Edge channel transport in the InAs/GaSb topological insulating phase

Transport in InAs/GaSb heterostructures with different InAs layer thicknesses is studied using a six-terminal Hall bar geometry with a 2-$\mu$m edge channel length. For a sample with a 12-nm-thick InAs layer, nonlocal resistance measurements with various current/voltage contact configurations reveal that the transport is dominated by edge channels with a negligible bulk contribution. Systematic nonlocal measurements allow us to extract the resistance of individual edge channels, revealing sharp resistance fluctuations indicative of inelastic scattering. Our results show that the InAs/GaSb system can be tailored to have conducting edge channels while keeping a gap in the bulk region and provide a way of studying two-dimensional topological insulators even when quantized transport is absent.

Figure 1

Schematic illustrations of the samples: (a) cross-sectional view of the layer structure, (b) top view of the Hall bar, and (c) band profile along the growth direction. Three samples with different InAs layer thicknesses ($w$ $=$ 10, 12, and 14 nm) were examined.

Figure 2

(a) Longitudinal resistance $R_{14,23}$ ($=V_{23}/I_{14}$) at 0.25 K as a function of $V_{\mathrm{FG}}$ for samples with different InAs thicknesses $w$. (b) Temperature dependence of $R_{xx}$ peaks plotted vs inverse temperature (1/$T$). Solid lines indicate fits to the high-temperature section of the data, which provide a crude estimate of the gap, $\Delta =0.5$ and 0.2 meV, for $w=10$ and 12 nm, respectively. (c)–(e) Schematic illustrations of energy band diagrams. The band overlap increases from (c) to (e). In (e), the region near the anticrossing is shown in an enlarged view (right).

Figure 3

(a) Effective circuit model describing edge transport in the topological insulating phase. Left and right panels represent current/voltage contact configurations for the nonlocal measurements in (b) and (c), respectively. See main text for details. Nonlocal resistances of the $w=12$ nm sample at 0.25 K as a function of $V_{\mathrm{FG}}$: (b) $R_{16,23}$ and $R_{16,34}$ and (c) $R_{65,23}$ and $R_{65,34}$. (d) Ratio between nonlocal resistances measured with adjacent voltage-probe pairs (2–3 and 3–4) at 0.25 K. The two traces show results for different current paths (1–6 and 6–5). (e) Magnified view of the data in the dashed rectangle in (d). (f) Similar measurements to those in (d) at 4.3 K.

Figure 4

Results of nonlocal measurements on the $w=12$ nm sample for various current/voltage contact configurations. In the gap region (gray area) corresponding to the the longitudinal $R_{\mathrm{xx}}$ peak, the ratios coincide with each other for all contact configurations.

Figure 5

(a) Resistances of individual edge channels deduced from nonlocal resistance measurements at 0.25 K. (b) Comparison of longitudinal resistance $R_{14,65}$ reconstructed from the individual channel resistances shown in (a) with that measured directly.

Figure 6

Nonlocal resistance $R_{32,45}$ as a function of $V_{\mathrm{FG}}$ for the $w$ $=$ 12 nm sample. The resistance was monitored at each $V_{\mathrm{FG}}$ for 60 s with a 1-s interval. The red curve represents the average taken at each $V_{\mathrm{FG}}$, and the vertical bars indicate the amplitude of the time-dependent fluctuation at each$V_{\mathrm{FG}}$. The dashed rectangle area in (a) is enlarged in (b).