Comparative angle-resolved photoemission spectroscopy study of ${\mathrm{CaRuO}}_3$ and ${\mathrm{SrRuO}}_3$ thin films: Pronounced spectral weight transfer and possible precursor of lower Hubbard band

In the prototypical $4d$ system $(\mathrm{Sr},\mathrm{Ca}){\mathrm{RuO}}_3$, the degree and origin of electron correlations, and how they correlate with physical properties, still remain elusive, though extensive studies have been performed. In this work we present a comparative electronic structure study of high-quality epitaxial ${\mathrm{CaRuO}}_3$ and ${\mathrm{SrRuO}}_3$ thin films, by means of reactive molecular beam epitaxy and in situ angle-resolved photoemission spectroscopy. We found that while ${\mathrm{SrRuO}}_3$ possesses sharp features signaling the Fermi liquid state, the isostructural ${\mathrm{CaRuO}}_3$ exhibits broad features and its spectral weight is markedly transferred from the Fermi level to $-1.2$ eV forming a “hump” structure which resembles the Mott-Hubbard system $(\mathrm{Sr},\mathrm{Ca}){\mathrm{VO}}_3$. We suggest that this hump is the precursor of the lower Hubbard band, and the $U/W$ ($U$ and $W$ represent the on-site Coulomb interactions and bandwidth, respectively) of our ${\mathrm{CaRuO}}_3$ thin film is much larger than that of ${\mathrm{SrRuO}}_3$. In addition, we discuss the origin of electron correlations as well as the ferromagnetism in ${\mathrm{SrRuO}}_3$ which is absent in ${\mathrm{CaRuO}}_3$. Our findings put constraints on future studies, and also show that perovskite ruthenates are indeed an experimentally tunable system for the study of electron correlations.

Figure 1

Growth and characterization of CRO thin films. (a) Schematic of CRO thin films epitaxially grown on NGO substrates. (b) Typical RHEED patterns of CRO thin film (b1) and the NGO substrate (b2) along the (100)${}_p$ azimuthal. Sharp diffraction spot-streak structures and visible Kikuchi lines in the CRO RHEED pattern indicate its high quality with long-range ordered surface structure. (c) X-ray diffraction pattern around (001)${}_p$ peak for a CRO thin film. The red and black markers denote (001)${}_p$ peaks of the CRO thin film and the NGO substrate, respectively. The red dotted line is the good fitting curve which gives a thickness of 41 unit cells (UCs) and out-of-plane constant of 3.832 Å as a result of the tensile strain. (d) Temperature dependence of the resistances of this 41-UC-thick CRO film (i) and 21-UC-thick SRO film (ii). In (i), no signature of a phase transition in our CRO film was observed within the temperature range of our measurement; the inset shows that the low-temperature resistance has a linear dependence of $T^{3/2}$ ($T$ is short for temperature). In (ii), the red arrow marks the kink corresponding to the ferromagnetic transition of SRO; the inset shows that the low-temperature resistance has a linear dependence of $T^2$, which indicates the Fermi liquid ground state. The growth details of high-quality SRO thin films with well-controlled thicknesses can be found in our previous work [34].

Figure 2

Valence bands of CRO and SRO. (a) Valence band spectra of CRO and SRO. Their high-binding energy dispersions (below $-2.4$ eV) are similar. (b) Integrated valence bands of CRO (blue curve) and SRO (red curve). The normalization of these two curves was done within the energy window of $[-4.2$ eV, $-3.4$ eV] to clearly show their near-$E_F$ structures. In (i), near-$E_F$ parts were shown with the normalization within the energy window of $[-2.2$ eV, 0]. The inset is the schematic of the first Brillouin zone (corresponding to the pseudocubic substrate) with the cut #1.

Figure 3

Side-by-side comparison of representative long-cut dispersions of CRO and SRO along the cut #1. (a) Long-cut spectrum (down to $-2.5$ eV) of CRO. The hump structure enclosed by black dotted line is evidently shown to be located around $-1.2$ eV. (b) EDCs of the spectrum (a) exhibiting broad peaks. (c) Two representative EDCs cutting across the hump structure marked by the purple shadow. (d) Long-cut spectrum of SRO. No hump structure is observed [no evident spectral weight in the same black-dotted-line region as (a)]. The inset shows sharp band dispersions near $E_F$. (e) EDCs of the spectrum (d), exhibiting sharp peaks. The inset shows EDCs near $E_F$. (f) Two representative EDCs at the same positions as (c) showing no signatures of hump in SRO.

Figure 4

(a) Integrated EDCs of CRO (blue curve) and SRO (red curve) (i), and schematic of CRO (ii) showing a bigger ${\mathrm{GdFeO}}_3$-type distortion than SRO (iii). (b) Pronounced spectral weight transfer in Mott-Hubbard system $\mathrm{Sr}\left(\mathrm{Ca}\right){\mathrm{VO}}_3$, which is reproduced from Ref. [7] (i). Schematic of spectral weight transfer with increasing the $U/W$ in a typical Mott-Hubbard system (from the DMFT solution of the Hubbard model; Refs. [41, 42]) (ii). (c) Simple phase diagram of CRO and SRO. The $y$ axis and the bottom and top $x$ axes are the ${\mathrm{GdFeO}}_3$-type distortion and the $U/W$ and magnetism, respectively. Ruthenium perovskites are indeed a tunable system to explore the electron correlations and related quantum criticality, with structural distortion as the knob (can be experimentally modified via epitaxial strain and doping).