Scanning tunneling spectroscopy investigations of superconducting-doped topological insulators: Experimental pitfalls and results

Recently, the doping of topological insulators has attracted significant interest as a potential route towards topological superconductivity. Because many experimental techniques lack sufficient surface sensitivity, however, definite proof of the coexistence of topological surface states and surface superconductivity is still outstanding. Here we report on highly surface sensitive scanning tunneling microscopy and spectroscopy experiments performed on Tl-doped ${\mathrm{Bi}}_2{\mathrm{Te}}_3$, a three-dimensional topological insulator which becomes superconducting in the bulk at $T_{\mathrm{C}}=2.3$ K. Landau level spectroscopy as well as quasiparticle interference mapping clearly demonstrated the presence of a topological surface state with a Dirac point energy $E_{\mathrm{D}}=-(118\pm 1)$ meV and a Dirac velocity $v_{\mathrm{D}}=(4.7\pm 0.1)\times {10}^5$ m/s. Tunneling spectra often show a superconducting gap, but temperature- and field-dependent measurements show that both $T_{\mathrm{C}}$ and ${\mu }_0{H}_{\mathrm{C}}$ strongly deviate from the corresponding bulk values. Furthermore, in spite of a critical field value which clearly points to type-II superconductivity, no Abrikosov lattice could be observed. Experiments performed on normal-metallic Ag(111) prove that the gapped spectrum is caused only by superconducting tips, probably caused by a gentle crash with the sample surface during approach. Nearly identical results were found for the intrinsically $n$-type compound Nb-doped ${\mathrm{Bi}}_2{\mathrm{Se}}_3$. Our results suggest that the superconductivity in superconducting-doped V-VI topological insulators does not extend to the surface where the topological surface state is located.

Figure 1

(a) Topographic image of cleaved ${\mathrm{Tl}}_{0.6}{\mathrm{Bi}}_2{\mathrm{Te}}_3$. Defects characteristic for $p$-doped ${\mathrm{Bi}}_2{\mathrm{Te}}_3$ can be recognized. No signs of Tl intercalation are visible. The inset shows an atomic resolution image. (b) Differential conductance $\left(dI/dU\right)$ spectrum of ${\mathrm{Tl}}_{0.6}{\mathrm{Bi}}_2{\mathrm{Te}}_3$ recorded on the clean surface at a lateral distance of more than 3 nm from a defect. The Dirac point energy $E_{\mathrm{D}}$, valence band maximum $E_{\mathrm{VB}}$, and conduction band minimum $E_{\mathrm{CB}}$ are marked with blue dashed lines. Scan/stabilization parameters are $T=1.7$ K and $U=-0.3\mathrm{V},I=50\mathrm{pA}$ in the main panel of (a), $U=-0.2\mathrm{V},I=50\mathrm{pA}$ in the inset of (a), and $U_{\mathrm{set}}=-0.4\mathrm{V},I_{\mathrm{set}}=100\mathrm{pA},U_{\text{mod}}=5\mathrm{mV}$ in (b).

Figure 2

(a) STS data of ${\mathrm{Tl}}_{0.6}{\mathrm{Bi}}_2{\mathrm{Te}}_3$ measured at various magnetic fields at $T=1.7$ K. (b) Landau level energies plotted versus $k_n$, confirming the linear dependence expected for a TSS. (c) Schematic equipotential surface (black), in-plane spin polarization (blue and red arrows), and potential scattering vectors $({\mathit{q}}_{\mathit{1}}/{\mathit{q}}_{\mathit{2}})$ for a TI close to the Dirac point (top panel) and in the warped energy regime (bottom). (d) QPI map of ${\mathrm{Tl}}_{0.6}{\mathrm{Bi}}_2{\mathrm{Te}}_3$, with its Fourier transformation in the inset $(T=4.8$ K). No backscattering can be recognized. Scan parameters are (a) $U=-150\mathrm{mV},I=600\mathrm{pA},U_{\text{mod}}=1\mathrm{mV}$ and (d) $U=400\mathrm{mV},I=50\mathrm{pA},U_{\text{mod}}=10$ mV.

Figure 3

(a) Normalized $dI/dU$ spectra of ${\mathrm{Tl}}_{0.6}{\mathrm{Bi}}_2{\mathrm{Te}}_3$. The spectrum taken with tip 1 shows a superconducting gap at $T=1.5$ K (black line). However, the same tip exhibits a very similar gap on Ag(111) (red line), indicating that a superconducting cluster was accidentally picked up, presumably due to a gentle collision with the sample surface. After careful tip handling, even at $T=0.32$ K no gap can be found (tip 2; blue line). (b) Temperature-dependent $dI/dU$ spectra as measured with a superconducting tip. The gap vanishes at 5 K, i.e., a temperature significantly higher than $T_{\mathrm{C}}$. (c) Field-dependent STS data measured with the same tip at $T=1.5$ K. The superconducting gap can be recognized up to 4.5 T. Parameters are $U=-5\mathrm{mV},I=50–200\mathrm{pA},U_{\text{mod}}=0.1\mathrm{mV}$.

Figure 4

(a) Temperature-dependent magnetic susceptibility measured after STM/STS measurements. No significant changes in critical temperature and the superconducting volume fraction were found. (b) Overview spectroscopy data taken with a W tip on ${\mathrm{Tl}}_{0.6}{\mathrm{Bi}}_2{\mathrm{Te}}_3$ for different Cu doping amounts. With an increasing amount of Cu on the sample surface the bulk band gap shifts to lower energies.

Figure 5

(a) High-resolution STS data for Nb-intercalated ${\mathrm{Bi}}_2{\mathrm{Se}}_3$ acquired with a superconducting tip (1; black line) and a normal-metallic W tip (2; blue line) at a nominal temperature $T=1.5$ K. While for tip 1 a pronounced gap is visible, no superconductivity can be recognized with tip 2. (b) Temperature-dependent normalized $dI/dU$ signal taken on ${\mathrm{Nb}}_x{\mathrm{Bi}}_2{\mathrm{Se}}_3$ with a superconducting tip. (c) Magnetic-field-dependent STS data taken with a superconducting tip. The gap can be recognized up to about 3 T. For higher fields additional variations caused by the fine structure of the Landau levels are visible. Stabilization parameters are $U=-5\mathrm{mV},I=200\mathrm{pA},U_{\text{mod}}=0.1\mathrm{mV}$.