Spin and orbital excitations in undoped infinite layers: a comparison between superconducting PrNiO2 and insulating CaCuO2

TL;DR

This study investigates spin and orbital excitations in the nominally undoped infinite-layer nickelate PrNiO$_2$ and compares them to the insulating cuprate CaCuO$_2$ using momentum- and polarization-resolved RIXS. The authors find that in-plane magnetic exchange is smaller in PrNiO$_2$ ($J_1 \approx 46$ meV) than in CaCuO$_2$ ($J_1 \approx 82$ meV), while the out-of-plane coupling is similar, together indicating three-dimensional antiferromagnetic correlations in both materials. Orbital excitations ($dd$) largely agree with single-ion predictions, but the Ni $d_{xy}$ peak disperses with the opposite sign to Cu $d_{xy}$, attributed to different orbital superexchange pathways (NN-dominated in PrNiO$_2$ versus NNN-dominated in CaCuO$_2$). The work shows that infinite-layer nickelates share much of the cuprate spin and orbital physics, albeit at reduced energy scales and with self-doping-induced spectral continua, offering insights into the mechanisms that govern superconductivity in these systems.

Abstract

Infinite-layer nickelates are among the most promising cuprate-akin superconductors, although relevant differences from copper oxides have been reported. Here, we present momentum- and polarization-resolved RIXS measurements on chemically undoped, superconducting PrNiO2, and compare its magnetic and orbital excitations with those of the reference infinite layer cuprate CaCuO2. In PrNiO2, the in-plane magnetic exchange integrals are smaller than in CaCuO2, whereas the out-of-plane values are similar, indicating that both materials support a three-dimensional antiferromagnetic order. Orbital excitations, associated to the transitions within 3d states of the metal, are well reproduced within a single-ion model and display similar characteristics, except for the Ni-dxy peak which, besides lying at significantly lower energy, shows an opposite dispersion to that of Cu-dxy. This is interpreted as a consequence of orbital superexchange coupling between nearest neighbor sites, which drives the orbiton propagation. Our observations demonstrate that infinite layer cuprates and nickelates share most of the spin and orbital properties, despite their markedly different charge-transfer energy Delta.

Figures (6)

• Figure 1: Comparison of RIXS spectra on PNO and CCO. (a) Infinite-layer structure (ochre planes) of PNO, showing the local tetragonal coordination of the nickel atoms to the surrounding oxygens; CCO has the same structure, with Ni and Pr replaced by Cu and Ca respectively; the dark blue arrows on metal sites depict antiferromagnetic spin order. (b-c) RIXS spectra stacks for PNO and CCO respectively, along the $[H,0]$ cut of the Brillouin zone. (d-e) Momentum-dependent RIXS maps of CCO (d) and PNO (e) along the triple paths of the Brillouin zone shown in the (d) inset for the two samples. All (d-e) spectra have been normalized to the $dd$ energy integral along the interval $(1;3.5)$ eV.
• Figure 2: Analysis of RIXS spectra. (a) Fitting of PNO spectrum with $\pi$ incident polarization, as described in the main text. Data points were obtained as the sum of different spectra along $[H,0]$ with $H$ between 0.4 and 0.475 r.l.u. at incident energy 852.48 eV. (b) Same fit for CCO, with $H=0.465$ r.l.u.; a second Gaussian is added here (light green) representing the phonon overtone, while the rare-earth peak tail has been replaced by the bimagnon continuum (dark blue). Elastic, phonon and magnon colors are the same as in (a). (c-d) Same fittings as (a-b) respectively, but focusing on the orbital excitations area.
• Figure 3: Comparison of magnon energy and damping. (a) Fitting of dispersions of PNO and CCO with the Linear Spin Wave model fit. The definitions and values (in meV) of the four in-plane exchange coupling constants are displayed on the right peng2017influence. (b) Comparison of the damping coefficients of PNO and CCO with those extracted from analogous fittings on Bi2201 cuprate at different doping levels in Ref. peng2018dispersion (UD = underdoped, OP=optimally doped, OD=overdoped). Error bars in both panels represent 95% confidence intervals from the fittings.
• Figure 4: Comparison between polarimetric RIXS data and theory. (a-f) Polarization-resolved RIXS spectra for PNO (a–c) and CCO (d–f), measured under the geometry and polarization conditions specified into each panel. (g–l) Corresponding single-ion RIXS cross-section calculations for the same polarization and geometry conditions as in (a–f), respectively.
• Figure 5: Orbital excitation dispersion. (a) Momentum-resolved map in the $d_{xy}$ energy range of PNO; dots highlight the dispersion of the peak, while the continuous line represents the charge transfer model fitting (see text). Horizontal dashed lines quote the maximum and minimum energies of each branch, with their distance being reported. The followed BZ cuts are the same shown in the inset of Figure 1(d). (b) Stack of spectra of PNO in the whole $dd$ range, with vertical lines fixed at the minimum of the dispersion of the $d_{xy}$ and $d_{xz/yz}$ peak. (c-d) Same as (a-b), for CCO. Error bars in (a)-(c) represent 95% confidence intervals from the fittings.