Superconductivity in type-II Weyl-semimetal WTe₂ induced by a normal metal contact

Topological materials attract a lot of attention in modern condensed matter physics. This interest stems from intriguing fundamental properties and great potential for practical applications. The especially interesting class of topological materials is topological superconductors, promising to revolutionize quantum computing due to the inherent error protection.¹ Topological superconductivity could be obtained by inducing superconductivity in a topologically non-trivial system. There are several theoretical predictions of different topological superconducting states in Dirac and Weyl-semimetal based systems, including Fulde–Ferrell–Larkin–Ovchinnikov superconductors,^2–4 the time-reversal invariant topological superconductor,⁵ chiral non-Abelian Majorana fermions,⁶ and flatband superconductivity.⁷ 

$WTe2$ is a layered transition-metal dichalcogenide with rich topological properties. As a bulk material, it is a type-II Weyl semimetal with bulk Weyl nodes connected by Fermi-arc surface states.^8,9 Recently, it has been predicted to be a higher-order topological insulator with one-dimensional hinge states,¹⁰ and experimental evidence of these states has been obtained.^11–13 In a single layer form, $WTe2$ is a two-dimensional topological insulator with helical edge states.^14,15 In addition to all these topological phases, $WTe2$ has a tendency of becoming superconducting under different conditions: under pressure,^16,17 electron doping,¹⁸ or electrostatic gating.^19,20 The combination of these properties makes $WTe2$ a particularly promising candidate for topological superconductivity.

In this paper, we demonstrate the emergence of superconductivity at the interface between the normal metal palladium and few-layer thick $WTe2$⁠. Studying the transport properties in magnetic field and at different temperatures, we deduce the main parameters characterizing the superconducting state including the critical temperature, the coherence length, and the London penetration depth. These parameters correspond to a low Fermi velocity and a high density of states at the Fermi level. This hints to a possible origin of superconductivity due to the formation of flatbands. Moreover, the measured in-plane critical field exceeds the Pauli limit, suggesting non-trivial superconducting pairing. The coexistence of the observed superconductivity with topological states in $WTe2$ makes it a promising platform for studying topological superconductivity and applications for quantum computing.

The single crystals of $WTe2$ were grown with a flux growth method.²¹ We obtained few-layer thick $WTe2$ flakes by mechanically exfoliating single crystals with an adhesive tape on an oxidized Si substrate with a 295 nm $SiO2$ layer. To avoid oxidation of $WTe2$⁠, the exfoliation has been carried out in a glovebox with low oxygen content. We selected few-layer thick (5–12 single layers) stripe shaped flakes. Suitable flakes have been identified with an optical contrast method²² and were picked up and transferred using the polycarbonate assisted pick-up technique²³ on the device chip that already contained prepatterned contacts. The contacts were defined before using standard e-beam lithography and metal deposition of 3 nm titanium and 12 nm palladium. In the final stack, $WTe2$ is protected from oxidation by an hBN layer that covers the $WTe2$⁠. All the measurements were performed in a dilution refrigerator with a base temperature of 60 mK. Similar superconducting properties have been observed in multiple devices, and data presented in the paper were collected from 3 samples.

Figure 1(a) shows an optical image of an encapsulated $WTe2$ crystal with a contact pattern that resembles a standard Hall-bar configuration. Note that the visible Pd contacts are at the bottom, followed by a few-layer $WTe2$ crystal with a rectangular shape and a high-aspect ratio oriented vertically, followed by an hBN layer that has the weakest contrast in the image. The drawn electrical schematics correspond to the measurement of the longitudinal resistance $Rxx$ given by $Vxx/I$⁠.

Figure 1(b) displays $Rxx$ as a function of perpendicular magnetic field $B⊥$⁠. At 4 K, the resistance shows a non-saturating magnetoresistance characteristic for $WTe2$⁠.²⁴ The small thickness (7 layers) of our $WTe2$ crystal results in a relatively small magnetoresistance.²⁵ Another evidence of the high quality of our samples is the presence of Shubnikov–de Haas oscillations at low temperature. The frequency of the oscillations $f1/B∼100$ T corresponds to an electron density $n2D=e/(πℏf1/B)∼5⋅1012$ cm$−2$ and a Fermi wavevector $kF=πn2D∼0.4$ nm$−1$ (here, the two electron pockets of $WTe2$ are taken into account). The oscillation visibility at around 5 T suggests a mobility of at least 2000 cm$2$ V$−1$ s$−1$⁠, which yields an electron mean free path of $lmfp=kFℏμ/e∼50$ nm.

At low temperature, additional features develop in $Rxx(B⊥)$⁠: at zero field, the resistance goes to zero, and in small fields, it has an intermediate state between zero and high-temperature values. The asymmetry in $Rxx(B)$ is defined by the sweep direction and rate; thus, it is likely connected with heating during the magnetic field sweep. The intermediate resistance state in Fig. 1(b) is a result of the formation of a superconducting state in $WTe2$ above the Pd leads. Furthermore, these superconducting regions could be connected by the Josephson effect, as illustrated in Fig. 1(c), leading to a zero longitudinal resistance. The zero resistance state appears only for smaller distances between the contacts, excluding intrinsic superconductivity in our $WTe2$ samples. This explanation is further supported by $Rxx(T)$ dependence in Fig. 1(d). With decreasing temperature, the first superconducting transition takes place in the range of 1.05–1.2 K, followed by the gradual developing of the Josephson effect at lower temperature achieving zero resistance below 350 mK.

To understand the properties of the superconducting state, we studied the evolution of $Rxx(B⊥)$ with increasing temperature, as shown in Fig. 2(a). Upon temperature increase, both transitions in the resistance are shifting toward zero field. The zero resistance state connected to the Josephson coupling disappears first above 0.75 K, and the second transition connected to the suppression of superconductivity by magnetic field $Bc2$ persists up to 1.1 K. We define $Bc2(T)$ as the magnetic field where the $Rxx(B⊥)$ crosses the fixed resistance value $Rxx=45$ $Ω$⁠, which approximately corresponds to half of the resistance step. Figure 2(b) shows the extracted dependence of the critical magnetic field as a function of temperature $T$⁠. The $Bc2(T)$ dependence is linear as expected for a 2D superconductor,

where $Φ0$ is the magnetic flux quantum, $ξGL$ is the Ginzburg–Landau coherence length at zero temperature, and $Tc$ is the critical temperature at zero magnetic field. Fitting the experimental data with Eq. (1), we obtain $Tc∼1.2$ K and relatively short $ξGL∼14$ nm.

Disorder can cause the reduction of the coherence length, but we do not think this is the case in our samples since we have found $lmfp>ξ$ and non-saturating magnetoresistance. In the clean limit at low temperatures, the Ginzburg–Landau coherence length is similar to the Bardeen–Cooper–Schrieffer (BCS) coherence length $ξGL∼ξ$⁠. Knowing the coherence length and the critical temperature, we can estimate the Fermi velocity $vF=ξπΔ/ℏ$⁠, where we take for $Δ(Tc)$ the BCS relation $Δ∼1.76kBTc$⁠, yielding $vF∼1.2⋅104$ ms$−1$⁠. The obtained small value of Fermi velocity could suggest superconductivity due to the formation of flatbands.²⁶ 

We further investigate the superconducting properties by looking at the $Rxx$ dependence on the in-plane magnetic field $B∥$⁠, as shown in Fig. 3(a). Compared with the perpendicular field, both changes in the resistance have shifted to higher magnetic fields. We extracted the critical field values as a function of temperature and plotted them in Fig. 3(b). In this case, $Bc2(T)$ follows the known empirical law for superconductors $Bc2(T)=Bc2(0)[1−(T/Tc)2]$⁠,²⁷ as evident from the very good agreement between measured points and the fit. Both fits of the critical field as a function of $B⊥$ and $B∥$ converge to the same temperature $Tc∼1.2$ K.

A notable feature of the parallel critical field is its large value, which exceeds the Pauli paramagnetic limit $BP$⁠. The latter is given by $BP∼1.76kBTc2/gμB∼1.86Tc∼2.3$ T. This expression is based on the BCS theory for weak-coupling superconductors and a free electron $g$-factor of $g=2$⁠.²⁸ This effect has also been observed in a gated monolayer^19,20 and doped bulk $WTe2$¹⁸ and ultrathin films of other materials.^29,30 Several mechanisms could be responsible for superconductivity exceeding the Pauli limit, including Ising-type superconductivity²⁸ or a diminishing of the effective $g$-factor due to strong spin–orbit coupling.³¹ Further studies are required to resolve this.

The London penetration depth $λL$ is another important characteristic of a superconductor. While RF-measurements are a common way of measuring the penetration depth,³² it can also be estimated by measuring the critical current of a Josephson junction in a magnetic field $B⊥$⁠. A Josephson junction placed in a perpendicular magnetic field demonstrates an oscillating critical current. One period of the oscillations corresponds to the magnetic flux quantum $Φ0$ through the effective area of the junction $Seff=WLeff=W(L+2λ)$⁠, where $W$ and $L$ are the junction’s width and length, respectively, and $λ$ is the magnetic field penetration depth;²⁷ see Fig. 4(a). For a bulk superconductor $λ=λL$⁠, but for a thin film superconductor with thickness $d$⁠, the penetration depth is a function of the thickness $λ(d)=λLcoth⁡(d/λL)$⁠.³² In the limit of small thickness $d≪λL$⁠, the previous expression is equal to the Pearl’s penetration depth $λP=λL2/d$⁠.

Figure 4(b) demonstrates several examples of $Ic(B⊥)$ dependencies for Josephson junctions in $WTe2$ where the superconducting regions on top of Pd play the role of superconducting contacts. These dependencies have a SQUID-like character due to hinge states¹¹ with a rapidly decaying Fraunhofer contribution due to the Fermi-arc surface states^33,34 or the bulk conductivity. The SQUID-like oscillations with many visible periods allow one to determine the period with a high precision. For junctions 1 and 2, with $L=1$ $μ$m and $W=4.3$ $μ$m, we obtain a period of $△B=0.27$ mT. This period corresponds to $Leff=1.77$ $μ$m and a penetration depth of $λ=380$ nm. For junction 3 (⁠$L=500$ nm, $W=4.2$ $μ$m), we obtain $△B=0.41$ mT, yielding $λ=350$ nm. The obtained penetration depth is much larger than the thickness of the $WTe2$ flakes $d∼7$ nm (approximately 10-layer thick) so that the extracted penetration depth is given by Pearl’s limit $λ=λP=λL2/d$⁠. Using this expression, we estimate the London penetration depth to be $λL∼50$ nm. The ratio between the London penetration depth and the coherence length $κ=λL/ξ$ is $κ∼3>1/2$⁠, suggesting type-II superconductivity.²⁷ 

The obtained London penetration depth is comparable to typical values for metals and is surprisingly small considering the semimetalic nature of $WTe2$⁠. This additionally speaks against the presence of disorder in our samples since the penetration depth is expected to be higher in dirty superconductors.²⁷ An estimate of the superconducting electron density yields a quite high value $ns=m/μ0λL2e2∼3⋅1021$ cm$−3$⁠, where $m∼0.3me$ is the effective mass of the electrons in $WTe2$⁠.³⁵ This value is higher than the typical carrier densities in $WTe2$ $n∼1019$ cm$−3$³⁶ and corresponds to a density per single layer of $ns1L∼2⋅1014$ cm$−2$⁠, which is an order of magnitude higher than the electron density in monolayer $WTe2$ with gate induced superconductivity^19,20 but comparable to the predicted optimal charge carrier density.³⁷ Furthermore, a large superconducting carrier density implies a high density of states at the Fermi level $g(EF)∼ns/2Δ∼8⋅1024$ cm$−3$ eV$−1$⁠, which is a signature of flatbands.

The emergence of superconductivity at the interface of two non-superconducting materials is quite surprising despite the fact that it has been observed previously in different Weyl and Dirac semimetals.^38–42 While the underlying mechanism for the superconductivity is unclear, in our case of the $WTe2$/Pd interface, several material specific reasons for the superconductivity could be proposed. First, the structural change at the interface could lead to the superconductivity similar to the pressure induced superconductivity in $WTe2$⁠.^16,17 Second, electron doping from palladium⁴³ could create a superconducting state similar to what was seen in monolayer^19,20 or bulk doped $WTe2$⁠.¹⁸ The latter seems to be the more probable explanation since the in-plane critical field exceeds the Pauli limit, which has been observed in doped $WTe2$^18–20 but not in the pressure induced superconductivity.^16,17 Another possibility is interdiffusion of Pd and Te with a formation of superconducting $PdTe2$⁠, which has been recently reported in samples with Pd deposited on $(Bi1−xSbx)2Te3$⁠.⁴⁴ We think that this mechanism is unlikely in our case since $WTe2$ and Pd are merely placed in contact by stacking and the samples were never heated above 180 $°$C.

Even more intriguing is the possibility of the flatband superconductivity,^45,46 as suggested by the small Fermi velocity and high density of states at the Fermi level. Flatbands are ubiquitous in van der Waals (vdW) heterostructures. For example, high carrier density combined with a low Fermi velocity has been observed close to van-Hove singularities in the band structure of superlattices formed in hBN-encapsulated graphene.⁴⁷ Also, the presence of the flatbands is known to stimulate superconductivity.^26,48,49

Establishing the presence of the flatband superconductivity at the $WTe2$/Pd interface and understanding the reasons for it will require further experiments, but some explanations could be outlined already. Flatband superconductivity can be formed as a result of a topological phase transition due to strain at the interface.^7,46 Furthermore, flatband superconductivity has been observed in vdW systems with a Moiré pattern.²⁶ This possibility is feasible since the mismatch between the lattice constant in Pd (0.389 nm⁵⁰) and the a-axis lattice constant in $WTe2$ (0.349 nm¹⁷) is only about 10%. However, we deem this scenario unlikely since Moiré patterns strongly depend on the mutual orientation of the lattices, while we observe superconductivity in multiple samples without any intentional alignment of the lattices.

An alternative explanation for the experimental data could be a multiband superconductivity in our samples. In this situation, a sublinear dependence of $Bc2⊥(T)$ and an exceeding Pauli limit by $Bc2∥$ could be expected in the dirty limit of the superconductivity,⁵¹ which seems not to be the case in our samples.

We demonstrate the emergence of superconductivity at the interface between type-II Weyl-semimetal $WTe2$ and normal metal palladium. Studying the transport properties in a magnetic field and at different temperatures, we deduce the key parameters that characterize the superconducting state, including the critical temperature $Tc$⁠, the coherence length $ξ$⁠, and the London penetration depth $λL$⁠. The combined set of parameters hint to a possible origin of superconductivity being due to the formation of flatbands. Moreover, the measured in-plane critical field exceeds the Pauli limit, suggesting non-trivial superconducting pairing. The coexistence of superconductivity with topological states makes $WTe2$ a promising platform for topological superconductivity and applications for quantum computing.

We thank A. Baumgartner for helpful discussions. A.K. was supported by the Georg H. Endress foundation. This project has received further funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (Grant Agreement No. 787414 TopSupra), by the Swiss National Science Foundation through the National Centre of Competence in Research Quantum Science and Technology (QSIT), and by the Swiss Nanoscience Institute (SNI). K.W. and T.T. acknowledge support from the Elemental Strategy Initiative conducted by MEXT, Japan and the CREST (JPMJCR15F3), JST. D.G.M. and J.Y. acknowledge support from the U.S. Department of Energy (U.S. DOE), Office of Science—Basic Energy Sciences (BES), Materials Sciences and Engineering Division. D.G.M. acknowledges support from the Gordon and Betty Moore Foundation’s EPiQS Initiative (Grant No. GBMF9069).

The data that support the findings of this study are openly available in a numerical form in Zenodo at https://doi.org/10.5281/zenodo.3934680, Ref. 52.