Superfluid stiffness of a KTaO3

Nature Communications volume 13, Article number: 4625 (2022) Cite this article

3770 Accesses

7 Citations

90 Altmetric

Metrics details

After almost twenty years of intense work on the celebrated LaAlO3/SrTiO3system, the recent discovery of a superconducting two-dimensional electron gas (2-DEG) in (111)-oriented KTaO3-based heterostructures injects new momentum to the field of oxides interface. However, while both interfaces share common properties, experiments also suggest important differences between the two systems. Here, we report gate tunable superconductivity in 2-DEGs generated at the surface of a (111)-oriented KTaO3 crystal by the simple sputtering of a thin Al layer. We extract the superfluid stiffness of the 2-DEGs and show that its temperature dependence is consistent with a node-less superconducting order parameter having a gap value larger than expected within a simple BCS weak-coupling limit model. The superconducting transition follows the Berezinskii-Kosterlitz-Thouless scenario, which was not reported on SrTiO3-based interfaces. Our finding offers innovative perspectives for fundamental science but also for device applications in a variety of fields such as spin-orbitronics and topological electronics.

Potassium tantalate KTaO3 is a band insulator with a 3.6 eV gap that retains a cubic perovskite structure down to the lowest temperature1. Like strontium titanate (SrTiO3), it is a quantum paraelectric material on the verge of a ferroelectric instability that is characterized by a large permittivity at low temperature (ϵr ≃ 5000)1,2. Both materials can be turned into metal by electron doping, through oxygen vacancies, for example. Because of their common properties, it was suggested that superconductivity should also occur in doped KTaO3. However, while superconductivity was discovered more than half a century ago in bulk SrTiO33, all the attempts to induce bulk superconductivity in KTaO3have failed so far4. Using ionic gating, Ueno et al. could generate a superconducting 2-DEG at the surface of (001)-KTaO3, albeit at a very low temperature (Tc ≃ 40 mK)5. Later explorations of KTaO3 2-DEGs did not evidence any superconductivity until the beginning of the year 2021, when two articles reported the discovery of superconducting 2-DEG formed at the interface between (111)-KTaO3 and insulating overlayers of LaAlO3or EuO6,7. An empiric increase of Tc with electron density was proposed with a maximum value of 2.2 K for doping of ≈ 1.04 × 1014e− × cm−2 6, which is almost one order of magnitude higher than in the LaAlO3/SrTiO3 interface8. An electric field effect control of the Tc was also demonstrated in a Hall bar device7 and a dome-shaped superconducting phase diagram similar to that of SrTiO3-based interfaces was derived9,10. Following this discovery, the (110)-oriented KTaO3 interface was also found to be superconducting with Tc ≃ 1 K11. It was recently proposed that the soft transverse optical mode, involved in the quantum paraelectricity, could be responsible for electron pairing in KTaO3 interfaces. The coupling amplitude between this phonon mode and electrons is expected to be maximum in the (111) orientation and minimum in the (001) one, which would explain the hierarchy in Tc observed in these superconducting 2-DEGs12.

In conventional superconductors, well described by the Bardeen–Cooper–Schrieffer (BCS) theory, the superconducting transition is controlled by the breaking of Cooper pairs as the temperature exceeds the energy scale set by the superconducting gap. However, in two-dimensional superconductors, the superfluid stiffness, i.e., the energy associated with the phase rigidity of the superconducting condensate, can be comparable to the pairing energy, allowing for a Tc suppression driven by the loss of phase coherence. In this case, the transition is expected to belong to the Berezinskii–Kosterlitz–Thouless (BKT) universality class, where the transition is controlled by the unbinding of topological vortex-antivortex pairs13,14,15. Critical magnetic field measurements in (111)-KTaO3 2-DEGs, both in the perpendicular and in the parallel geometry, set an upper bound, d ≈ 5 nm, on the extension of the 2-DEG in the substrate6. This is lower than the superconducting coherence length, ξ ≈ 10–15 nm6, which confirms that the superconducting 2-DEG is within the 2D limit. In addition, the presence of disorder, which has been identified in this system6,7, is also expected to lower the superfluid rigidity and reinforce the role of phase fluctuations. Even though the measurements of the current-voltage characteristics in ref. 6 could be compatible with indirect signatures of a BKT transition, a direct measurement of the superfluid stiffness is required to properly address this issue16.

Here, we show that a 2-DEG can be generated at the surface of a (111)-oriented KTaO3 crystal simply by sputtering a very thin Al layer. The deposition of Al leads to the reduction of Ta ions as evidenced by X-ray photoelectron spectroscopy (XPS) and leads to the formation of an interfacial gate-tunable superconducting 2-DEG. We use resonant microwave transport to measure the complex conductivity of the 2-DEG and extract the temperature-dependent superfluid stiffness Js(T). Our results are consistent with a node-less superconducting order parameter in a rather strong-coupling regime (\({{\Delta }}(0)/{k}_{{{{{{{{\rm{B}}}}}}}}}{T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\) = 2.3). Taking into account the presence of disorder and finite-frequency effect, we show that the superconducting transition follows the Berezinskii–Kosterlitz–Thouless model, which was not observed on SrTiO3-based interfaces.

2-DEGs were generated by dc sputtering of a very thin Al layer on (111)-oriented KTaO3`substrates at a temperature between 550 and 600 °C. The preparation process is detailed in the Methods section. Prior to deposition, we measured the in-situ X-ray photoelectron spectra (XPS) of the Ta 4f valence state (Fig. 1a top) of the KTaO3 substrate. The spectra show the sole presence of Ta5+ states (4f5/2 and 4f3/2), indicating the expected stoichiometry of the substrate. The Ta 4f core levels were then measured after growing 1.8–1.9 nm of Al and transferring the sample in vacuum to the XPS setup. The bottom graph in Fig. 1a shows the Ta 4f core level spectra with additional peaks corresponding to reduced states of Ta i.e., Ta4+ and Ta2+. The deeper and lighter shades of same-colored peaks correspond to 4f5/2 and 4f3/2 split peaks. The reduction of Ta5+ to Ta4+ upon Al deposition indicates the formation of oxygen vacancies at the surface of KTaO3, which in turn suggests the formation of a 2-DEG. The Ta2+ signal may be due to the presence of a small amount of Ta in the AlOx layer (akin to the situation in AlOx/STO17) or reflects the presence of small clusters of oxygen vacancies around some Ta ions, reducing their apparent valence state. We monitored the Al oxidation state by measuring the Al 2p core levels after exposure of the sample to the atmosphere, which evidenced full oxidation of the Al layer into AlOx. Thus, as in the AlOx/SrTiO3 system, the 2-DEG is formed through a redox process by which oxygens are transferred from the KTaO3 substrate to the Al overlayer17,18,19.

a X-ray photoelectron spectra near the 4f core level binding energy of Ta for a KTaO3substrate prior to deposition (top) and after deposition of 1.8 nm of Al (bottom). The fitted peaks for Ta5+, Ta4+, and Ta2+ are shown in cyan, green and violet colors, respectively. The deeper and lighter shades of the same colors represent the 4f5/2 and 4f2/2 valence states of the respective peaks. The data and sum fit envelope are shown in red circles and black lines. b (top) HAADF scanning transmission electron microscopy image at the cross-section of KTaO3 (111) and AlOx interface. The KTaO3 is observed along the [112] direction. (down) EELS maps (Al-L2,3, Ta-O2,3, and K-L2,3 edges) showing the presence of Al on top of the interface without any interdiffusion but with some limited diffusion of Ta and K inside the AlOx layer.

The structure of the AlOx/KTaO3 (111) interface has been imaged by scanning transmission electron microscopy (STEM). Figure 1b depicts the high-angle annular dark field (HAADF) - STEM image in cross-section. The electron energy loss spectroscopy (EELS) indicates that a small amount of K and Ta diffuse into the AlOx layer. In contrast, the Al signal decays very rapidly in KTaO3, indicating no Al diffusion into KTaO3. Our fabrication method based on the sputtering of a thin Al film has already been successfully implemented to generate 2-DEGs on (001)-oriented KTaO3 substrates showing a fivefold enhancement of the Rashba spin–orbit coupling as compared to SrTiO320. In the present work, four samples, labeled A, B, C, and D, have been investigated by transport measurement at low temperature in a dilution refrigerator (see Methods section for fabrication parameters).

Figure 2a shows the resistance vs temperature curve of sample A on a wide temperature range revealing a superconducting transition at Tc ≃ 0.9 K. In Fig. 2b, we plot the Tc as a function of the 2D carrier density, n2D for the different samples studied and compare their values with those extracted from ref. 6. Our results confirm the trend observed in the literature (Tc increases with the carrier density) and demonstrate that our growth method, while being much easier to implement than the molecular beam epitaxy of a rare-earth element such as Eu or the pulsed laser deposition of a complex oxide, is able to produce good quality samples with similar Tc. The resistance vs temperature curves of sample B measured for different values of a magnetic field applied perpendicularly to the sample plane are shown in Fig. 2c. The temperature dependence of the critical magnetic field is consistent with a Landau-Ginsburg model near Tc, \({\mu }_{0}{H}_{c}(T)=\frac{{{{\Phi }}}_{0}}{2\pi {\xi }_{\parallel }^{2}(T)}\), taking into account an in-plane superconducting coherence length \({\xi }_{\parallel }={\xi }_{\parallel }(0){(1-\frac{T}{{T}_{{{{{{{{\rm{c}}}}}}}}}})}^{-\frac{1}{2}}\). We found ξ∥(T = 0) ≈ 27 nm, which is comparable with the value reported in ref. 6.

a Sheet resistance of sample A as a function of temperature (log scale) showing a superconducting transition at Tc ≃ 0.9 K. b Summary of the superconducting critical temperature as a function of the carrier density for all samples studied in this work compared with results from ref. 6. c Sheet resistivity of sample B as a function of temperature for increasing magnetic fields between 0 and 750 mT. d Perpendicular critical magnetic field defined by a 50% drop of the normal-state resistance measured at T = 1.65 K. The orange line corresponds to a linear fit with the Ginsburg Landau formula.

Although KTaO3 is a quantum paraelectric material like SrTiO3, its permittivity is reduced by a factor of five as compared to SrTiO3, making the electric field effect less efficient in a back-gating configuration1,2. To overcome this difficulty, we prepared an AlOx/KTaO3 sample using a thinner substrate (150 μm). After cooling the sample, the back-gate voltage was first swept to its maximum value VG = 200 V while keeping the 2-DEG at the electrical ground. This forming procedure is commonly applied on SrTiO3-based interfaces to ensure the reversibility of the gate sweeps in further gating sequences21. Figure 3 shows the sheet resistance of sample C as a function of temperature for different values of the gate voltage between −40 and 200 V. Electrostatic gating induces both a modulation of the normal-state resistance and a variation of the superconducting critical temperature. For negative gate voltages corresponding to a depleted quantum well, R vs T curves exhibit a quasi-reentrant behavior: the resistance first decreases and then upturns upon further cooling22,23. This is characteristic of disordered superconducting thin films in which superconductivity only exists locally, forming a network of isolated islands surrounded by an insulating medium that precludes percolation. While the decrease of resistance marks the emergence of superconductivity inside the islands, the upturn of resistance at low-temperature results from the opening of a gap in the excitation spectrum, which prevents the flow of quasiparticles across islands. Hence, the resistance does not reach zero, indicating that the superconducting order does not extend at a long range. As carriers are added upon increasing the gate voltage, the resistance curves flatten at low temperature, and the 2-DEG eventually reaches a true zero resistive state (VG > −25 V). Long-range superconducting order is established through Josephson coupling between the islands. Further doping makes the network of islands denser and increases the coupling between islands resulting in a "homogeneous-like" superconducting 2-DEG at high doping. The resulting superconducting phase diagram is shown in Fig. 3b, where the resistance is plotted in color scale as a function of temperature and electron density extracted by combining the Hall effect and gate capacitance measurements10,24. In this experiment, the carrier density was tuned from n2D ≃ 0.95 × 1013e− × cm−2 to n2D ≃ 2.2 × 1013e− × cm−2, which is not sufficient to explain the modulation of the normal resistance by more than one order of magnitude. This indicates that the gate voltage not only controls the carrier density but also modifies deeply the electronic properties of the 2-DEG, in particular the electronic mobility, in agreement with the previous reports7.

a Temperature dependence of the sheet resistance of sample C for different values of the back-gate voltage in the range −40 to 200 V. b Normalized sheet resistance in color scale as a function of the carrier density extracted from Hall measurements and temperature. The dashed line indicates the critical temperature defined by an 80% drop of the resistance with respect to the normal resistance RN at T = 0.9 K. c Sheet resistance at 0.9 K as a function of the back-gate voltage.

We further investigated the superconducting KTaO3 2-DEG by measuring its superfluid stiffness Js, which is the energy scale associated with the phase rigidity of the superconducting condensate. Js is related to the imaginary part of the complex conductivity σ(ω) = σ1(ω) − iσ2(ω) of the superconductor that accounts for the transport of Cooper pairs at finite-frequency ω. This is a direct probe of the superconducting order parameter that provides important information on the nature of the superconducting state. In the low-frequency limit, ℏω ≪ Δ, a superconductor displays an inductive response to an ac electrical current and σ2(ω) = \(\frac{1}{{L}_{{{{{{{{\rm{k}}}}}}}}}\omega }\), where Lk is the kinetic inductance of the superconductor that diverges at Tc25. The superfluid stiffness is then directly related to Lk

where ℏ is the reduced Planck constant and e is the electron charge.

We used resonant microwave transport to extract Lk below Tc and determine the superfluid stiffness of sample D as a function of temperature. The method, which was successfully applied to superconducting SrTiO3-based interfaces, is illustrated in Fig. 4a and described in detail in refs. 26,27. In short, the KTaO3 sample is embedded into a parallel RLC resonant electrical circuit made with surface mount microwave devices (SMD). The capacitance of the circuit is dominated by the KTaO3 substrate contribution (CKTO) due to its large intrinsic permittivity. The total inductance of the circuit, \({L}_{{{{{{{{\rm{tot}}}}}}}}}(T)=\frac{{L}_{1}{L}_{{{{{{{{\rm{k}}}}}}}}}(T)}{{L}_{1}+{L}_{{{{{{{{\rm{k}}}}}}}}}(T)}\), includes the contribution of an SMD inductor (L1) and the contribution of the kinetic inductance Lk of the superconducting 2-DEG below Tc. Finally, an SMD resistor R1 imposes that the dissipative part of the circuit impedance remains close to 50 Ω in the entire temperature range, ensuring a good impedance matching with the microwave circuitry. The circuit resonates at the frequency \({\omega }_{0}=\frac{1}{\sqrt{{L}_{{{{{{{{\rm{tot}}}}}}}}}{C}_{{{{{{{{\rm{KTO}}}}}}}}}}}\), which is accessed by measuring the reflection coefficient of the sample circuit \({{\Gamma }}(\omega )=\frac{{A}^{{{{{{{{\rm{in}}}}}}}}}}{{A}^{{{{{{{{\rm{out}}}}}}}}}}=\frac{Z(\omega )-{Z}_{0}}{Z(\omega )+{Z}_{0}}\). The resonance manifests itself as a peak in the real part of the circuit impedance, Z(ω), accompanied by a π phase shift26. The height and the width of the peak are controlled by the dissipative part of the circuit impedance. In the superconducting state, the 2-DEG conductance acquires a kinetic inductance Lk that generates a shift of ω0 towards high frequencies with respect to the normal state (Fig. 4b). The temperature-dependent superfluid stiffness \({J}^{\exp }\), extracted from the resonance shift and Eq. (1) is presented in Fig. 4c (blue circles).

a Schematics of the measurement setup adapted from ref. 26. The contribution of the sample is represented by the 2-DEG impedance, Z2DEG, and the parallel capacitive contribution of the KTaO3 substrate, CKTO. SMD components, resistor R1 = 75 Ω, inductor L1 = 6.5 nH, are placed in parallel to define a resonating circuit. Large SMD capacitors Cp = 2μF block the dc signals in L1 and R1 without influencing the signals at microwave frequencies. The reflection coefficient at the circuit sample Γ(ω) = \(\frac{{A}^{{{{{{{{\rm{out}}}}}}}}}}{{A}^{{{{{{{{\rm{in}}}}}}}}}}\) is extracted from the measurement as described in ref. 26. A bias-tee is used to separate the dc current and the microwave one. b Top: Real (full lines) and imaginary (dashed lines) parts of the 2-DEG impedance as a function of frequency in the normal state at T = 2.5 K and in the superconducting state at T = 0.2 K after a calibration procedure26. Bottom: corresponding phases of the 2-DEG impedance in the normal and superconducting state. The resonance frequency of the sample circuit can be clearly identified, for instance, as the maximum of the peak in the real part of the impedance or as a π phase shift in its phase. c Superfluid stiffness \({J}_{{{{{{{{\rm{s}}}}}}}}}^{\exp }\) extracted from the resonance frequency and Eq. (1) as a function of the temperature. The dashed purple line shows an attempt to fit the experimental data within a standard BCS model (\({J}_{{{{{{{{\rm{s}}}}}}}}}^{{{{{{{{\rm{BCS}}}}}}}}}\)), which provides a mean field \({T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\) = 2.3 K. A better agreement is obtained by using a BKT model (\({J}_{{{{{{{{\rm{s}}}}}}}}}^{{{{{{{{\rm{BKT}}}}}}}}}\)). On the right axis, the figure also shows the sheet resistance curve (R) fitted with the Halperin and Nelson formula (RHN).

The flattening of the \({J}^{\exp }\) curve below 1 K supports a fully gapped behavior, i.e., an absence of nodes in the order parameter. The purpled dashed line (JBCS) shows an attempt to fit the experimental curve with a standard BCS expression \({J}_{{{{{{{{\rm{s}}}}}}}}}^{{{{{{{{\rm{BCS}}}}}}}}}(T)/{J}_{{{{{{{{\rm{s}}}}}}}}}(0)=({{\Delta }}(T)/{{\Delta }}(0))\tanh ({{\Delta }}(T)/{k}_{{{{{{{{\rm{B}}}}}}}}}T)\)25, where Δ(T) is the superconducting gap obtained numerically by a self-consistent solution of the BCS equation, so that it vanishes at the mean-field temperature \({T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\) (i.e., the temperature at which Cooper pairs form). Since Js(0) is fixed by the experimental value at the lowest temperature, the only free parameter is then the ratio \({{\Delta }}(T=0)/{k}_{{{{{{{{\rm{B}}}}}}}}}{T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\), that determines the curvature of the Js(T) curve. As one can see, even using a relatively strong-coupling value \({{\Delta }}(0)/{k}_{{{{{{{{\rm{B}}}}}}}}}{T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\) = 2.3, from the fit of the low-temperature curve, one obtains \({T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\) ≃ 2.2 K, that is larger than the experimental Tc. To fit the data in the whole temperature range with the BCS expression only, one would then need an unreasonably large value (\({{\Delta }}(0)/{k}_{{{{{{{{\rm{B}}}}}}}}}{T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\) ≃ 6), a result that we checked to hold irrespectively of the exact functional BCS form used to fit the stiffness. Here we follow a different approach and interpret the rapid drop of Js(T) below the BCS fit as a BKT signature, as we will discuss below. This interpretation is supported by a second striking observation that holds regardless of any specific consideration about its temperature dependence: the T = 0 value of the stiffness Js(T = 0) ≃ 7.3 K is of the same order as Tc ≃ 2.2 K. It is worth noting that in conventional superconductors, where the superfluid density ns(T = 0) is close to the carrier density n2D, the stiffness at zero temperature is of the order of the Fermi energy, and then several orders of magnitude larger than \({T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\). A strong reduction of Js(0) is instead observed in 2D-superconductors, where disorder strongly reduces ns with respect to n2D already at T = 028,29,30,31,32,33,34,35,36,37,38. In the dirty limit, in which the elastic scattering rate 1/τ is much larger than the superconducting gap, only a fraction of carriers, ns/n2D ≃ 2Δ(0)/(ℏ/τ), forms the superconducting condensate. In a single-band picture, an estimate of the superfluid stiffness is obtained from Δ(0) and the normal resistance RN, \({J}_{{{{{{{{\rm{s}}}}}}}}}\simeq \frac{\pi \hslash {{\Delta }}(0)}{4{e}^{2}{R}_{{{{{{{{\rm{N}}}}}}}}}}\). Using the previously estimated value of Δ(0) ≃ 5 K and RN ≃ 1300 Ω, we obtain Js ≃ 11.8 K, close to the measured value (\({J}_{{{{{{{{\rm{s}}}}}}}}}^{\exp }(T=0)\) ≃ 7.3 K), which is consistent with the dirty limit.

The superfluid density of the 2-DEG can be directly deduced from the stiffness through the formula \({n}_{{{{{{{{\rm{s}}}}}}}}}=\frac{4m}{{\hslash }^{2}}{J}_{{{{{{{{\rm{s}}}}}}}}}\), where m is the effective mass of superconducting electrons. In the case of (111)-KTaO3 2-DEGs, the conduction band is derived from the bulk J = 3/2 states with a Fermi Surface formed by a hexagonal contour inside a sixfold symmetric star-shaped contour39,40. Considering an average effective mass m ≃ 0.5m0, the corresponding superfluid density ns extracted from \({J}_{{{{{{{{\rm{s}}}}}}}}}^{\exp }\) is ns ≃ 1.8 × 1012 e− cm−2, which is about 2.5% of the total carrier density (n2D = 7.5 × 1013 e− cm−2 for sample D). This very low ratio is comparable with previous findings in LaAlO3/SrTiO3 interfaces26,41,42. Although such reduced superfluid density is consistent with the dirty limit, KTaO3-(111) 2-DEG is a multiband system40, in which superconductivity may involve only specific bands, as also suggested in SrTiO326.

The reduced dimensionality and the suppression of the energy scale associated with the stiffness represent the prerequisites to observe BKT13,14,15 physics, since it makes the BKT temperature scale TBKT associated with the unbinding of vortex-antivortex pairs far enough from \({T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\)43. The most famous hallmark of the BKT transitions is the discontinuous jump to zero of Js at TBKT < Tc with a universal ratio Js(TBKT)/TBKT = 2/π44. Such a prediction, theoretically based on the study of the 2D XY model13,14,15, has been successfully confirmed in superfluid He films45. In practice, the experimental observation of the BKT transition in real superconductors is more subtle. Indeed, in thin films, the suppression of ns (and then Js) with disorder comes along with an increasing inhomogeneity of the SC background, that is predicted to smear out the discontinuous superfluid-density jump35,46,47,48 into a rapid downturn, as observed experimentally via the direct measurement of the inverse penetration depth31,32,33,34,35,36,37,38 or indirectly via the measurement of the exponent of the nonlinear IV characteristics near Tc16,28,29,30. In the case of SrTiO3-based interfaces, the direct measurement of Js is rather challenging, and the few experimental reports available so far do not evidence a BKT jump26,41,42.

Within the BKT approach, the effect of vortex-like topological excitations provides additional suppression of Js with respect to the BCS dependence discussed above, driven only by quasiparticle excitations. To provide a fit of \({J}_{{{{{{{{\rm{s}}}}}}}}}^{\exp }\), we then solved numerically the renormalization-group (RG) equations of the BKT theory for the superfluid stiffness and vortex fugacity. As input parameters of the RG equations, we used the BCS temperature dependence of the stiffness. As mentioned above, the low-temperature part is fully captured by the BCS approximation, and for the estimated \({{\Delta }}(0)/{k}_{{{{{{{{\rm{B}}}}}}}}}{T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\) ratio, the dirty-limit and the clean-limit expressions of \({J}_{{{{{{{{\rm{s}}}}}}}}}^{{{{{{{{\rm{BCS}}}}}}}}}\) provide the same result. We also included the finite-frequency effects in our calculation43,49,50. Indeed, even though the resonance frequency (about 0.5 GHz) is still small as compared to the optical gap (2Δ ~ 10 K ~ 200 GHz) it can nonetheless lead to non-negligible effects, in particular a rounding of the jump and suppression of the stiffness at a temperature slightly larger than the one where the dc resistivity vanishes32,33,38, as indeed observed in our case. The resistivity itself is consistently fitted with the interpolating Halperin–Nelson formula50, which accounts for BKT-like fluctuations between TBKT and \({T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\), and for standard Gaussian fluctuations above \({T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\). Finally, to account for spatial inhomogeneities, we introduce a gaussian distribution of local Tc and Js with variance σG centered around \({T}_{{{{{{{{\rm{c}}}}}}}}}^{0}\) and \({J}_{{{{{{{{\rm{s}}}}}}}}}^{\exp }(0)\). As seen in Fig. 4c, the result of the fitting procedure (dashed red line) is in very good agreement with experimental data both for the superfluid stiffness and the resistance, considering a very small inhomogeneity, σG = 0.02. Details on the fitting procedure are given in the Method Section.

Although KTaO3 and SrTiO3 have many common properties, the superconducting phases of their interfacial 2-DEG exhibit noticeable differences. Whereas a pure BCS weak-coupling limit with Δ(0)/kBTc ≃ 1.76 provides a very good description of superconductivity in SrTiO3-based interfaces26,27, we found a stronger value of the coupling for KTaO3 (Δ(0)/kBTc ≃ 2.3). Such an important difference, which must be traced back to the pairing mechanism, is a strong constraint on the possible origin of superconductivity in these two materials. In addition, BKT physics was not observable in SrTiO3 for which a simple BCS model without phase fluctuations was sufficient to fit the Js(T) curves with a very good accuracy27. This may suggest more bosonic-like superconductivity in KTaO3-based interfaces (in the highly doped regime), as evidenced by the large separation between the pairing scale, set by Δ, and the phase-coherence scale, set by the small value of the superfluid stiffness. Recent measurements of the in-plane critical field in KTaO3-based interfaces suggested that the order parameter could be a mixture of s-wave and p-wave pairing components induced by strong spin-orbit coupling51. While we can not rule out this possibility, the saturation of the Js(T) curve below Tc/2 seen in Fig. 4b suggests a dominance of the fully gapped s-wave component. In addition, despite the complex band structure of the KTaO3-(111) interfaces, we have not observed any signatures suggesting multi-gap superconductivity in our data. Further experiments, including tunneling spectroscopy, are therefore necessary to understand the nature of superconductivity in KTaO3-based interfaces.

Prior to deposition, KTaO3 (111) substrates from MTI corporation were annealed at 600 ∘C for 1 h in vacuum. Then, the thin Al layer was deposited in a dc magnetron sputtering system (PLASSYS MP450S) under a base pressure of the vacuum chamber lower than 5 × 10−8 mbar. During Al deposition, the Ar partial pressure and the dc power were kept fixed at 5 × 10−4 mbar and 10 W, respectively. The deposition rate for Al was 0.66 Å/s. Table 1 below summarizes the deposition parameters for the different samples.

X-ray photoelectron spectroscopy (XPS) was measured using a non-monochromatized Mg Kα source (hν = 1253.6 eV) in an Omicron NanoTechnology GmbH system with a base pressure of 5 × 10−10 mbar. The operating current and voltage of the source was 20 mA and 15 kV, respectively. Spectral analysis to determine different valence states of Ta were carried out using the CasaXPS software. Adventitious carbon was used as a charge reference to obtain the Ta 4f5/2 peak position for the fitting. The energy difference and the ratio of the area between 4f5/2 and 4f3/2 peaks for all the Ta valence states were constrained according to the previously reported values.

STEM-HAADF and STEM-EELS measurements have been done at 100 keV using a Cs corrected Nikon STEM microscope and a Gatan modified EELS spectrometer equipped with a MerlinEM detector.

In order to account for vortex excitations, we solved the BKT RG equations15,43,44 for the vortex fugacity \(g=2\pi {e}^{-\mu /({k}_{B}T)}\), with μ the vortex-core energy, and the rescaled stiffness K ≡ πJs/kBT:

where \(\ell=\ln (a/{\xi }_{0})\) is the RG-scaled lattice spacing with respect to the coherence length ξ0, that controls the vortex sizes and appears as a short-scale cut-off for the theory. The initial values at ℓ = 0 are set by the BCS fitting JBCS(T) of \({J}_{{{{{{{{\rm{s}}}}}}}}}^{\exp }\), and the renormalized stiffness is given by the large-scale behavior, Js = (kBT/π)K(ℓ → ∞). The ratio μ/Js = 0.87, similar to the one found in other conventional superconductors34,35,37,38, is used as a free (temperature-independent) parameter, which controls the strength of stiffness renormalization due to bound vortices below TBKT43. To account for finite-frequency effects, we further include a dynamical screening of vortices49,50 via an effective frequency-dependence dielectric function ε(ω) which enters in the complex conductivity of the film as \(\sigma (\omega )=-\frac{4{J}^{{{{{{{{\rm{BCS}}}}}}}}}{e}^{2}}{i\omega {\hslash }^{2}\varepsilon (\omega )}\). At zero frequency ε(ω) is real and ε1(0) = K(0)/K(ℓ → ∞) = JBCS/Js so one recovers the usual static result. At finite-frequency ε(ω) develops an imaginary part due to the vortex motion, that can be expressed in the first approximation49 as \({\varepsilon }_{2}\simeq {({r}_{\omega }/\xi )}^{2}\), where ξ is the vortex correlation length and rω is a finite length scale set in by the finite frequency of the probe, i.e. \({r}_{\omega }=\sqrt{\frac{14{D}_{v}}{\omega }}\), with Dv the vortex diffusion constant of the vortices. The main effect of ε2 is to induce a small tail above TBKT for the finite-frequency stiffness, as given by Js = ℏ2ωσ2(ω)/(4e2), as we indeed observe in the experiments. Here we follow the same procedure outlined in ref. 38 to compute ε(ω), and in full analogy, with this previous work, we find a very small vortex diffusion constant Dv ~ 1010 nm2/s. The correlation length ξ(T) also enters the temperature dependence of the resistivity above TBKT, that follows the usual scaling R/RN = 1/ξ2(T). To interpolate between the BKT and Gaussian regime of fluctuations, we use the well-known Halperin–Nelson expression43,46,50 \({\xi }_{HN}(T)=\frac{2}{A}\sinh \left(\frac{b}{\sqrt{t}}\right)\) where t = (T − TBKT)/TBKT, and we set A = 2.5 and b = 0.27, consistent with the theoretical estimate of b ≃ 0.2 that we obtain from the value of μ35,38,43. Finally, to account for the possible inhomogeneity of the sample, we consider the extension of the previous method to the case where the overall complex conductivity of the sample is computed in the self-consistent effective-medium approximation52 as the solution of the following equation:

Here σi(ω) denotes the complex conductivity of a local superconducting puddle with stiffness Ji and local \({T}_{{{{{{{{\rm{c}}}}}}}}}^{i}\), that are taken with a Gaussian distribution Pi with variance σG centered around the BCS fit of \({J}_{{{{{{{{\rm{s}}}}}}}}}^{\exp }\). For each realization Ji we then compute the Js,i from the solution of the BKT equations (2) -(3), we determine the corresponding complex conductivity σi(ω) and we finally solve Eq. (4) to get the average \({J}_{{{{{{{{\rm{s}}}}}}}}}^{{{{{{{{\rm{BKT}}}}}}}}}=({\hslash }^{2}/4{e}^{2})\omega {\sigma }_{2}(\omega )\) below Tc and the average σ1(ω = 0) ≡ 1/RHN above Tc, i.e., the dashed lines reported in Fig. 4c. Further details about the implementation of the effective-medium approximation can be found in refs. 16,38. The main effect of inhomogeneity is to contribute slightly to the suppression of Js with respect to JBCS before TBKT. In our case, we checked that inhomogeneity, if present, is very small, and a σG = 0.02 is enough to account for the measured temperature dependences.

The authors declare that the data that support the findings of this study are available within the article. All other relevant data are available from the corresponding authors upon request.

Fujii, Y. & Sakudo, T. Dielectric and optical properties of KTaO3. J. Phys. Soc. Jpn. 41, 888–893 (1976).

Article ADS CAS Google Scholar

Fleury, P. A. & Worlock, J. M. Electric-field-induced Raman scattering SrTiO3 and KTaO3. Phys. Rev. 174, 613 (1968).

Article ADS CAS Google Scholar

Schooley, J. F., Hosler, W. R. & Cohen, M. L. Superconductivity in semiconducting SrTiO3. Phys. Rev. Lett. 12, 474–475 (1964).

Article ADS CAS Google Scholar

Thompson, J. R., Boatner, L. A. & Thomson, J. O. Very low-temperature search for superconductivity in semiconducting KTaO3. J. Low Temp. Phys. 47, 467 (1982).

Article ADS CAS Google Scholar

Ueno, K. et al. Discovery of superconductivity in KTaO3 by electrostatic carrier doping. Nature Nano. 6, 408–412 (2011).

Article ADS CAS Google Scholar

Liu, C. et al. Two-dimensional superconductivity and anisotropic transport at KTaO3 (111) interfaces. Science 371, 716–721 (2021).

Article ADS CAS PubMed Google Scholar

Chen, Z. et al. Electric field control of superconductivity at the LaAlO3/KTaO3 (111) interface. Science 372, 721–724 (2021).

Article ADS CAS PubMed Google Scholar

Reyren, N. et al. Superconducting interfaces between insulating oxides. Science 317, 1196–1199 (2007).

Article ADS CAS PubMed Google Scholar

Caviglia, A. D. et al. Electric field control of the LaAlO3/SrTiO3 interface ground state. Nature 456, 624–627 (2008).

Article ADS CAS PubMed Google Scholar

Biscaras, J. et al. Two-dimensional superconducting phase in LaTiO3/SrTiO3 heterostructures induced by high-mobility carrier doping. Phys. Rev. Lett. 108, 247004 (2012).

Article ADS CAS PubMed Google Scholar

Chen, Z. et al. Two-dimensional superconductivity at the LaAlO3/KTaO3 (110) heterointerface. Phys. Rev. Lett. 126, 026802 (2021).

Article ADS CAS PubMed Google Scholar

Liu, C. et al. Tunable superconductivity at the oxide-insulator/KTaO3 interface and its origin. Preprint at arXiv:2203.05867 (2022).

Berezinskii, V. L. Destruction of long-range order in one-dimensional and two-dimensional systems possessing a continuous symmetry group. II. Quantum systems. Sov. Phys. JETP 34, 610 (1972).

ADS Google Scholar

Kosterlitz, J. M. & Thouless, D. J. Ordering, metastability and phase transitions in two-dimensional systems. J. Phys. C 6, 1181 (1973).

Article ADS CAS Google Scholar

Kosterlitz, J. M. The critical properties of the two-dimensional xy model. J. Phys. C 7, 1046 (1974).

Article ADS Google Scholar

Venditti, G. et al. Nonlinear I-V characteristics of two-dimensional superconductors: Berezinskii-Kosterlitz-Thouless physics versus inhomogeneity. Phys. Rev. B 100, 064506 (2019).

Article ADS CAS Google Scholar

Vaz, D. C. et al. Mapping spin-charge conversion to the band structure in a topological oxide two-dimensional electron gas. Nature Mat. 18, 1187–1193 (2019).

Article ADS CAS Google Scholar

Vicente-Arche, L. M. et al. Metal/SrTiO3 two-dimensional electron gases for spin-to-charge conversion. Phys. Rev. Mat. 5, 064005 (2021).

CAS Google Scholar

Rödel, T. C. et al. Universal fabrication of 2D electron systems in functional oxides. Adv. Mater. 28, 1976–1980 (2016).

Article PubMed Google Scholar

Vicente-Arche, L. M. et al. Spin-charge interconversion in KTaO3 2D electron gases. Adv. Mater. 33, 2102102 (2021).

Article CAS Google Scholar

Biscaras, J. et al. Limit of the electrostatic doping in two-dimensional electron gases of LaXO3 (X = Al, Ti)/SrTiO3. Sci. Rep. 4, 6788 (2014).

Article CAS PubMed PubMed Central Google Scholar

Jaeger, H. M., Haviland, D. B., Orr, B. G. & Goldman, A. M. Onset of superconductivity in ultrathin granular metal films. Phys. Rev.B 40, 182–196 (1989).

Article ADS CAS Google Scholar

Orr, B. G., Jaeger, H. M. & Goldman, A. M. Local superconductivity in ultrathin Sn films. Phys. Rev.B 32, 7586(R) (1985).

Article ADS Google Scholar

Singh, G. et al. Effect of disorder on superconductivity and Rashba spin-orbit coupling in LaAlO3/SrTiO3 interfaces. Phys. Rev. B 96, 024509 (2017).

Article ADS Google Scholar

Tinkham, M. Introduction to Superconductivity 2nd edn (Dover Publications, Inc., 2004).

Singh, G. et al. Competition between electron pairing and phase coherence in superconducting interfaces. Nat. Commun. 9, 407 (2018).

Article ADS CAS PubMed PubMed Central Google Scholar

Singh, G. et al. Gap suppression at a Lifshitz transition in a multi-condensate superconductor. Nature Mat. 18, 948–954 (2019).

Article ADS CAS Google Scholar

Epstein, K., Goldman, A. M. & Kadin, A. M. Vortex-antivortex pair dissociation in two-dimensional superconductors. Phys. Rev. Lett. 47, 534 (1981).

Article ADS CAS Google Scholar

Epstein, K., Goldman, A. M. & Kadin, A. M. Renormalization and the Kosterlitz-Thouless transition in a two-dimensional superconductor. Phys. Rev. B 27, 6691 (1983).

Article ADS Google Scholar

Fiory, A. T., Hebard, A. F. & Glaberson, W. I. Superconducting phase transitions in indium/indium-oxide thin-film composites. Phys. Rev. B 28, 5075 (1983).

Article ADS CAS Google Scholar

Turneaure, S. J., Lemberger, T. R. & Graybeal, J. M. Effect of thermal phase fluctuations on the superfluid density of two-dimensional superconducting films. Phys. Rev. Lett. 84, 987 (2000).

Article ADS CAS PubMed Google Scholar

Crane, R. W. et al. Fluctuations, dissipation, and nonuniversal superfluid jumps in two-dimensional superconductors. Phys. Rev. B 75, 094506 (2007).

Article ADS CAS Google Scholar

Liu, W., Kim, M., Sambandamurthy, G. & Armitage, N. P. Dynamical study of phase fluctuations and their critical slowing down in amorphous superconducting films. Phys. Rev. B 84, 024511 (2011).

Article ADS CAS Google Scholar

Kamlapure, A. et al. Measurement of magnetic penetration depth and superconducting energy gap in very thin epitaxial NbN films. Appl. Phys. Lett. 96, 072509 (2010).

Article ADS CAS Google Scholar

Mondal, M. et al. Role of the vortex-core energy on the Berezinskii-Kosterlitz-Thouless transition in thin films of NbN. Phys. Rev. Lett. 107, 217003 (2011).

Article ADS PubMed CAS Google Scholar

Misra, S., Urban, L., Kim, M., Sambandamurthy, G. & Yazdani, A. Measurements of the magnetic-field-tuned conductivity of disordered two-dimensional Mo43Ge57 and InOx. Superconducting films: evidence for a universal minimum superfluid response. Phys. Rev. Lett. 110, 037002 (2013).

Article ADS CAS PubMed Google Scholar

Yong, J., Lemberger, T., Benfatto, L., Ilin, K. & Siegel, M. Robustness of the Berezinskii-Kosterlitz-Thouless transition in ultrathin NbN films near the superconductor-insulator transition. Phys. Rev. B 75, 184505 (2013).

Article ADS CAS Google Scholar

Ganguly, R., Chaudhuri, D., Raychaudhuri, P. & Benfatto, L. Slowing down of vortex motion at the Berezinskii-Kosterlitz-Thouless transition in ultrathin NbN films. Phys. Rev. B 91, 054514 (2015).

Article ADS CAS Google Scholar

Bareille, C. et al. Two-dimensional electron gas with six-fold symmetry at the (111) surface of KTaO3. Sci. Rep. 4, 3586 (2014).

Article CAS PubMed PubMed Central Google Scholar

Bruno, F. Y. et al. Band structure and spin-orbital texture of the (111)-KTaO32D electron gas. Adv. Electron. Mater. 5, 1800860 (2019).

Article CAS Google Scholar

Manca, N. et al. Bimodal phase diagram of the superfluid density in LaAlO3/SrTiO3 revealed by an interfacial waveguide resonato. Phys. Rev. Lett. 122, 036801 (2019).

Article ADS CAS PubMed Google Scholar

Bert, J. A. et al. Gate-tuned superfluid density at the superconducting LaAlO3/SrTiO3 interface. Phys. Rev. B 86, 060503(R) (2012).

Article ADS CAS Google Scholar

Benfatto, L., Castellani, C. & Giamarchi, T. in 40 Years of Berezinskii-Kosterlitz-Thouless Theory (ed. Josè, J. V.) Ch. 5 (World Scientific,2013).

Nelson, D. R. & Kosterlitz, J. M. Universal jump in the superfluid density of two-dimensional superfluids. Phys. Rev. Lett. 39, 1201–1205 (1977).

Article ADS CAS Google Scholar

McQueeney, D., Agnolet, G. & Reppy, J. D. Surface superfluidity in dilute 4He-3He mixtures. Phys. Rev. Lett. 52, 1325 (1984).

Article ADS CAS Google Scholar

Benfatto, L., Castellani, C. & Giamarchi, T. Broadening of the Berezinskii-Kosterlitz-Thouless superconducting transition by inhomogeneity and finite-size effects. Phys. Rev. B 80, 214506 (2009).

Article ADS CAS Google Scholar

Mondal, M. et al. Phase fluctuations in a strongly disordered s-wave NbN superconductor close to the metal-insulator transition. Phys. Rev. Lett. 106, 047001 (2011).

Article ADS PubMed CAS Google Scholar

Maccari, I., Benfatto, L. & Castellani, C. Broadening of the Berezinskii-Kosterlitz-Thouless transition by correlated disorder. Phys. Rev. B 96, 060508 (R) (2017).

Article ADS Google Scholar

Ambegaokar, V., Halperin, B. I., Nelson, D. R. & Siggia, E. D. Dynamics of superfluid films. Phys. Rev. B 21, 1806 (1979).

Article ADS Google Scholar

Halperin, B. I. & Nelson, D. R. Resistive transition in superconducting films. J. Low. Temp. Phys. 36, 599 (1979).

Article ADS CAS Google Scholar

Zhang, G. et al. Spontaneous rotational symmetry breaking in KTaO3 interface superconductor. Preprint at arXiv:2111.05650v2 (2021).

Kirkpatrick, S. Percolation and conduction. Rev. Mod. Phys. 45, 574 (1973).

Article ADS Google Scholar

Download references

This work was supported by the ANR QUANTOP Project-ANR-19-CE47-0006 grant, by the QuantERA ERA-NET Cofund in Quantum Technologies (Grant Agreement N. 731473) implemented within the European Union's Horizon 2020 Program (QUANTOX), by EU under project MORE-TEM ERC-SYN (grant agreement No 951215) and by Sapienza University of Rome, through the projects Ateneo 2019 (Grant No. RM11916B56802AFE) and Ateneo 2020 (Grant No. RM120172A8CC7CC7), and by the Italian MIUR through the Project No. PRIN 2017Z8TS5B.

These authors contributed equally: S. Mallik, G. Ménard, G. Saïz.

Unité Mixte de Physique, CNRS, Thales, Université Paris-Saclay, 1 Avenue Augustin Fresnel, 91767, Palaiseau, France

S. Mallik, H. Witt & M. Bibes

Laboratoire de Physique et d’Etude des Matériaux, ESPCI Paris, PSL University, CNRS, Sorbonne Université, Paris, France

G. C. Ménard, G. Saïz, H. Witt, J. Lesueur & N. Bergeal

Laboratoire de Physique des Solides, Université Paris-Saclay, CNRS UMR 8502, 91405, Orsay, France

A. Gloter

Department of Physics and ISC-CNR, Sapienza University of Rome, Rome, Italy

L. Benfatto

You can also search for this author in PubMed Google Scholar

N.B. and M.B. proposed and supervised the study. S.M. and H. W. prepared the samples and performed XPS experiments and their analysis. A.G. performed the STEM and EELS analysis. G.C.M., G.S., and S.M. performed the dc and microwave transport experiments and analysed the data with input from M.B., J.L., and N.B. L.B. conducted the BKT analysis of microwave data. N.B., M.B., and L.B. wrote the manuscript with input from all authors. All authors discussed the results and contributed to their interpretation.

Correspondence to N. Bergeal.

The authors declare no competing interests.

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work.

Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

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/.

Reprints and Permissions

Mallik, S., Ménard, G.C., Saïz, G. et al. Superfluid stiffness of a KTaO3-based two-dimensional electron gas. Nat Commun 13, 4625 (2022). https://doi.org/10.1038/s41467-022-32242-y

Download citation

Received: 18 February 2022

Accepted: 21 July 2022

Published: 08 August 2022

DOI: https://doi.org/10.1038/s41467-022-32242-y

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Nature Communications (2023)

Communications Physics (2023)

Nature Communications (2023)

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.

News