Andreev Reflection in an $s$-Type Superconductor Proximized 3D Topological Insulator

We investigate transport and shot noise in lateral normal-metal–3D topological-insulator–superconductor contacts, where the 3D topological insulator (TI) is based on Bi. In the normal state, the devices are in the elastic diffusive transport regime, as demonstrated by a nearly universal value of the shot noise Fano factor $F_N\approx 1/3$ in magnetic field and in a reference normal-metal contact. In the absence of magnetic field, we identify the Andreev reflection (AR) regime, which gives rise to the effective charge doubling in shot noise measurements. Surprisingly, the Fano factor $F_{\mathrm{AR}}\approx 0.22\pm 0.02$ is considerably reduced in the AR regime compared to $F_N$, in contrast to previous AR experiments in normal metals and semiconductors. We suggest that this effect is related to a finite thermal conduction of the proximized, superconducting TI owing to a residual density of states at low energies.

Figure 1

Magnetotransport data and sample layout. Left: Linear response resistances as functions of perpendicular magnetic field for the normal-metal–TI–normal-metal ($N$-TI-$N$) device ($n$, symbols) and normal-metal–TI–superconductor ($N$-TI-$S$) devices (s1–s4, lines) measured at $T=0.6\mathrm{K}$. For the devices s1 and s3, the data were multiplied by factors of 2 and 0.5, respectively. Right: A scanning electron micrograph of the inner part of one of our $N$-TI-$S$ devices.

Figure 2

Shot noise and transport in the reference $N$-TI-$N$ device. Left axis: Bias voltage dependence of the measured shot noise spectral density at $B=0\mathrm{T}$ (circles) and $B=1.2\mathrm{T}$ (crosses). The fit using Eq. (1) with $F=0.33$ is shown by the dashed line. Right axis: Differential resistance $R=(dV/dI)$ as a function of the bias voltage across the device at $B=0$ (circles) and $B=1.2\mathrm{T}$ (crosses).

Figure 3

Transport and shot noise in $N$-TI-$S$ devices. (a) Typical bias voltage dependence of the differential resistance measured in a $N$-TI-$S$ device s2 at $B=0\mathrm{T}$ and $B=0.3\mathrm{T}$. (b) Measured shot noise spectral density (symbols) as a function of the bias voltage across the device in zero magnetic field. The fits using Eq. (1) with $q=e$, $0.38\le F\le 0.48$ and $q=2e$, $0.19\le F\le 0.24$ are shown by solid and dashed lines, respectively. The data sets are multiplied by 1, 1.5, and 2, respectively, from top to bottom. (c) $S_I(V)$ (symbols) in perpendicular magnetic field. The fits with $q=e$, $0.31\le F\le 0.36$ and $q=2e$, $0.15\le F\le 0.18$ are shown by solid and dashed lines, respectively. Different traces correspond to devices s1, s2, and s3 as marked in the figure.

Figure 4

Finite bias $2e\to e$ transition and proximity gap. (a) Body: Measured noise temperature vs $V$ in the AR regime (thick solid line) and in the normal transport regime (thick dashed line offset by 0.32 K) in device s3, $T\approx 0.53\mathrm{K}$. Corresponding $2e$ and $e$ fits are shown by thin dotted lines. The $2e\to e$ transition at the edge of the proximity gap $|V|\approx 0.35\mathrm{mV}$ is marked by arrows. Note that thick solid and dashed lines closely coincide up to $|V|\approx 5\mathrm{mV}$, which is not shown in figure. Insets: Similar $2e\to e$ transitions in the devices s1 and s2. (b) Proximity effect in a lateral $N$-TI-$S$ junction. A superconducting gap ${\mathrm{\Delta }}_{\mathrm{STI}}^{\prime }$ opens in a strongly proximized TI (STI) beneath the Nb film. Lateral interface between the STI and normal TI relevant for noise and transport in the AR regime is indicated by an arrow.