line) and intentionally roughened (dashed line) SE surface, after removal of the epitaxial Al. (b) Electron mobility and carrier density as a function of top barrier thickness (for standard structures 2 MLs of GaAs cap and after Al wet etching).

Figure 7

(a) False-colored SEM image of the measured Josephson junction, highlighting the top-gate (gold) and the epitaxial Al leads of the JJ (blue). The terminals for the bias current $I$ = $I_{\mathrm{SD}} + I_{\mathrm{AC}}$, measured four-probe voltage $V_{\mathrm{DC}}$ and $V_{\mathrm{AC}}$ and gate voltage $V_{\mathrm{G}}$ are highlighted. (b) Differential resistance $R$ = $V_{\mathrm{AC}}/I_{\mathrm{AC}}$ of the JJ as a function of source-drain current $I_{\mathrm{SD}}$ and gate voltage $V_{\mathrm{G}}$. (c) Dependence of $R$ of the JJ as a function of $I_{\mathrm{SD}}$ and out-of-plane magnetic field $B_{\perp }$. (d) Dependence of $V_{\mathrm{DC}}$ (red, left axis) and differential resistance $R$ (blue, right axis) on $I_{\mathrm{SD}}$, measured at a base temperature of 18 mK. The black dashed line shows a linear fit to the I-V trace above $\sim$ 0.4 mV. The individual MAR peaks in $R$ are labeled by $n$. They are related to the measured voltage via the I-V trace (gray dotted lines). The black arrows point to the position of the switching and excess current, respectively. (e) Temperature dependence of $R$ as function of bias current $I_{\mathrm{SD}}$. For clarity, each temperature trace is offset by 50 $\mathrm{\Omega }$. Extracted $V_{\mathrm{DC}}$ positions of the MAR peaks in $R$ for $n$ = 1, 2, 3 as a function of temperature are shown in the inset. The dashed lines correspond to fits to the BCS theory for the nominal Al superconducting gap.