Re-entrant superconductivity at an oxide heterointerface

Abstract

A magnetic field typically suppresses superconductivity by either breaking Cooper pairs via the Zeeman effect or inducing vortex formation. However, under certain circumstances, a magnetic field can stabilize superconductivity instead. This seemingly counterintuitive phenomenon is associated with magnetic interactions and has been extensively studied in three-dimensional materials. By contrast, this phenomenon, hinting at unconventional superconductivity, remains largely unexplored in two-dimensional systems, with moiré-patterned graphene being the only known example. Here, we report the observation of re-entrant superconductivity (RSC) at the epitaxial (110)-oriented LaTiO3-KTaO3 interface. This phenomenon occurs across a wide range of charge carrier densities, which, unlike in three-dimensional materials, can be tuned in-situ via electrostatic gating. We attribute the re-entrant superconductivity to the interplay between a strong spin-orbit coupling and a magnetic-field driven modification of the Fermi surface. Our findings offer new insights into re-entrant superconductivity and establish a robust platform for exploring novel effects in two-dimensional superconductors.

Figures (3)

• Figure 1: Properties of the (110)-oriented LaTiO$_3$-KTaO$_3$ interface. a) Sketch of the interface between LaTiO$_3$ and KTaO$_3$. The backside of KTaO$_3$ acts as an electrode to apply an electrostatic field to tune the in-situ interface electronic properties. b) Relativistic band structure of a 2D electron gas emerging at the (110) interface, which was simulated from first principles. Nearly flat-band regions around $E_F$ = 0 along $\Gamma$–X near the Brillouin zone boundary are formed almost equally by highly localized $d_{z^2}$ orbitals of Ta and Ti. Solid lines are bands that are important for the theory model presented below. Dashed lines are other bands. The inset shows the unit cell in real space. c) The backgate voltage can effectively tune the sample resistance. Inset depicts the photograph of the device under study. The device allows to flow the current along the [001] and [1-10] crystallographic directions independently. d) Superconducting transition at different backgate voltages $V_\text{BG}$. Note that the resistance is normalized to its value. The superconducting transition temperature $T_\text{c}$ is defined as a temperature at which the resistance has dropped by 50$\%$ from its value at normal conducting state. e) Gate dependence of the superconducting transition temperature $T_\text{c}$ and the charge carrier density $n$.
• Figure 2: Experimental visualization of re-entrant superconducting state. a) Scheme of the experimental arrangement. b) Color rendition plot of the longitudinal resistance $R_\text{xx}$ as functions of the gate voltage and the magnetic field. The re-entrant superconductivity visualizes as the temperature increases. c) Magnetoresistance $R_\text{xx}$-traces for various temperature at $V_\text{BG}=+50$ V. A resistive peak emerges at $B=0.9$ T as the temperature increases. d) Gate voltage dependence of the amplitude of the resistive peak at several temperatures.
• Figure 3: Theoretical model to explain the observed re-entrant superconductivity. a) Left panel: Brillouin zone and Fermi surface demonstrating the van Hove singularity in the absence of spin-orbit coupling at the Brillouin zone boundary. The lattice constants are $a_{[1\overline{1}0]}$ and $a_{[100]},$ respectively. Right panel: zoom in on the states near the van Hove singularity taking into account spin-orbit coupling and magnetic field. Dashed lines correspond to the Fermi surface without spin-orbit coupling. Solid lines, marked with the corresponding spin orientations, show the deformation of the Fermi surface by spin-orbit coupling and magnetic field. The Fermi surface states contributing to the Cooper pairing are shown as filled circles. b) Numerical result demonstrating the existence of the required for the re-entrant superconductivity minimum in the $T_{c}(B)-$dependence. $T_{c}$ is normalized to its value at $B=0$.