Angle-resolved photoemission spectroscopy of superconducting (La,Pr)₃Ni₂O₇/SrLaAlO₄ heterostructures

ABSTRACT

Ruddlesden-Popper bilayer nickelate thin film superconductors discovered under ambient pressure enable great possibilities for investigating electronic structures of the superconducting state. Here, we report angle-resolved photoemission spectroscopy (ARPES) measurements of 1, 2, and 3 unit-cell epitaxial La_2.85Pr_0.15Ni₂O₇ films grown on SrLaAlO₄ substrates, through pure-oxygen in situ sample transportation. Evidence obtained using photons with distinct probing depths shows that conduction is localized primarily at the first unit cell near the interface. Scanning transmission electron microscopy (STEM), together with energy-dispersive X-ray spectroscopy (EDS) and electron energy loss spectroscopy (EELS), indicates that interfacial Sr diffusion and pronounced p-d hybridization gradient may collectively account for the interfacial confinement of conduction. Fermi surface maps reveal hole doping compared to non-superconducting ambient-pressure bulk crystals. Measurements of dispersive band structures suggest contributions from both Ni d_x²-y² and d_z² orbitals at the Fermi level. Density functional theory (DFT) + U calculations capture qualitative features of the ARPES results, consistent with a hole-doped scenario. These findings constrain theoretical models of the superconducting mechanism and suggest potential for enhancing superconductivity in nickelates under ambient pressure.

The long-sought electronic band structure of nickelate high-temperature superconductor has been revealed through angle-resolved photoemission spectroscopy, enabled by atomically flat thin film growth of the complex Ruddlesden-Popper bilayers.

INTRODUCTION

High-temperature superconductivity has been a central topic of condensed matter physics for decades [1–3], particularly within the cuprate family, where the superconducting mechanism is predominantly associated with the d_x²-y² single-band Fermi surface observed by angle-resolved photoemission spectroscopy (ARPES) [4–6]. Although Ni and Cu are distinct elements, infinite-layer nickelates, a family of superconductors discovered in 2019, exhibit a lattice structure analogous to that of cuprates and share a similar nominal 3d⁹ electron occupancy [7–9]. ARPES studies of infinite-layer nickelates have confirmed that, despite the presence of rare-earth 5d orbitals crossing the Fermi level, the electronic structure is fundamentally governed by d_x²-y² bands, again similar to the cuprates [10,11]. In contrast, bilayer nickelates, discovered in 2023 to be superconducting under high pressure over 14 GPa with transition temperature near 80 K, while also comprising Ni, exhibit an average 3d electron occupancy of ∼7.5, markedly different from the 3d⁹ configuration in cuprates [12–16]. Furthermore, bilayer nickelates feature apical oxygen atoms between the superconducting Ni-O layers, a structural characteristic absent in cuprates. This apical oxygen induces coupling between the two Ni-O layers via the d_z² orbitals, prompting theoretical debates regarding the precise role of these orbitals in the superconducting mechanism [17–34]. Critical aspects of these debates involve whether the d_z²-dominant bands intersect the Fermi level and whether the superconducting state is hole-doped or electron-doped relative to the non-superconducting ambient-pressure parent. ARPES measurements on non-superconducting bulk bilayer nickelates at ambient pressure demonstrated that the d_z² bands lie below the Fermi level [35,36]. However, the high-pressure condition that is necessary for superconductivity is incompatible with photoemission experimental techniques, and the presence of mixed phases further complicates the analysis [37–40], leaving the electronic structure of the superconducting state undetermined.

Recently, ambient-pressure superconducting bilayer nickelate ultrathin films have been reported [41,42], offering an opportunity for investigating their electronic structures. Unlike bulk superconducting samples subjected to near-isotropic external pressure, the thin films experience biaxial epitaxial strain from the substrate, resulting in an in-plane lattice compression and out-of-plane lattice elongation. This distinct lattice deformation may lead to variations in electronic structures, differing from the assumptions of many existing theoretical and computational studies that are based on high-pressure, isotropic compression [43,44]. As such, the strained bilayer nickelate ultrathin films provide a unique platform for exploring high-temperature superconducting mechanisms. In this work, we have performed ARPES measurements on these recently discovered bilayer nickelate superconducting ultrathin films, providing initial evidence to the open questions listed above.

RESULTS

Figure 1 shows the magnetotransport measurements and ARPES Fermi surface maps of one unit-cell (1UC), 2UC and 3UC La_2.85Pr_0.15Ni₂O₇ films on treated (001)-oriented SrLaAlO₄ substrates, epitaxially grown and annealed under the same conditions. Samples with all different thicknesses exhibit superconducting transition in transport measurements (Fig. 1a–f), though with notable thickness-dependent variations. Degradation of superconducting properties in 2UC and 1UC samples is attributed to the inevitable oxygen loss during the relatively slow cooling process (300 K to 200 K, cooling rate limited by the cryogen-free cryostat) under vacuum. This oxygen loss is increasingly significant with reduced film thickness. With the existence of interfacial scattering (e.g. step terraces from substrate miscut angles and interdiffusions), the reduced dimensionality leads to disproportionately severe degradation of macroscopic transport characteristics in the monolayer limit. Nonetheless, these transport measurements confirm that the ARPES measurement conducted at 10 K reflect the superconducting state across 1UC to 3UC samples. For ARPES measurements, samples were transported in a 0.2-MPa pure oxygen suitcase from the growth chamber to the synchrotron end station, and then were rapidly quenched to <100 K on the manipulator (temperature constantly set to 10 K). This protocol minimizes oxygen loss compared to transport measurement procedures, which is particularly critical for the ultrathin 1UC sample.

Figure 1.

Fermi surface maps of La_2.85Pr_0.15Ni₂O₇/SrLaAlO₄ heterostructures. (a–c) Structural schematics of La_2.85Pr_0.15Ni₂O₇ films with thickness of 1 unit cell (1UC), 2UC and 3UC, respectively, grown on treated SrLaAlO₄ substrates. (d–f) Magnetotransport measurements in 1UC, 2UC and 3UC films, with insets zoomed in on the onset region. Arrows indicate the onset temperatures of magnetic field responses. Note that surface oxygen loss during cooling down in vacuum causes degradation of superconductivity in 1UC and 2UC samples. To reduce oxygen loss, the 1UC sample for transport was capped with around half nm SrTiO₃ after growth, while 2UC and 3UC samples are measured without capping. (g–i) Fermi surface maps of 1, 2 and 3UC thin films measured via angle-resolved photoemission spectroscopy (ARPES), using linear horizontally polarized (LH) photons with an energy of 200 eV. The red solid squares denote the first Brillouin zone (BZ) with high symmetry points Γ, M and X marked in panel g. The regions enclosed by the red dashed lines represent the second BZ. (j) Experimental geometry. The mirror plane is parallel to the (11) direction of the sample surface. (k) Momentum distribution curves (MDCs) at the Fermi level for the 1, 2 and 3UC thin films along cut 1 (shown in panel (g)). Black solid curve is the same MDC extracted from Ref. [35]. Vertical dashed lines are guides for the eye. (l) Schematic Fermi surface maps comparing this work and Ref. [35].

Utilizing linear horizontally (LH) polarized photons of 200 eV, the Fermi surface maps exhibit similar topologies across the three heterostructures, while the 2UC and 3UC samples display more smeared features compared to that of the 1UC sample (Fig. 1g–i, constant energy contours in Fig. S2). The experimental geometry is defined in Fig. 1j. The LH polarization maximize the capability of probing photoelectrons from different electron orbitals (Fig. S1). By extracting the momentum distribution curves (MDCs) along MXM (cut 1 in Fig. 1g) for the three films (Fig. 1k), we observe that the peak separations are similar across the films, suggesting the absence of significant chemical potential shifts across samples with varying film thickness. Notably, these peak separations are smaller than that of the extracted MDC from bulk La₃Ni₂O₇ non-superconducting compound at ambient pressure along the same direction, indicating hole doping in the superconducting films (Fig. 1l). Weak intensity is observed at the Γ point, suggesting that the Fermi level may be nearly touching a band bottom at this location. Moreover, the diffuse yet pronounced intensity around the M point suggests the presence of an additional band. The diffuse appearance of these features may stem from the underlying nature of this band, which will be discussed later.

Before presenting detailed band structure measurements and analysis, we first investigate the origin of smearing for 2UC and 3UC samples’ Fermi surface measurements, by characterizing the electronic structures across the film thickness. The integrated energy distribution curves (EDCs) obtained from 200 eV synchrotron photons (Fig. 2a) and 21.2 eV Helium-I photons (Fig. 2b) are compared. The photoelectron mean free path with 21.2 eV incident photons is significantly smaller than that with 200 eV photons, resulting in a much shorter probing depth and much stronger surface sensitivity. While all three films (1UC, 2UC and 3UC) exhibit metallic behavior, as indicated by integrated EDCs crossing the Fermi level under 200 eV illumination, distinct features are observed for each film when measured with 21.2 eV photons: in contrast to the 1UC sample, which retains a significant amount of electron density of states near the Fermi level, the 2UC sample shows a substantial suppression of intensity in this region, and the 3UC sample exhibits an absence of intensity near the Fermi level, suggesting an insulating-like behavior (intensities are normalized at higher binding energies, see Fig. 2b inset). The thin insulating layer allows any excess charge to tunnel away efficiently, thus no obvious charging effect is observed (Fig. S3). A plausible scenario involves a pronounced conductivity gradient across the film thickness, as shown in Fig. 2c, where the 1st UC near the interface is highly conductive, while the 3rd UC is almost insulating.

Figure 2.

Conductivity gradient across the film thickness. (a, b) Momentum-integrated energy distribution curves (EDCs) near the Fermi level of 1, 2 and 3 UC samples using 200 eV synchrotron photons and 21.2 eV Helium-Ⅰ photons, respectively. Insets show larger energy windows. (c) Schematics of the possible different photoemission processes across the film thickness under 200 eV and 21.2 eV photons. (d) High-angle annular dark-field (HAADF) image overlayed with atomically resolved energy-dispersive X-ray spectroscopy (EDS, for Ni, Al and Sr, respectively) images of a 3UC superconducting sample. The yellow curve is the integration of intensity in horizontal pixels for Sr EDS. White dashed line represents the interface. White dotted lines separate different unit cells. White boxes are guides to the eye for the stacking structure of the 327 phase. (e) Electron energy loss spectroscopy (EELS) of the oxygen K edge, obtained by integrating signals within areas in the substrate and each different UC across a 3UC superconducting sample. The black squared brackets in panel (d) approximately represent the integration windows. Black arrow points to the pre-peak feature. Note that additional oxygen loss may be introduced during STEM sample preparation. (f) The pre-peak intensity of the curves in panel (e), integrated from 525 eV to 530 eV. The dark grey solid curve is a schematic profile representing the interfacial confinement.

The origin of such a significant conductivity gradient may be elucidated through scanning transmission electron microscopy (STEM) techniques (Fig. 2d and insets, full-range image in Fig. S4). Atomically resolved energy-dispersive X-ray spectroscopy (EDS) reveals Sr diffusing across the interface with the diffusion length ∼1UC. Since Sr has a +2 valence, its substitution for the +3 valence La (or Pr) induces hole doping. This implies that Sr concentrations, on top of the substrate epitaxial compressive strains, may help the emergence of superconductivity. Additionally, electron energy loss spectroscopy (EELS) analysis of the oxygen K edge pre-peak intensity, a signature of p-d hybridization [45,46] (Fig. 2e), exhibits a marked gradient across the film thickness (Fig. 2f). One possible scenario is that oxygen loss at the surface induces insulating behavior at the top layer [41,42], and the resulting oxygen content variations throughout the film leads to the observed p-d hybridization gradient [34]. The existence of top layers may have protected the 1st UC from significant oxygen depletion. Both Sr diffusion, which may induce hole doping at the interface, and oxygen loss, which may cause reduced p-d hybridization at the surface, may collectively contribute to the confinement of superconductivity near the interface in the measured La_2.85Pr_0.15Ni₂O₇/SrLaAlO₄ system. In this context, improvement of experimental technique is needed to control and maintain oxygen content across the films during the sample growth, storage, and transfer, thereby enhancing superconducting properties, particularly in the top layers. Given the similar Fermi surface topologies among 1UC, 2UC and 3UC samples and the higher spectroscopic quality of the 1UC, we focus mostly on the 1UC below for band structure analysis.

Fig 3a displays the Fermi surface maps of the 1UC film in LV polarization (constant energy contours in Fig. S5). Due to the matrix element effect (Fig. S1) of ARPES, these maps clearly differ from those in LH polarization as shown in Fig. 1. Fig. 3b exhibits the summed Fermi surface intensity obtained from both LH and LV polarized photons, which contains more comprehensive orbital information. We reproduce a schematic of the orbital-characterized Fermi surface map [6,47] in Fig. 3c, showing contributions from both d_x²-y² and d_z² orbitals (denoted as α, β, and γ pockets). Based on area integrals of this Fermi surface map, we estimate a hole doping level of 21% ± 5% (i.e. 0.21 holes per Ni) relative to that of the non-superconducting parent bulk, assuming the two-dimensional nature of the electronic structure [35] (reproduced in Fig. 1i). The second derivative spectra exhibit band dispersions for both β and γ pockets crossing the Fermi level in the first and second Brillouin zones (Fig. 3d and e). The quantitative determination of the band dispersions near the Fermi level (Fig. 3f) can be obtained from Fig.  3g–l. The small peak in the Fig. 3j MDC at the Fermi level near the Γ point may be corresponding to the same weak intensity observed in the Fermi surface map (Fig. 1d), implying the existence of a band bottom right above Fermi level. The band dispersions (Fig. 3g and h) and the corresponding MDCs (Fig. 3j and k) confirm the presence of the γ pocket around the M point crossing the Fermi level.

Figure 3.

Band dispersions. (a) Fermi surface maps of 1UC sample measured by 200 eV LV polarized photons. The red arrows indicate the position of γ pocket dominated by d_z² orbital, which is more clearly resolved in the dispersions. (b) Fermi surface map of the 1UC sample summing intensity obtained from LH and LV photons. The black arrow indicates the small α pocket from d_x²-y² orbital which is clearer in the second BZ. Note that in the deflection mode of the hemispheric photoelectron analyzer, polarization analysis deviates from the ideal case at higher angles (e.g. out of the first Brillouin zone). (c) Schematic of experimentally resolved Fermi surface map with α, β, and γ pockets different orbitals characterizations. The light blue lines represent α and β pockets dominated by Ni 3d_x²-y² orbital and the bold red lines represent the γ pocket dominated by Ni 3d_z² orbital. (d, e) Second derivative spectra along cuts 1 and 2 in panel (c) with LH photons overlapped with dashed guidelines, respectively. (f) Schematic of experimentally resolved band dispersion with different orbitals characterized. (g, h) ARPES spectra along cuts 1 and 2 in panel (c) with LH photons, respectively. (i) ARPES spectrum along cut 3 with LV photons extracted from the deflection-mode Fermi surface map as shown in panel (a). Yellow curves are MDCs at the Fermi level. (j, k, l) MDC stacks of spectra from panel (g, h) and (i), respectively. The color bar shows the ARPES spectra intensity. Blue and red dots mark the position of fitted Lorentzian peaks in MDCs, corresponding to β and γ bands, respectively. Dashed lines are extended guides for the eye.

Density functional theory (DFT) + U calculations of a nickelate bilayer based on strained thin film lattice structures [42] without Sr substitution are presented in Fig. 4. The larger energy window in Fig. 4a displays the distribution of the d_x²-y² and d_z² orbital related bands. Figure 4b is a zoom-in version of the same set of data, deliberately plotted with the similar energy window and high-symmetry path as in Fig. 3f. Note that the Fermi level is calibrated for comparison with the experiment, based on the proximity of a band minimum to the Fermi level at the Γ point. The Fermi level for the undoped case is also shown for reference. The calculated band dispersions qualitatively capture the ARPES results. The quantitative differences hint at the need for strongly correlated algorithms [48]. The diffuse manifestation of the d_z² Fermi surface observed in ARPES may also arise from electron correlations in this d_z² band, which features a partially flat region above the Fermi level in DFT calculations. Fig. 4c and d presents the Fermi surfaces for undoped and hole-doped cases, with the latter qualitatively agreeing with experiments (Fig. 3e). Figure 4e is the associated orbital dependent density of states.

Figure 4.

DFT + U calculations. (a) Band structures of La₃Ni₂O₇ with larger energy window, calculated based on thin film lattice parameters [42]. The red (blue) dots demonstrate the projected weights for the orbital d_z² (d_x²-y²) of Ni atoms. The horizontal grey dashed line indicates the undoped case, while the horizontal black dashed line denotes the hole-doped case corresponding to experiments. (b) The same data with magnified energy window and high symmetry path similar to Fig. 3i. (c, d) The Fermi surfaces of undoped and hole-doped cases, respectively. (e) Orbital-dependent partial density of states.

CONCLUSION

In summary, angle-resolved photoemission spectroscopy measurements indicate that the superconducting state in La_2.85Pr_0.15Ni₂O₇/SrLaAlO₄ heterostructures is hole-doped, with contributions from both d_x²-y² and d_z² orbitals at the Fermi level. Photoemission spectroscopy, together with atomically resolved energy-dispersive X-ray spectroscopy and electron energy loss spectroscopy, indicates a combined effect of interfacial Sr diffusions and surface suppressions of p-d hybridization, confining conduction near the interface. These findings provide an initial basis for further exploration of the superconducting mechanisms in bilayer nickelates and offer insights into strategies for enhancing their superconducting properties.

METHODS

Samples

High quality La_2.85Pr_0.15Ni₂O₇ thin films with different thickness were grown using the Gigantic-Oxidative Atomically Layer-by-Layer Epitaxy (GOALL-Epitaxy) method [49], followed by a 30-minute post annealing at 575C within the growth chamber. The SrLaAlO₄ (001) substrates (MTI, China) were annealed by placing a LaAlO₃ substrate face to face with a LaSrAlO₄ substrate under atmospheric conditions at 1030°C or 1080°C for 2 hours. Platinum electrodes were deposited on the substates/film before ozone annealing for grounding. The grown samples were transferred to a synchrotron beamline station and helium-lamp ARPES station in suitcases, filled with oxygen at ∼0.2 MPa. The purity of the oxygen was 99.999% and the base-pressure of the suitcases were maintained below 1 × 10⁻⁸ mbar prior to filling with oxygen.

ARPES measurements

Synchrotron-ARPES measurements were performed at beamline BL09U of Shanghai Synchrotron Radiation Facility (SSRF) in China and the overall energy resolution was set to be better than 20 meV at 200 eV photon energy while the angular resolution was ∼0.2 degree for measurements. Base pressure of the beamline station is better than 4 × 10⁻¹¹ Torr. Helium-lamp (21.2 eV) ARPES measurements were performed at a lab-based ARPES system with a base pressure better than 5 × 10⁻¹¹ Torr. After testing various photon energies from different light sources, 200 eV was found to provide optimal resolution for band dispersions. During ARPES measurements, the remaining samples were kept under oxygen environment at ∼0.2 MPa. Sample transfers in ultrahigh vacuum were conducted as quickly as possible to minimize surface oxygen loss. Measurement temperature was ∼10 K.

Band structure calculations

A simple thin film crystal structure was constructed considering only half unit-cell of La₃Ni₂O₇ using the lattice constants. Ni-O-Ni angles and distances and La-La distance were determined experimentally [42]. The actual length of vacuum in the calculation was >30 Å. The positions of oxygen atoms were relaxed while fixing the lattice constants and positions of La and Ni atoms using the VASP package [50–52]. The band structures were calculated using the DFT + U method with preliminary parameters U = 5 eV and J = 1 eV. The tight-binding (TB) model Hamiltonian for the bands near the Fermi energy was obtained from the maximally-localized Wannier function method implemented in the package of wannier90 [53,54]. A 125 × 125 k-point mesh grid in the first Brillouin zone was used to calculate the iso-energy surfaces for different binding energies based on the TB model Hamiltonian.

Scanning transmission electron microcopy (STEM)

Specimens for the cross-section STEM were prepared utilizing the focused ion beam (FIB) technique on a FEI Helios G4 HX dual-beam FIB/SEM machine. The HAADF images were photographed using a double spherical-aberration corrected FEI Themis Z system at 200 kV, with a high-brightness field-emission gun (X-FEG) equipped with a monochromator installed onto this microscope. The probe semi-convergence angle was 25 mrad and the collection angles for the STEM images (β1 and β2) were 90 and 200 mrad, respectively. The EDS mappings were obtained on a Super X FEI System in STEM mode. The beam currents of 40 and 136 pA were used for HAADF imaging and EDS analyses, respectively. EELS spectra of the O-K edge were performed on a FEI Themis Z equipped with a high stability stage, a Gatan Quantum ER/965 spectrometer, operated at 200 kV with 200 pA beam current. The probe currents were controlled at <200 pA to minimize radiation damage with the scan parameters used: ∼1-Å probe, 9-Ų scan pixel size (applied 16 × 16 sub-pixel scan), and 300 µs per pixel dwell time. Before extraction and thinning, platinum and carbon layers were deposited by electron beam and ion beam, respectively, to prevent ion beam damage of the sample surface. All operations were done at room temperature.

Transport measurements

Electric transport measurements were performed in a closed-cycle helium-free cryostat with a base temperature of ∼1.8 K and a magnetic field of up to 14 T. Platinum Hall bar electrodes were fabricated on 5 × 5 mm² samples using magnetron sputtering through a pre-patterned hard shadow mask. Prior to sample loading, the sample chamber was cooled to 275 K, followed by helium gas purging to remove residual air. The final pressure in the chamber was maintained at ∼10 Torr. To minimize oxygen loss, the sample was cooled below 200 K in 10 minutes, to minimize the time under conditions that leads to oxygen loss.

FUNDING

This work was supported by the National Key R&D Program of China (2024YFA1408101 and 2022YFA1403101), the National Natural Science Foundation of China (92265112, 12374455 and 52388201), the Guangdong Provincial Quantum Science Strategic Initiative (GDZX2401004 and GDZX2201001), the Shenzhen Science and Technology Program (KQTD20240729102026004) and the Shenzhen Municipal Funding Co-Construction Program Project (SZZX2301004 and SZZX2401001). Yueying Li acknowledges support by the China Postdoctoral Science Foundation (2024M761276). Changming Yue acknowledges support by the National Natural Science Foundation of China (12474231) and Guangdong Provincial Key Laboratory of Advanced Thermoelectric Materials and Device Physics (2024B1212010001). Jianchang Shen, Yu Miao and Junfeng He acknowledge support by the National Key R&D Program of China (2024YFA1408100), the International Partnership Program of the Chinese Academy of Sciences (123GJHZ2022035MI) and the Innovation Program for Quantum Science and Technology (2021ZD0302802). Tian Qian acknowledges support by the National Key R&D Program of China (2022YFA1403800) and the National Natural Science Foundation of China (U22A6005).

AUTHOR CONTRIBUTIONS

Q.K.X. supervised the entire project. Z.C. initiated the study and coordinated all the research efforts. P.L. and Y.L. led synchrotron-ARPES measurements, with assistance from L.X., J.S., Y.M. and W.S., and under supervision from Z.C., J.H. and Y.J.S. G.Z. and W.L. prepared samples customized for ARPES measurements, with assistance from Z.N. and Y.C. C.Y. performed DFT + U calculations. H.H. and J.L. provided STEM, EDS and EELS analysis. J.S. and Y.M. performed helium-lamp–ARPES measurements under supervision from J.H. H.W. conducted oxygen loss analysis. W.C. provided theoretical support. Y.H. and Z.H.C. supported ARPES measurements at the beamline. T.Q. provided beamline resources and participated in discussions. Z.C., P.L. and Y.L. wrote the manuscript with input from C.Y., J.H., Y.J.S. and all other authors.

Conflict of interest statement. None declared.