Visualizing superconductivity in a doped Weyl semimetal with broken inversion symmetry

The Weyl semimetal $\mathrm{Mo}{\mathrm{Te}}_2$ offers a rare opportunity to study the interplay between Weyl physics and superconductivity. Recent studies have found that Se substitution can boost the superconductivity up to 1.5 K, but suppresses the $T_d$ structure phase that is essential for the emergence of the Weyl state. A microscopic understanding of the possible coexistence of enhanced superconductivity and the $T_d$ phase has not been established so far. Here, we use scanning tunneling microscopy to study an optimally doped superconductor $\mathrm{Mo}{\mathrm{Te}}_{1.85}{\mathrm{Se}}_{0.15}$ with bulk $T_c\sim 1.5\mathrm{K}$. By means of quasiparticle interference imaging, we identify the existence of a low-temperature $T_d$ phase with broken inversion symmetry where superconductivity globally coexists. Furthermore, we find that the superconducting coherence length, extracted from both the upper critical field and the decay of density of states near a vortex, is much larger than the characteristic length scale of the existing chemical disorder. Our findings of robust superconductivity arising from a Weyl semimetal normal phase in $\mathrm{Mo}{\mathrm{Te}}_{1.85}{\mathrm{Se}}_{0.15}$ make it a promising candidate for realizing topological superconductivity.

Figure 1

(a) Crystal structure of $1T^{\prime }$ and $T_d$ phases of $\mathrm{Mo}{\mathrm{Te}}_2$. The solid dark lines denote one unit cell. (b) An illustration of the phase diagram of $\mathrm{Mo}{\mathrm{Te}}_{2–x}{\mathrm{Se}}_x$. The dark gray and pink stars denote the two doping levels studied in this work ($x=0$ and 0.15). (c) Resistivity as a function of temperature between 100 and 350 K for the pristine (dark gray) and Se-doped samples (pink), respectively. The anomaly at 250 K shows a $1T^{\prime }$ to $T_d$ phase transition. The onset of the superconducting transition in $\mathrm{Mo}{\mathrm{Te}}_{1.85}{\mathrm{Se}}_{0.15}$ occurs around 3 K as shown in the inset. (d) Typical topographic images taken on the (001) cleaved surface at 0.3 K $(V_S=100\mathrm{mV},I_t=50\mathrm{pA})$. The inset shows an atomically resolved image (10 mV, 50 pA) and the FFT of (d), where the dashed red box indicates the first Brillouin zone. The topographic corrugation has a typical length scale of ∼20 nm (Fig. S5 [35]). (e) $dI/dV$ spectra taken on the pristine (dark gray) and Se-doped (pink) samples $(V_S=–500\mathrm{mV},I_t=0.5\mathrm{nA})$. The dashed lines and arrows mark the features in the density of states that are related to the top of two bulk bands [33, 34]. (f) Raman shifts of $\mathrm{Mo}{\mathrm{Te}}_{1.85}{\mathrm{Se}}_{0.15}$ at 77 and 295 K. (g) Normalized Raman spectra $I$(77 K)/$I$(295 K). The yellow and purple lines denote the Raman shifts for the $T_d$ and $T^{\prime }$ phases of $\mathrm{Mo}{\mathrm{Te}}_2$, respectively.

Figure 2

(a) Topographic image $\left(V_S=10\mathrm{mV},I_t=50\mathrm{pA}\right)$. The white line indicates the trace on which the spectra shown in (b) are obtained. (b) $dI/dV$ spectra taken at 0.3 K showing the homogeneous superconductivity in the doped sample. (c,d) Temperature dependence of the tunneling spectra and superconducting gap. The inset of (d) shows a BCS fitting of the spectra at 0.3 K. (e,f) Magnetic field dependence of the tunneling spectra and the gap. The dark line indicates the expectation of Ginzburg-Landau theory.

Figure 3

(a) Differential conductance map measured with a magnetic field of 0.04 T perpendicular to the surface. (b) Spatially resolved tunneling spectra measured across a single vortex along the dashed yellow line (A-B) shown in (a). (c) Tunneling spectra as a function of distance (up to 95 nm) from the vortex core, azimuthally averaged for a range of ${90}^{\circ }$ in the bottom-right corner (marked by the white arcs). (d) Distance dependence of the zero-bias conductance (ZBC) from vortex core up to 200 nm and the fit with an exponential decay $g\left(r\right)=g_0+A\exp \left(-r/\xi \right)$. The data are obtained along the dashed yellow line shown in (a), from the center of the vortex to B point. Data are taken at 0.3 K.

Figure 4

Spectroscopic evidence for the broken inversion symmetry in $\mathrm{Mo}{\mathrm{Te}}_{1.85}{\mathrm{Se}}_{0.15}$. (a) The schematic for two inequivalent surfaces obtained by cleaving the opposite side of a bulk crystal. (b,c) Tunneling spectra obtained on two different surfaces. (d) Cartoon illustration of the constant energy contours and the related scattering vectors. (e) QPI patterns obtained on surface A at energies of −93, −12, 69, and 150 mV, respectively. (f) QPI patterns obtained on surface B at energies of −90, −10, 70, and 150 mV, respectively. The images are mirror symmetrized with respect to $K_x=0$. We have identified multiple scattering channels as follows: ${\mathrm{Q}}_1$, intraband scattering of the trivial surface states; ${\mathrm{Q}}_2$, scattering between surface states and the bulk electron pockets; ${\mathrm{Q}}_3$, scattering between the surface states and the projected bulk states at the $\overline{Y}$ points. The yellow circles mark the Bragg peaks along the $K_y$ direction, and the dashed ellipses highlight the differences in the QPI patterns. Data are taken at 0.3 K.

Figure 5

Comparison between QPI patterns of $\mathrm{Mo}{\mathrm{Te}}_2$ and $\mathrm{Mo}{\mathrm{Te}}_{1.85}{\mathrm{Se}}_{0.15}$, for both surfaces A and B. One can find a reasonably good analogy for QPI patterns of surfaces A and B by simply shifting the chemical potential between doped and nondoped samples. The yellow circles mark the Bragg peaks along the $K_y$ direction and the dashed ellipses highlight the differences in the QPI patterns. Here, the data of pristine $\mathrm{Mo}{\mathrm{Te}}_2$ are from our previous work [34].