Boosting superconductivity in ultrathin YBa$_2$Cu$_3$O$_{7-δ}$ films via nanofaceted substrates

Abstract

In cuprate high-temperature superconductors the doping level is fixed during synthesis, hence the charge carrier density per CuO$_2$ plane cannot be easily tuned by conventional gating, unlike in 2D materials. Strain engineering has recently emerged as a powerful tuning knob for manipulating the properties of cuprates, in particular charge and spin orders, and their delicate interplay with superconductivity. In thin films, additional tunability can be introduced by the substrate surface morphology, particularly nanofacets formed by substrate surface reconstruction. Here we show a remarkable enhancement of the superconducting onset temperature $T_{\mathrm{c}}^{\mathrm{on}}$ and the upper critical magnetic field $H_{c,2}$ in nanometer-thin YBa$_2$Cu$_3$O$_{7-δ}$ films grown on a substrate with a nanofaceted surface. We theoretically show that the enhancement is driven by electronic nematicity and unidirectional charge density waves, where both elements are captured by an additional effective potential at the interface between the film and the uniquely textured substrate. Our findings show a new paradigm in which substrate engineering can effectively enhance the superconducting properties of cuprates. This approach opens an exciting frontier in the design and optimization of high-performance superconducting materials.

Figures (5)

• Figure 1: Engineering the film-substrate interface: The origin of dramatic modifications in the YBCO ground state.a, By annealing (110)-oriented MgO substrates, nanoscale facets are created on the surface, leading to significant differences in atomic coordination between the valleys and the facet edges, particularly at the apexes (shown in the gradient of blue, with stronger color indicating lower coordination of surface atoms). When the YBCO film (grey slab) is deposited, it interacts with this anisotropic substrate matrix, since the YBCO unit cell height is comparable to the facet height. This induces strong coupling between the film and the substrate, mainly at the under-coordinated apex regions (darker blue), where hybridization occurs to saturate dangling bonds. The zoom-in on top illustrates the YBCO planar tight-binding structure at the atomic scale (red dots), with hopping parameters $t$ (nearest neighbor) and $t'$ (next nearest neighbor). The coupling between the under-coordinated substrate regions (dark blue dots) and the YBCO atoms is mediated by a coupling parameter $t^{\perp}$, while the coupling to the more coordinated regions (light blue dots) is negligible. This is evident in both transport and spectroscopic measurements. b,c, Resistivity $\rho(T)$ measured along the $a$- (blue line) and $b$-axis (pink line) for a $50$ nm thick film (b) and a $10$ nm thick film (c). The black dashed lines are linear fits to the high-$T$ resistivity, while $T=T_{\mathrm{L}}^{a,b}$ (blue diamonds and pink circles) mark the temperatures at which the linear $\rho(T)$ ends. In the 10 nm film, two key effects are observed: on one hand, the linear-in-$T$ resistivity along the $b$-axis extends to much lower temperatures compared to the $a$-axis, as previously discussed in Ref. wahlberg2021; second, the onset of the resistive transition $T_{\mathrm{c}}^{\mathrm{on}}$, defined at 90% of the normal state resistivity, is approximately 15 K higher in both in-plane directions compared to the $d=50$ nm film. d,e, CDW signal, measured using Resonant Inelastic X-ray Scattering (RIXS) at $T=T_{\mathrm{c}}=70$ K for the $a$-axis (d) and $b$-axis (e). For each direction, the 50 nm film (triangles) and the 10 nm film (hexagons) are compared. The CDW is thickness independent along the $b$-axis, whereas it disappears along the $a$-axis for the 10 nm film.
• Figure 2: Phase diagram of strained YBCO thin films. Open and filled symbols are from measurements of unpatterned thin films and patterned Hall bars, respectively. All lines are guides for the eye. $T_{\mathrm{c}}^{\mathrm{on}}$ (red squares) is the superconducting transition onset temperature as defined in the main text. For comparison the red dash-dotted line shows $T_{\mathrm{c}}^{\mathrm{on}}$ measured in relaxed YBCO thin films and crystals and the red striped area the suppression of $T_{\mathrm{c}}^{\mathrm{on}}$ from quadratic $p$ dependence due to competition with the CDW order. $H_{c,2}$ is the upper critical field in YBCO single crystals.
• Figure 3: Thickness dependence of $H_{c,2}$ in underdoped YBCO thin films. a,b, The magnetic field dependence of the resistive transition in two YBCO thin films of different thickness and similar doping level. The magnetic field is applied along the $c$-axis, perpendicular to the current which is applied along the $a$-axis. The gray dashed lines indicate where the resistance has dropped to 50% of the normal state resistance $R_N$ at $T_c$. c, Linear fits to $T_c^{0.5R_N}(H)$ for the two films in (a) and (b). The blue and pink symbols show the linear extrapolated values of $H_{c,2}$ at $T=0$. The inset is a blow up of the data points to highlight the different slopes of the linear fits. The error bars are estimated from the uncertainty of the $T$=0 value of the linear fits. d,e, Vortex lattice melting fits for Hall bars on films with similar doping to (a) and (b). The violet squares are values of $B_M$ extracted from measurements of the resistive transition in an out-of-plane ($c$-axis) pulsed magnetic field reaching 55 T. The current is applied along the $a$-axis.
• Figure 4: Doping dependence of $H_{c,2}$ in strained YBCO films. a, Linear fits of $T_c^{0.5R_N}(H)$ for films with different values of $p$. The current is applied along the $a$-axis. The pink hexagons show the linear extrapolated values of $H_{c,2}$ at $T_c^{0.5R_N}=0$. The error bars are estimated from the uncertainty of the $T$=0 value of the linear fits.b, Doping dependence of $H_{c,2}$ in $d=10$ nm (pink hexagons) and $d=50$ nm (blue triangles) YBCO films. The blue dashed line shows the doping dependence of $H_{c,2}$ measured in YBCO crystals grissonnanche2014direct. c, Difference between $T_c^{on}$ (red squares) and $H_{c,2}$ (pink hexagons) in $d=10$ nm and $d=50$ nm films.
• Figure 5: Theoretically calculated dependence of the magnetic field on temperature.a, Magnetic field $h$ versus temperature $T$ for "Model 1" (checkerboard CDW with $Q=(\pi,\pi)$ and no nematicity $\alpha=0$ representing 50 nm thick films, blue line and squares), "Model 2" (uniaxial CDW with $Q=(\pi,0)$ and nematic Fermi surface $\alpha=0.06$ representing 10 nm thick films, green line and triangles), and reference (no CDW and no nematicity $\alpha=0$, grey line and diamonds). Solid lines are fits to a function $h=A+BT^2$, with $A, B$ as fit parameters. Magnetic field $h$ as reduced dimensionless field ($h = eH / (\hbar c)$) and temperature $T$ as dimensionless quantity (temperature in Kelvin given by $T \cdot t / k_B$). b,c, Fermi surfaces considered for "Model 1" (b) and "Model 2" (c). Red lines indicate CDW wave vectors, with red points marking Fermi surface points nested by these wave vectors. Yellow dashed lines represent the superconducting $d$-wave nodal lines.Since the nested points in "Model 1" are close to the antinodes, a CDW gap affects $d$-wave superconductivity more significantly than in "Model 2", where the nested points are closer to the nodes.