Charge order and magnetism in NdNiO2 infinite layers

The quest for nickel-based, high-critical-temperature (T_c) superconductors dates back to the end of the twentieth century, when superconductivity was theoretically predicted in LaNiO₃/LaAlO₃ superlattices [1]. This result stimulated experimental activities aimed at mimicking the CuO₂ electronic structure of copper-based high-T_c superconductors. However, superconductivity was only recently demonstrated (with a T_c = 15K) in Nd_0.8Sr_0.2NiO₂ thin films deposited on the (001) surface of SrTiO₃ (STO) [2].  The main step to achieve this result was an oxygen de-intercalation of the pristine perovskite Nd_1−xSr_xNiO₃ phase via a topotactic reduction by using a CaH₂ powder as a reagent. This soft chemistry process leads to an infinite-layer structure with Ni¹+ ions characterised by a 3d^(9−x) electronic configuration in a square planar lattice, similar to that of hole-doped cuprates.

Several analogies were found between cuprates and infinite-layer nickelates, both theoretically and experimentally, including, recently, the observation of magnons in NdNiO₂ [3] capped with epitaxial STO, dispersing similarly to those of cuprates, but on a smaller energy range. However, electron energy loss spectroscopy, X-ray absorption spectroscopy (XAS) and resonant inelastic X-ray scattering (RIXS) experiments [4] also evidenced some crucial differences among infinite-layer nickelates and cuprates. In particular, a lower charge-transfer energy and a higher on-site potential places undoped infinite-layer nickelates in a region of the Zaanen–Sawatzky–Allen classification scheme, intermediate between Mott-Hubbard and charge-transfer insulators. Moreover, in infinite-layer nickelates, a sizeable contribution of the Nd5d bands at the Fermi level leads to an important hybridisation of Nd5d and Ni3d states [4], whereas in cuprates the inter-layer cations do not contribute to the electronic states relevant to transport.

In this work, Ni-L₃ edge high-energy-resolution RIXS at beamline ID32 was used to study electronic and magnetic excitations and charge density correlations in Nd_1−xSr_xNiO₂ thin films with and without an SrTiO₃ capping layer. Figure 1 summarises the main result:

 

Click figure to enlarge

Fig. 1: Summary of the RIXS results for uncapped and STO-capped NSNO(0). a,c) Energy-loss/in-plane-momentum scattering intensity maps along the high symmetry directions indicated in the insets, excited at incident photon energy 852.5 eV (Ni¹⁺ XAS peak) using p polarisation. b,d) Energy loss/excitation-energy maps across the Ni L₃ edge, at 10° grazing incidence. Lateral panels show sketches of the structure of (left) uncapped and (right) capped samples.

In uncapped NdNiO₂, when the photon energy is adjusted to the Ni¹⁺ X-ray absorption resonance (which probes Ni-3d⁹ related excitations in the NiO₂ planes), a modulation can be observed of the quasi-elastic peak intensity as a function of the in-plane momentum along the (010) and (100) directions (disappearing along the (110) direction), with a peak at wave vector (1/3,0) (Figure 1a), evocative of the charge order (CO) (or charge density wave, CDW) in hole-doped cuprates. The peak weakens at x= 0.05 and disappears in the superconducting x=0.20 film.

Surprisingly, STO-capped NdNiO₂ films do not show any CO/CDW, but well-dispersing magnon excitations (Figure 1c). The uncapped and capped samples also present a different degree of Ni3d-Nd5d hybridisation, identifiable in the RIXS map as a well-defined 3d⁸R feature at -0.6eV energy loss, and of anisotropy of the Ni3d occupation. In particular, uncapped NdNiO₂ shows much larger Ni3d-Nd5d hybridisation, and lower anisotropy with respect to capped samples.

Other reports of a CO in infinite-layer nickelates have also been published recently [5-6]. While details on the specific material studied and the results differed, these findings confirmed another analogy between the physics of cuprates and infinite-layer nickelates, namely, the instability toward the formation of a CO. However, they also raised different questions. In particular, a CO is not observed in undoped cuprates, showing that NdNi_O2 is somewhat special as it is actually self-doped even in the stoichiometric phase, probably due to the aforementioned Nd5d-Ni3d hybridisation, as suggested by theoretical calculations.

Furthermore, the differences between capped and uncapped samples, and the doping evolution, suggest a correlation between the degree of NdNiO₂ hybridisation and the stabilisation of a CO/CDW, i.e., stronger hybridisation corresponds to a larger tendency to form a CO/CDW. On the other hand, the CO/CDW instability clearly competes with magnetic excitations, related to (short-range) antiferromagnetic fluctuation of the 3dx2-y2 spins, in infinite-layer nickelates. Superconductivity and magnetic excitations are re-established when the role of the out-of-plane cation states is diminished, which can be controlled be doping and/or by the presence (or absence) of a capping layer. In summary, these findings highlight the role of the STO capping layer in the unique electronic properties of NdNiO₂.

Principal publication and authors
Charge and Spin Order Dichotomy in NdNiO₂ Driven by the Capping Layer, G. Krieger (a) L. Martinelli (b), S. Zeng (c), L.E. Chow (c), K. Kummer (d), R. Arpaia (e), M. Moretti Sala (b), N.B. Brookes (d), A. Ariando (c), N. Viart (a), M. Salluzzo (f), G. Ghiringhelli (b,g), D. Preziosi (a), Phys. Rev. Lett. 129, 027002 (2022); https://doi.org/10.1103/PhysRevLett.129.027002
(a) Université de Strasbourg, CNRS, IPCMS UMR 7504, Strasbourg (France)
(b) Dipartimento di Fisica, Politecnico di Milano, Milan (Italy)
(c) Department of Physics, Faculty of Science, National University of Singapore (Singapore)
(d) ESRF
(e) Quantum Device Physics Laboratory, Department of Microtechnology and Nanoscience,
Chalmers University of Technology, Goteborg (Sweden)
(f) CNR-SPIN Complesso di Monte S. Angelo, Naples (Italy)
(g) CNR-SPIN, Dipartimento di Fisica, Politecnico di Milano, Milan (Italy)

References
[1] V.I. Anisimov et al., Phys. Rev. B 59, 7901 (1999).
[2] D. Li et al., Nature 572, 624 (2019).
[3] H. Lu et al., Science 373, 213 (2021).
[4] M. Hepting et al., Nat. Mater. 19, 381 (2020).
[5] M. Rossi et al., Nat. Phys. 18, 869-873 (2022).
[6] C.C. Tam et al., Nat. Mater. 21, 1116-1120 (2022).