Abstract
Tunnel junctions, an established platform for high resolution spectroscopy of superconductors, require defectfree insulating barriers; however, oxides, the most common barrier, can only grow on a limited selection of materials. We show that van der Waals tunnel barriers, fabricated by exfoliation and transfer of layered semiconductors, sustain stable currents with strong suppression of subgap tunneling. This allows us to measure the spectra of bulk (20 nm) and ultrathin (3 and 4layer) NbSe_{2} devices at 70 mK. These exhibit two distinct superconducting gaps, the larger of which decreases monotonically with thickness and critical temperature. The spectra are analyzed using a twoband model incorporating depairing. In the bulk, the smaller gap exhibits strong depairing in inplane magnetic fields, consistent with high outofplane Fermi velocity. In the fewlayer devices, the large gap exhibits negligible depairing, consistent with outofplane spin locking due to Ising spin–orbit coupling. In the 3layer device, the large gap persists beyond the Pauli limit.
Introduction
Superconductors of the transition metal dichalcogenide (TMD) family have seen a revival of interest subsequent to developments in device fabrication by mechanical exfoliation^{1,2,3,4,5,6,7}. The isolation of ultrathin NbSe_{2}^{1,2} has yielded indications of a Berezinskii–Kosterlitz–Thouless transition, which occurs in 2D superconductors. NbSe_{2}^{1}, gated MoS_{2}^{5,6}, and gated WS_{2}^{7} devices also remain superconducting in inplane magnetic fields well beyond the Pauli limit \(H_{\mathrm{p}} = {{\Delta /}}\left( {\sqrt g \mu _{\mathrm{B}}} \right)\)^{8,9}. (Here, Δ is the superconducting energy gap, g is the Landé gfactor and μ_{B} the Bohr magneton.) This is likely due to Ising spinorbit coupling (ISOC): The broken inversion symmetry of the monolayer TMD in the plane is expected to lead to the formation of Cooper pairs whose constituent spins are locked in the outofplane direction, in a singlet configuration. Interestingly, zeroresistance states have been observed in parallel magnetic fields exceeding the Pauli limit even in fewlayer devices, where inversion symmetry is recovered^{1}. This suggests that the interlayer coupling is not strong enough to overcome the outofplane spinlocking due to ISOC, perhaps in part due to the presence of spinlayer locking^{10}.
These previous studies^{1,2,3,4,5,6,7} used inplane electronic transport at high magnetic field and temperatures close to the critical temperature, T_{c}(H = 0), to determine the upper critical field H_{c2}, which depends on the magnitude of the spinorbit field H_{SO}. Tunneling spectroscopy can provide complementary information. For example, tunneling and other probes (see ref.^{11} and references therein) suggest that bulk NbSe_{2} is a 2band superconductor. The role of both bands in ultrathin devices, and their response to magnetic field, can be addressed by tunnel spectroscopy but not by electronic transport (which addresses the condensate) or heat conductivity measurements^{12}. The latter, specifically, would be shorted by the substrate. Tunneling can also probe the effect of magnetic field at the full temperature range and in fields ranging from zero to H_{c2}.
To carry out tunneling measurements on devices of variable thickness, at variable magnetic field conditions, it is necessary to develop a devicebased architecture suitable for integration with TMDs. Oxidebased tunnel barriers, such as those used since the days of Giaever^{13}, have now reached technological maturity; however, there is a limited number of oxides which form highquality insulating, nonmagnetic barriers, and they do not grow well on all surfaces. It is therefore of interest to explore alternatives based on van der Waals (vdW) materials^{14}, ultrathin layers of which can be precisely positioned on many surfaces^{15}. Indeed, such barriers have proven effective when integrated with graphene^{16,17,18} and appear to be promising candidates for integration with TMD superconductors^{19}.
In this work we study tunneling devices utilizing ultrathin TMDs as barriers. We find that these devices exhibit a hard gap with strong subgap conductance suppression. In the spectra of bulk NbSe_{2}, we clearly resolve the presence of two superconducting order parameters, which exhibit a distinct evolution upon application of inplane magnetic fields. Spectra of ultrathin NbSe_{2} are stable against the application of magnetic fields, even beyond the Pauli limit^{8,9}.
Results
Hard gap vanderWaals tunnel junctions
We fabricate NormalInsulatorSuperconductor (NIS) tunnel junctions with either MoS_{2} or WSe_{2}—both van der Waals (vdW) materials—as the insulating barrier. The barrier material is placed on top of 2HNbSe_{2} (hereafter NbSe_{2}), a vdW superconductor with bulk T_{c} ≈ 7.2 K, using the ‘dry transfer’ fabrication technique^{14,15}. Crucially, such heterostructures can comprise NbSe_{2} flakes of varying thicknesses, from bulk (≳6 layers) to fewlayer, often within a single device. Figure 1a shows a typical junction consisting of a 20 nmthick NbSe_{2} flake partially covered by a 4–5 layer thick MoS_{2} barrier (cf. Supplementary Note 1 and Supplementary Fig. 1). The junction has an area A = 1.6 μm^{2}, and we evaluate its transparency to be \({\cal T}\sim 10^{  8}\) (Supplementary Note 2 and Supplementary Fig. 2). Figure 1b shows the differential conductance G = dI/dV as a function of V obtained with the device, at T = 70 mK. This spectrum has two striking features: first, the very low subgap conductance (G_{0}R_{N} ≈ 1/500), which indicates all bands crossing the Fermi energy are fully gapped. Second, the intricate structure of the quasiparticle peak differs from a standard BCS densityofstates (DOS) by having a relatively low peak and a shoulder at lower energies. The latter feature can be clearly seen in the second derivative (Fig. 1c) where the slope separates into a double peak feature, similar to STS scans of bulk NbSe_{2}^{11,20}. Based on this, the flake can be considered bulk in terms of the zero field superconducting properties, and is hence referred to as the bulk sample.
Twoband model
Density functional theory calculations^{21}, and ARPES data^{22} show that five independent electronic bands cross the Fermi energy in NbSe_{2}. Of these, four are Nbderived bands with roughly cylindrical Fermi surfaces, centered at the Γ and K points. The fifth is derived from the Se p_{z} orbitals, which give rise to a small ellipsoid pocket around the Γ point. Ref. ^{11} uses a twoband model to fit NbSe_{2} tunneling data, which can be justified noting that the Se and Nbderived bands differ in the density of states and value of the electronphonon coupling parameter^{23}. We follow ref.^{11} in fitting our data using the same twoband model, which was developed in various forms by Suhl^{24}, Schopohl^{25}, and McMillan^{26} (below “SSM”).
The model entails a selfconsistent solution to the coupled equations for the energy dependent order parameters Δ_{ i }(E) in the two bands i:
\({{\Delta }}_i^0\) describes the intrinsic gap within each band i, that is generated by the electronphonon coupling and by the scattering rates of quasiparticles between the bands Γ_{ ij }. The extension of the twoband model to include AbrikosovGor’kov depairing^{27,28,29,30}—via the terms with \({{\Gamma }}_i^{{\mathrm{AG}}}\)—was done by Kaiser and Zuckermann^{31}. Here, depairing is due to magnetic field; thus, \({{\Gamma }}_i^{{\mathrm{AG}}}\) are set to 0 when no magnetic field is applied. The DOS of each band is then given by
where N_{ i }(E_{F}) is the DOS at the Fermi energy in the normal state in band i. The parameter α = 0.1, incorporates bandanisotropy. This anisotropy also affects the effective fit temperature (T = 0.44 K), which is higher than the sample temperature. Our fit indicates the presence of two independent order parameters. The larger, \({{\Delta }}_1^0 = 1.26 \pm 0.01{\kern 1pt} {\mathrm{meV}}\) can be determined with high fidelity. The smaller order parameter \(\Delta _2^0\), cannot be determined unambiguously, and could have any value between 0 and 0.3 meV. The suppressed intrinsic superconductivity in the second band is consistent with a band with small density of states and weak electron–phonon coupling. The Seband, having these properties^{23}, is thus a candidate for the 2nd band.
NbSe_{2} in inplane magnetic field: diffusive vs. ballistic regimes
We next investigate the evolution of the tunneling spectra vs. inplane magnetic field H_{} (model—Fig. 1d, data—Fig. 1e). The bulk sample appearing in Fig. 1 is thick enough to accommodate Meissner currents leading to orbital depairing. We track the peaks in d^{2}I/dV^{2}, which are associated with the two gaps. The high energy peak in d^{2}I/dV^{2} depends weakly on the field, whereas the low energy peak evolves nonlinearly towards lower energies. This trend persists up to H = 0.5 T which we interpret as H_{c1}. Bulk NbSe_{2} is regarded as a cleanlimit superconductor (coherence length ξ < l mean free path). This condition, however, might not hold equally for both bands, since the smaller gap is associated with a larger ξ. Indeed, the evolution of the low energy gap in H_{} does not agree with the relevant, Doppler shift model^{32,33,34}. Instead, the magnetic field evolution of its d^{2}I/dV^{2} features towards lower energies is wellreproduced by the diffusive KaiserZuckermann (KZ) model (Fig. 1d), assuming a depairing parameter \({{\Gamma }}_2^{{\mathrm{AG}}}\) quadratic in H_{}. For a thin sample of thickness d ≪ λ, \({{\Gamma }}_i^{{\mathrm{AG}}} = D_ie^2H_{}^2d^2{\mathrm{/}}6\hbar c^2\)^{35}, with D_{ i } the diffusion coefficient, d ≈ 20 nm and the penetration length λ ≈ 230 nm^{36}. We find \({{\Gamma }}_2^{{\mathrm{AG}}}\) = 640 μeV at 1 T, corresponding to D_{2} = 40 cm^{2} s^{−1}. The model therefore yields a very large value for the diffusion coefficient D_{2} = v_{F}l/3, indicating a large Fermi velocity. This lends further support to the identification of this feature with the Sederived band, where v_{F} is 4–5 times larger than in the Nbderived bands^{23}. In contrast, the high energy quasiparticle peak moves only slightly and linearly to higher energies at low H_{} (dashed line in Fig. 1e). From this, we estimate v_{F} ≈ 5 × 10^{4} m s^{−1}. This indicates that a comprehensive description of the field evolution of the full spectrum of bulk NbSe_{2} would require a model with arbitrary disorder bridging clean and diffusive limits.
Reduced gap in ultrathin NbSe_{2}
Our methods allow us to carry out a straightforward comparison of the tunneling spectra from devices with different NbSe_{2} thicknesses. Figure 2a shows differential conductance curves taken by tunneling into ultrathin NbSe_{2} flakes (3 layers, 4 layers), in comparison to the dI/dV of the bulk flake discussed above. The spectra of the thin devices are in good agreement with the SSM model (fits are shown in Supplementary Fig. 3, fit parameters are given in Supplementary Table 1). We extract the values of \({{\Delta }}_1^0\) from these fits, and separately evaluate T_{c} using the temperature dependence of the tunneling conductance (details in Supplementary Fig. 4). We find that \({{\Delta }}_1^0\) increases with T_{c} (Fig. 2b), however, deviating slightly from the BCS result Δ = 1.76k_{B}T_{c}. This deviation is plausible due to the multiband nature of superconductivity in NbSe_{2}. The T_{c} values measured here are lower than those reported elsewhere^{1,37}, likely due to higher disorder in our sample^{38,39,40}. Note that, based on ref.^{37}, it would seem that the T_{c} dependence on number of layers seen by all works to date is not due to strain or other substrate effects.
Ultrathin NbSe_{2} in magnetic field
We now turn to the response of the ultrathin devices to inplane magnetic fields. Figure 2 shows the tunneling spectra for the bulk device (Fig. 2a) and for the 4layer and 3layer devices (Fig. 2b and c, respectively). The KZ model, for all 3 devices, is shown in Fig. 2d–f, respectively. In the thin devices, unlike the bulk device, we find that the spectrum changes very little up to 3.5 T (which is the maximal field where a compensation coil keeps a zero perpendicular field). In both 3 and 4layer devices, there is a small reduction in the height of the quasiparticle peak; in the 3layer device, the low energy spectrum exhibits a more intricate evolution (discussed below). Thinner samples have shorter scattering lengths and longer coherence lengths, and should thus be closer to the diffusive limit. Indeed, the KZ model fits our data well, and we use it to quantify the reduction of the peak height of the 3 and 4layer devices (Fig. 2e, f), finding that the depairing term \({{\Gamma }}_1^{{\mathrm{AG}}} \approx 0.5\) μeV at 1 T (all model parameters are given in Supplementary Table 2).
Since orbital depairing is quadratic in sample thickness, we expect it to be diminished in the 3 and 4layer devices, allowing us to probe the spindependent interaction. The interaction of the spin with magnetic field should lead to Zeeman splitting of the spectrum, which we do not see. This could be due to two mechanisms. First, spin–orbit scattering can effectively randomize the spin, giving rise to a depairing parameter given by \({{\Gamma }}^{{\mathrm{AG}}} = \tau _{{\mathrm{SO}}}e^2\hbar H_{}^2{\mathrm{/}}2m^2\),where τ_{SO} is the spin orbit scattering time^{35}. Second, ISOC can align the spins in the outofplane direction with an effective field, H_{SO}, and the depairing term \({{\Gamma }}^{{\mathrm{AG}}}\sim 2\mu _{\mathrm{B}}H_{}^2{\mathrm{/}}H_{{\mathrm{SO}}}\)^{1,5}. The first scenario can be ruled out, since it yields τ_{SO} < 50 fs, shorter than the scattering time^{5}. The ISOC case is more likely, and the depairing energy of 0.5 μeV at 1 T (for the 3L device) yields H_{SO} ≈ 200 T. Using H_{p} = 4.9 T (extracted by setting Δ_{1} = 0.4 meV) we can estimate \(H_{{\mathrm{c2}}} = \sqrt {H_{{\mathrm{SO}}}H_{\mathrm{p}}}\) ≈ 30 T, consistent with transport experiments^{1}. We note that since the orbital term is not entirely suppressed, and we cannot estimate its contribution to the depairing, the estimate for H_{SO} is a lower bound. Further details concerning the possible interpretation of the depairing term are given in Supplementary Note 3 and Supplementary Table 2.
The stability of the larger gap can even be demonstrated above the Pauli limit, H_{p} = 4.9 T for the 3layer device. In Fig. 3c we present the density of states of this sample above the Pauli limit by applying an inplane field of 6.4 T. At this field, our measurement system did not allow us to compensate for angle misalignment leading to small component of perpendicular field (≈0.2 T) and possible vortex penetration. Nevertheless, the size of the gap remains unchanged. This lends further support to ISOC as the mechanism protecting NbSe_{2} superconductivity at high parallel fields.
Anomalous secondary gap
The subgap spectrum of the 3layer device appears to exhibit a secondary, wellformed small gap of Δ_{2} = 50 μV, which is suppressed at H_{} = 1.2 T. As we show in Fig. 3c, f, the KZ model reproduces this data remarkably well. Here, too, the depairing term is quadratic in H_{}, with \({{\Gamma }}_2^{{\mathrm{AG}}}\) ≈ 13 μeV at 1 T. This value is too big to be interpreted in terms of orbital depairing. The observed depairing could alternatively be associated with spinorbitdriven spinflip scattering, with τ_{SO} ≈ 1.5 ps. This depairing shows that unlike the band with larger gap, this band is not protected by ISOC. This difference can be explained by associating the smaller order parameter with the Sederived band. The outer gap, appearing immune to depairing, would then be associated with the Nbderived Kband. We note that this interpretation leaves open the question of the role of the Nbderived Γband. Addressing this question will require further spectral studies—in particular, of monolayer NbSe_{2} where the Sederived band does not cross the Fermi energy.
Discussion
Our results show that TMD semiconductors transferred on top of NbSe_{2} form stable tunnel barriers with a hard gap. We show that the SSM model, modified to include diffusive depairing, successfully reproduces the tunneling spectra in both the bulk and the ultrathin limits, in the presence of inplane magnetic fields. This allows us to probe the effect of the spin and orbital degrees of freedom on the spectra, thereby differentiating between the responses of the different bands to the field. The large gaps, in the 3 and 4layer devices, are remarkably stable to depairing by the inplane field, exhibiting very small depairing energies (<1 μeV), which place a tight cap on the spindepairing observed on this band, lending support to ISOC as the mechanism behind this stability. We suggest that our technique can be generalized to work with many other material systems, such as organic (super)conductors and other fragile systems which have hitherto not been investigated using tunneling spectroscopy.
Methods
Device fabrication and measurement
The vdW tunnel junctions were fabricated using the dry transfer technique^{41}, carried out in a glovebox (nitrogen atmosphere). NbSe_{2} flakes were cleaved using the scotch tape method, exfoliated on commercially available Gelfilm from Gelpack, and subsequently transferred to a SiO_{2} substrate. MoS_{2} and WSe_{2} flakes were similarly exfoliated and thin flakes suitable for the formation of tunnel barriers were selected based on optical transparency. The barrier flake was then transferred and positioned on top of the NbSe_{2} flake at room temperature. Ti/Au contacts and tunnel electrodes were fabricated using standard ebeam techniques. Prior to the evaporation of the ohmic contacts the sample was ion milled for 15 s. No such treatment was done with the evaporation of the tunnel electrodes. All transport measurements were done in a ^{3}He–^{4}He dilution refrigerator with a base temperature of 70 mK. The AC excitation voltage was modulated at 17 Hz; its amplitude was 15 μV at all temperatures for the bulk device and 10 μV for the fewlayer devices. Measurement circuit details are provided in Supplementary Fig. 5.
Data availability
The data that support the findings of this study are available from the corresponding author upon request.
Additional information
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
 1.
Xi, X. et al. Ising pairing in superconducting NbSe_{2} atomic layers. Nat. Phys. 12, 139–143 (2015).
 2.
Tsen, A. W. et al. Nature of the quantum metal in a twodimensional crystalline superconductor. Nat. Phys. 12, 208–212 (2015).
 3.
Ugeda, M. M. et al. Characterization of collective ground states in singlelayer NbSe_{2}. Nat. Phys. 12, 92–97 (2015).
 4.
Staley, N. E. et al. Electric field effect on superconductivity in atomically thin flakes of NbSe_{2}. Phys. Rev. B 80, 1–6 (2009).
 5.
Lu, J. M. et al. Evidence for twodimensional Ising superconductivity in gated MoS_{2}. Science 350, 1353–1357 (2015).
 6.
Saito, Y. et al. Superconductivity protected by spin–valley locking in iongated MoS_{2}. Nat. Phys. 12, 144–149 (2015).
 7.
Lu, J. et al. A full superconducting dome of strong Ising protection in gated monolayer WS_{2}. https://arxiv.org/abs/1703.06369 (2017).
 8.
Clogston, A. M. Upper limit for the critical field in hard superconductors. Phys. Rev. Lett. 9, 266–267 (1962).
 9.
Chandrasekhar, B. S. A note on the maximum critical field of highfield superconductors. Appl. Phys. Lett. 1, 7–8 (1962).
 10.
Jones, A. M. et al. Spinlayer locking effects in optical orientation of exciton spin in bilayer WSe_{2}. Nat. Phys. 10, 130–134 (2014).
 11.
Noat, Y. et al. Quasiparticle spectra of 2HNbSe_{2}: twoband superconductivity and the role of tunneling selectivity. Phys. Rev. B 92, 134510–134518 (2015).
 12.
Boaknin, E. et al. Heat conduction in the vortex state of NbSe_{2}: evidence for multiband superconductivity. Phys. Rev. Lett. 90, 117003 (2003).
 13.
Giaever, I. Energy gap in superconductors measured by electron tunneling. Phys. Rev. Lett. 5, 147–148 (1960).
 14.
Geim, A. K. & Grigorieva, I. V. Van der Waals heterostructures. Nature 499, 419–425 (2013).
 15.
Dean, C. R. et al. Boron nitride substrates for highquality graphene electronics. Nat. Nanotechnol. 5, 722–726 (2010).
 16.
Amet, F. et al. Tunneling spectroscopy of grapheneboronnitride heterostructures. Phys. Rev. B 85, 073405 (2012).
 17.
Britnell, L. et al. Electron tunneling through ultrathin boron nitride crystalline barriers. Nano. Lett. 12, 1707–1710 (2012).
 18.
Chandni, U., Watanabe, K., Taniguchi, T. & Eisenstein, J. P. Signatures of phonon and defectassisted tunneling in planar metalhexagonal boron nitridegraphene junctions. Nano. Lett. 16, 7982–7987 (2016).
 19.
Island, J. O., Steele, G. A., van der Zant, H. S. & CastellanosGomez, A. Thickness dependent interlayer transport in vertical MoS_{2} Josephson junctions. 2D Mater. 3, 031002 (2016).
 20.
Guillamón, I., Suderow, H., Guinea, F. & Vieira, S. Intrinsic atomicscale modulations of the superconducting gap of 2HNbSe_{2}. Phys. Rev. B 77, 134505 (2008).
 21.
Johannes, M. D., Mazin, I. I. & Howells, C. A. Fermisurface nesting and the origin of the chargedensity wave in NbSe_{2}. Phys. Rev. B 73, 205102 (2006).
 22.
Yokoya, T. et al. Fermi surface sheetdependent superconductivity in 2HNbSe_{2}. Science 294, 2518–2520 (2001).
 23.
Kiss, T. et al. Chargeordermaximized momentumdependent superconductivity. Nat. Phys. 3, 720–725 (2007).
 24.
Suhl, H., Matthias, B. T. & Walker, L. R. BardeenCooperSchrieffer theory of superconductivity in the case of overlapping bands. Phys. Rev. Lett. 3, 552–554 (1959).
 25.
Schopohl, N. & Scharnberg, K. Tunneling density of states for the twoband model of superconductivity. Solid State Commun. 22, 371–374 (1977).
 26.
McMillan, W. L. Tunneling model of the superconducting proximity effect. Phys. Rev. 175, 537–542 (1968).
 27.
Abrikosov, A. A. & Gor’kov, L. P. Contribution to the theory of superconducting alloys with paramagnetic impurities. Sov. Phys. JETP 12, 1243–1253 (1961).
 28.
Maki, K. The behavior of superconducting thin films in the presence of magnetic fields and currents. Progress. Theor. Phys. 31, 731–741 (1964).
 29.
Levine, J. L. Density of states of a shortmeanfreepath superconductor in a magnetic field by electron tunneling. Phys. Rev. 155, 373–378 (1967).
 30.
Millstein, J. & Tinkham, M. Tunneling into superconducting films in a magnetic field. Phys. Rev. 158, 325–332 (1967).
 31.
Kaiser, A. B. & Zuckermann, M. J. McMillan model of the superconducting proximity effect for dilute magnetic alloys. Phys. Rev. B 1, 229–235 (1970).
 32.
Fulde, P. Tunneling phenomena in solids. Gapless Superconducting Tunneling—Theory, 427–442 (Springer, 1969).
 33.
Fridman, I., Kloc, C., Petrović, C. & Wei, J. Y. T. Lateral imaging of the superconducting vortex lattice using Dopplermodulated scanning tunneling microscopy. Appl. Phys. Lett. 99, 192505 (2011).
 34.
Kohen, A. et al. Probing the superfluid velocity with a superconducting tip: the doppler shift effect. Phys. Rev. Lett. 97, 027001 (2006).
 35.
Tinkham, M. Introduction to superconductivity (Courier Corporation, 1996).
 36.
Garoche, P., Veyssié, J., P, M. & P, M. Experimental inverstigation of superconductivity in 2HNbSe_{2} single crystal. Solid State Commun. 19, 455–460 (1976).
 37.
Frindt, R. F. Superconductivity in ultrathin NbSe_{2} layers. Phys. Rev. Lett. 5, 299–301 (1972).
 38.
Finkel’shtein, A. M. Superconducting transition temperature in amorphous films. JETP 45, 37–40 (1987).
 39.
Finkel’shtein, A. M. Suppression of superconductivity in homogeneously disordered systems. Phys. B 197, 636–648 (1994).
 40.
Goldman, A. M. & Marković, N. Superconductorinsulator transitions in the two dimensional limit. Phys. Today 51, 39–44 (1998).
 41.
CastellanosGomez, A. et al. Deterministic transfer of twodimensional materials by alldry viscoelastic stamping. 2D Mater. 1, 011002 (2014).
Acknowledgements
We thank P. Février and J. Gabelli for helpful discussions on tunnel barriers, and T. Cren for the same on NbSe_{2}. This work was funded by a MaimondesIsrael grant from the IsraeliFrench High Council for Scientific & Technological Research and by an ANR JCJC grant (SPINOES) from the French Agence Nationale de Recherche. H.S. acknowledges support by ERC2014STG Grant No. 637298 (TUNNEL) and Marie Curie CIG Grant No. PCIG12GA2012333620. T.D. is grateful to the Azrieli Foundation for an Azrieli Fellowship. F.M. has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie SkodowskaCurie Grant No. 659247. L.A. and M.K. are supported by the Israeli Science Foundation through Grant No. 1287/15.
Author information
Affiliations
The Racah Institute of Physics, The Hebrew University of Jerusalem, Jerusalem, 91904, Israel
 T. Dvir
 , L. Attias
 , M. Khodas
 & H. Steinberg
Laboratoire de Physique des Solides (CNRS UMR 8502), Bâtiment 510, Université ParisSud/Université ParisSaclay, 91405, Orsay, France
 F. Massee
 , M. Aprili
 & C. H. L. Quay
Authors
Search for T. Dvir in:
Search for F. Massee in:
Search for L. Attias in:
Search for M. Khodas in:
Search for M. Aprili in:
Search for C. H. L. Quay in:
Search for H. Steinberg in:
Contributions
T.D. fabricated the devices. C.Q.H.L., T.D., and M.A. performed the measurements. T.D., F.M, L.A., M.K., C.Q.H.L., M.A., and H.S. contributed to data analysis and the writing of the manuscript.
Competing interests
The authors declare no competing financial interests.
Corresponding author
Correspondence to H. Steinberg.
Electronic supplementary material
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Further reading

1.
Tunnelling spectroscopy of gateinduced superconductivity in MoS2
Nature Nanotechnology (2018)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.