Superconductivity in infinite-layer nickelate La1-xCaxNiO2 thin films.

We report the observation of superconductivity in infinite-layer Ca-doped LaNiO2 (La1-xCaxNiO2) thin films and construct their phase diagram. Unlike the metal-insulator transition in Nd- and Pr-based nickelates, the undoped and underdoped La1-xCaxNiO2 thin films are entirely insulating from 300 K down to 2 K. A superconducting dome is observed at 0.15 < x < 0.3 with weakly insulating behavior at the overdoped regime. Moreover, the sign of the Hall coefficient RH changes at low temperature for samples with a higher doping level. However, distinct from the Nd- and Pr-based nickelates, the RH-sign-change temperature remains at around 35 K as the doping increases, which begs further theoretical and experimental investigation to reveal the role of the 4f orbital to the (multi)band nature of the superconducting nickelates. Our results also emphasize a notable role of lattice correlation on the multiband structures of the infinite-layer nickelates.

INTRODUCTION

The resemblance of the high critical temperature (high-Tc) superconducting cuprates in the nickel-based compounds has made superconductivity in the nickelates to be one of the holy grails in superconducting research in the past three decades. The recent discovery of superconductivity in infinite-layer Sr-doped NdNiO2 () has now further propelled the importance of nickelates as a route toward understanding and a material platform for searching for new high-Tc superconductors (–). The perovskite nickelates RNiO3 (R = rare-earth element) show a structural transition and exhibit a metal-insulator transition upon cooling when R = Nd and Pr, while it remains metallic when R = La (). To mimic the cuprate-like 3d9 electronic configuration, the infinite-layer LaNiO2 is required and can be realized by oxygen deintercalation of the LaNiO3 precursor (). Much effort has been devoted to dope carriers in this LaNiO2, but it failed to obtain superconductivity (). To increase the electronic bandwidth, the La was substituted by a rare-earth element with a smaller ionic radius (e.g., Nd atom), resulting in the observation of superconductivity in Nd0.8Sr0.2NiO2 (). This route to superconductivity contrasts the heavily studied copper- and iron-based superconductors in which superconductivity was first obtained in La-based compounds (, ), while up to now, the superconducting nickelate has only been observed in Sr-doped compounds with R = Nd or Pr (i.e., R1−SrNiO2) (, –). Moreover, the trilayer nickelate R4Ni3O8 (R = La and Pr) that has the same NiO2 square plane as in the infinite-layer nickelates and an effective one-third hole doping showed an insulating ground state in the case of La4Ni3O8, in contrast to the metallic state in the Pr4Ni3O8 (). Theoretical calculation attributed the absence of superconductivity to the intercalation of H atoms in energetically favorable LaNiO2 to form LaNiO2H during the reduction process (). It has also been shown that hybridization with Nd 4f orbitals is a non-negligible ingredient for the superconductivity in doped NdNiO2, while the f orbitals are far from the Fermi level in LaNiO2 (–). On the other hand, while earlier density functional theory (DFT) studies for NdNiO2 treated the 4f orbitals as a “frozen core” under the assumption that the 4f orbitals are localized and do not show around the Fermi level (, ), a recent DFT combined with dynamical mean-field theory calculation comparing nickelates with and without 4f orbitals also concluded that 4f orbitals are not essential to superconductivity in nickelates (). This suggests the importance of elucidating the role of 4f state of the electronic structure in the occurrence of superconductivity in nickelates (–). Therefore, the realization of superconductivity in the La series is a key step to understanding the superconducting properties in the nickelates.

In unconventional superconductors containing the rare-earth element, the substitution of an isovalent ion with a smaller radius causes an increase in inner chemical pressure due to the shrinkage of the crystal lattice. This is considered an important method to enhance Tc, as experimentally demonstrated in cuprates and pnictides (, ). In the Nd- and Pr-based nickelates, the Hall coefficient RH–sign-change temperature varies with Sr doping level, indicating that the Sr doping not only introduces charges but also modulates the band structures (, , , , –). Moreover, the lattice constant increases with increasing Sr doping level, and the superconducting dome occurs near the RH-sign-change boundary. It is therefore anticipated to induce superconductivity in LaNiO2 through doping with a smaller ionic radius. Here, we present the observation of superconductivity in infinite-layer La1−CaNiO2 thin films through Ca doping.

RESULTS

Doping-dependent structural properties

The perovskite La1−CaNiO3 precursor thin films were synthesized by a pulsed laser deposition (PLD) technique and reduced to an infinite-layer phase using the soft-chemistry topotactic reduction method. The x-ray diffraction (XRD) θ-2θ patterns of the as-grown La1−CaNiO3 thin films can be found in fig. S1. The prominent thickness oscillations in the vicinity of the (002) peak indicate the single phase and high quality of the perovskite films. After reduction, a clear transition of the diffraction peaks is seen, confirming the transformation from the perovskite to the infinite-layer structure, as shown in Fig. 1A for a representative Ca doping level of 0.23. Figure 1B shows a high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image of the 17-nm La0.77Ca0.23NiO2 thin film. A clear infinite-layer structure is observed with no obvious defect throughout the layer. Note that in Nd1−SrNiO2, the pure perovskite and the resultant infinite-layer structure can only survive up to ~10 nm, above which the Ruddlesden-Popper–type phase emerges (, , ). For the La-nickelate thin films, the pure infinite-layer phase is still observed for a thickness of 17 nm (Fig. 1B and fig. S1). We propose two possible explanations: (i) The lattice match between the La1−CaNiO3 and SrTiO3 substrates is much smaller than for the Nd1−SrNiO3, thus allowing better growth of the parent perovskite compound. (ii) A more similar ionic size between La3+ and the dopant ion Ca2+, as compared to Nd3+ versus Sr2+. While the observation of a stable infinite-layer phase at a larger thickness in La1−CaNiO2 is still preliminary, it offers an alternative route for realizing bulk-like superconducting nickelates. In addition, while the La1−CaNiO2 was only recently studied, we notice high reproducibility in realizing superconductivity in the system as compared to Nd1−SrNiO2. Figure 1C shows the XRD θ-2θ patterns of infinite-layer La1−CaNiO2 thin films with different Ca doping levels x from 0 to 0.35. The (00l) peak positions slightly shift toward a higher angle as Ca dopant increases, indicating a shrinking of the c-axis lattice constants d from ~3.405 Å at x = 0 to ~3.368 Å at x = 0.35 as plotted in Fig. 1D. The evolution of d is in agreement with the empirical expectation as the cation is replaced with an atom having a smaller ionic radius. The reciprocal space mappings around the (−103) diffraction peak indicate that the films are slightly relaxed, showing the larger in-plane lattice constants compared with that of the SrTiO3 substrate (Fig. 1D and fig. S2). The full width at half-maximum of the rocking curves for the (002) peaks of the infinite-layer films shows a value between 0.06o and 0.12o, supporting the high-quality thin films after reduction (fig. S3).

Fig. 1.

Structural characterization of infinite-layer La1−CaNiO2 thin films.

(A) The XRD θ-2θ intensity scan in arbitrary unit (a.u.) of the perovskite La0.77Ca0.23NiO3 and infinite-layer La0.77Ca0.23NiO2 thin films on a SrTiO3 (STO) substrate. The arrow denotes the transition of diffraction peaks related to the transformation from a perovskite to an infinite-layer structure. (B) The HAADF-STEM image of the 17-nm La0.77Ca0.23NiO2 on SrTiO3 substrate. (C) The XRD θ-2θ scan patterns of the La1−CaNiO2 thin films with different Ca doping levels x. The intensity is vertically displaced for clarity. (D) The room-temperature c-axis lattice constants, d, and in-plane lattice constant, a, as a function of x.

Electronic properties

Figure 2A shows the logarithmic-scale resistivity versus temperature (ρ-T) curves of the La1−CaNiO2 thin films with Ca doping level x from 0 to 0.35. For x ≤ 0.15, the samples show insulating behavior all the way below 300 K. This is different from Nd1−SrNiO2 and Pr1−SrNiO2 thin films, in which the undoped and underdoped samples show a metallic behavior at high temperatures with a resistivity minimum at the intermediate temperature below which a weakly insulating behavior appears (, –, ). The high-temperature metallic behavior in undoped Nd and Pr nickelates has been thought to be due to the self-doped nature of the parent compound (, , , ); however, theoretical simulations have not arrived at a dissimilar conclusion for La nickelates (). The zoomed-in and linear-scale ρ-T curves of the superconducting La1−CaNiO2 thin films with x from 0.15 to 0.27 are shown in Fig. 2B. For x = 0.15 and 0.18, the samples show transitions from a slightly metallic behavior to an insulating behavior and then an onset of superconducting transition with decreasing temperature, while for 0.2 ≤ x ≤ 0.27, the samples are superconducting with the ρ-T curves showing metallic behavior at high temperatures. The suppression of superconductivity with increasing out-of-plane magnetic field for a representative sample with x = 0.23 is shown in fig. S4. The fitting of the relationship between the upper critical field and midpoint critical temperature by the Ginzburg-Landau model gives the zero-temperature in-plane coherence length of 4.59 nm, comparable to that of Nd0.8Sr0.2NiO2 film (). At the overdoped regime with x ≥ 0.3, the samples show an increase in normal state resistivity compared with that of the optimally doped sample. Moreover, similar to the Nd and Pr nickelates, the overdoped La1−CaNiO2 thin films exhibit weakly insulating behavior at low temperature, suggesting a universal transport property in overdoped infinite-layer nickelates (, , ).

Fig. 2.

Electrical transport properties of infinite-layer La1−CaNiO2 thin films.

(A) The logarithmic-scale resistivity versus temperature (ρ-T) curves of the La1−CaNiO2 thin films with Ca doping level x from 0 to 0.35, measured from 300 to 2 K. (B) The zoomed-in and linear-scale ρ-T curves of the La1−CaNiO2 thin films with x from 0.15 to 0.27. For clarity, the resistivity of samples with x = 0.15, 0.18, and 0.27 is diminished to 0.3 of the initial value. (C) The temperature dependence of the normal-state Hall coefficients RH. (D) The RH at T = 300 and 20 K as a function of x. The dash lines are guides to the eye.

Similar to Nd and Pr nickelates, the room temperature RH is negative, and its magnitude decreases with increasing x and then saturates at x = 0.2 (Fig. 2, C and D), suggesting the multiband structure nature for La1−CaNiO2 films (, , , , , –). The RH remains negative below 300 K for samples with x ≤ 0.2. For samples with x ≥ 0.23 (except for x = 0.25), the RH undergoes a smooth transition from negative to positive sign as the temperature decreases. Figure 2D presents the RH at 20 K, clearly showing a sign change at x = 0.2 to 0.23 from negative to positive with increasing x. The doping level of the RH sign change is higher than that (x = 0.18 to 0.2) of Nd and Pr nickelates (, , ). While we notice that the RH-sign-change doping level for both La nickelates and Nd/Pr nickelates are at around the optimally doped regime, in Nd1−SrNiO2 and Pr1−SrNiO2, the RH-sign-change temperature increases with increasing doping level (, , ). However, for La1−CaNiO2 films, the RH-sign-change temperature does not change with Ca doping level, and it is at around 35 K for x ≥ 0.23. Such distinct saturation may suggest that hole doping in the superconducting lanthanide nickelates plays a more hidden role than merely charge carrier modulation.

Phase diagram and superconducting dome

Figure 3 depicts the phase diagrams of La1−CaNiO2 integrated with those of Nd1−SrNiO2 and Pr1−SrNiO2 (, , ). The onset of the critical temperature, Tc,90 % , is defined as the temperature at which the resistivity drops to 90% of the normal-state value at the onset of the superconductivity. A superconducting dome between 0.15 < x < 0.3 is seen for La1−CaNiO2. The doping range of the La1−CaNiO2 superconducting dome is comparable to that of Pr1−SrNiO2 and slightly wider than that of Nd1−SrNiO2 but is generally extended toward a higher doping level (Fig. 3B). At the two ends of the dome, the underdoped regime shows insulating behavior upon cooling, and the overdoped regime is weakly insulating at low temperature, different from the Nd and Pr nickelates in which both regimes show weakly insulating behavior (, , ). In general, the Tc of La1−CaNiO2 is lower than those of Nd and Pr series, possibly because of the large ionic radius of La, consistent with the experimental observation in Cu- and Fe-based superconductors that the Tc is enhanced as the ionic radius of the rare-earth element decreases (, ). This observation implies that the appearance of superconductivity in infinite-layer nickelates is not reliant on the 4f electrons of the rare-earth element and suggests that higher Tc in infinite-layer nickelates can be stabilized using smaller rare-earth ions.

Fig. 3.

Phase diagram of infinite-layer La1−CaNiO2 thin films and the comparison to Nd1−SrNiO2 and Pr1−SrNiO2 thin films.

(A) The critical temperature as a function of doping level x for La1−CaNiO2 in the present results. The Tc,90 % is defined as the temperature at which the resistivity drops to 90% of the value at 10 K (the onset of the superconductivity). (B) The combined phase diagram of La1−CaNiO2, Nd1−SrNiO2, and Pr1−SrNiO2. The data of Nd1−SrNiO2 are adapted from (, ), and the data of Pr1−SrNiO2 are adapted from ().

DISCUSSION

In addition to introducing charge carriers, the chemical doping in infinite-layer nickelates can cause the change of the multiband structures because of the modification of the lattice environment (, –). Therefore, the normal-state properties and the superconducting range will depend on the lattice constant upon doping, which can be seen from the doping-dependent RH and superconducting dome (, , ). Depending on the difference of cation and dopant radii, the extent of the lattice-constant change upon doping is different. In Nd1−SrNiO2 (, ), the extent of the lattice-constant change with Sr doping is ~0.53 Å per one Sr atom, larger than that (~0.42 Å per one Sr atom) for Pr1−SrNiO2 () and that (~0.11 Å per one Ca atom) for La1−CaNiO2 that is observed in the current result. It is found that the superconducting dome of the Nd1−SrNiO2 is narrower than that of the Pr1−SrNiO2 and La1−CaNiO2, suggesting the intrinsic link between the superconducting range and the modulation of the lattice constant. Moreover, theoretical calculations suggested that Sr doping in Nd1−SrNiO2 reduces the self-doping effect, and the nickelate behaves more like a system with a pure single-band picture (, ). This is consistent with the RH sign change upon doping, and the sign-change temperature monotonically increases with increasing Sr doping level in Nd and Pr nickelates (, , ). However, for La1−CaNiO2 films, the doping-dependent RH-sign-change temperature is negligible for x ≥ 0.23. Coupled with the small change in the lattice constant, this suggests the critical role of the dopant ionic nature on the electronic modulation in infinite-layer nickelates. For the infinite-layer bulk nickelates Nd1−SrNiO2 and Sm1−SrNiO2 in which superconductivity is absent, the c-lattice constant is smaller than those of superconducting thin films (–). Moreover, we also found that the strain effect induced electronic bandstructure modulation and the resultant change of RH in Nd0.8Sr0.2NiO2 (). At the same time as our report, superconductivity was also independently realized in (La, Sr)NiO2 (). These suggest the importance of the lattice environment on the superconductivity and normal-state properties in infinite-layer nickelates. The present result also suggests a new route for new superconductors and a unique way for tailoring normal-state properties through dopants in nickelate superconductors.

MATERIALS AND METHODS

We prepared and characterized our Ca-doped LaNiO2 samples following procedures previously described in ().

Sample preparation

The ceramic targets with nominal composition La1−CaNiO3 were prepared by the conventional solid-state reaction using high-purity La2O3 (99.999%; Sigma-Aldrich), NiO2 (99.99%; Sigma-Aldrich), and CaCO3 (99.995%; Sigma-Aldrich) powders as the starting materials. The mixed powders were sintered in air for 15 hours at 1150°, 1200°, and 1250°C, respectively, with thorough regrinding before each sintering. On the final sintering, the powder was pressed into a disk-shaped pellet. The perovskite thin films were grown on TiO2-terminated (001) SrTiO3 substrates using a PLD technique with a 248-nm KrF excimer laser. The deposition temperature and oxygen partial pressure PO2 for all samples were 600°C and 150 mtorr, respectively. The laser energy density on the target surface was set to be 1.8 J cm−2. After deposition, the samples were annealed for 10 min at 600°C and 150 mtorr and then cooled down to room temperature at a rate of 8°C/min. To obtain the infinite-layer structures, the as-grown films were then embedded with about 0.15 g of CaH2 powder and wrapped in aluminum foil and then placed into the PLD chamber for the reduction process. The wrapped sample was heated to 340 ° to 360°C at a rate of 25°C/min and kept for 80 min and then cooled down to room temperature at a rate of 25°C/min.

Electronic transport and structural characterization

The transport measurements were performed using a Quantum Design Physical Property Measurement System. The wire connection for the electrical transport measurement was made by Al ultrasonic wire bonding. The XRD measurement was done in the X-ray Diffraction and Development beamline at Singapore Synchrotron Light Source (SSLS) with an x-ray wavelength of λ = 1.5404 Å. The HAADF-STEM imaging was carried out at 200 kV with a JEOL ARM200F microscope, and the cross-sectional TEM specimens were prepared by a focused ion beam machine (FEI Versa 3D).