Strongly gapped spin-wave excitation in the insulating phase ofNaOsO3

Abstract

NaOsO3 hosts a rare manifestation of a metal-insulator transition driven by magnetic correlations, placing the magnetic exchange interactions in a central role. We use resonant inelastic x-ray scattering to directly probe these magnetic exchange interactions. A dispersive and strongly gapped (58 meV) excitation is observed indicating appreciable spin-orbit coupling in this 5d 3 system. The excitation is well described within a minimal model Hamiltonian with strong anisotropy and Heisenberg exchange (J1=J2=13.9 meV). The observed behavior places NaOsO3 on the boundary between localized and itinerant magnetism.The underlying mechanisms driving a metal-insulator transition (MIT) is an enduring focus of condensed matter physics [1]. Recent interest has extended investigations to 5d-based transition metal oxides that host new paradigms of competing interactions creating novel MITs [2]. For example spin-orbit coupling (SOC) in 5d 5 iridates dramatically influences the electronic ground state to allow even the presence of the reduced on-site Coulomb interaction (U ) to drive a relativistic Mott MIT [3]. Conversely, in 5d 3 osmium-based compounds MITs occur that cannot be reconciled with the reduced U in 5d systems, even when the large SOC is taken into account. These compounds therefore fall outside the Mott approximation. Of particular interest in this regard are the osmates NaOsO 3 and Cd 2 Os 2 O 7 which undergo a MIT that is continuous and coincident with the onset of magnetic order indicating the central role of magnetic interactions in the transition [4][5][6][7]. Accessing the collective role of the competing inter and intra-ion electron-electron interactions, SOC and magnetism in driving the MIT is required to gain an understanding of the novel MITs in these osmates.Outside of a Mott MIT several other mechanism exist to describe the transition, including an Anderson MIT driven by disorder [8] and a Peirels MIT driven by a structural distortion in a low dimensional system [9].Slater considered a route in which magnetism could drive a MIT with the central observation being that within a magnetically ordered system the potential created by an up-spin is different from that created by a down-spin [10]. By this definition three-dimensional magnetic ordering with oppositely aligned spins is a route to a MIT, and implicitly includes q=0 antiferromagnetic structures. NaOsO 3 exhibits several features consistent with Slater's general scenario [6,7,[11][12][13][14][15]. The MIT occurs concomitant with the onset of antiferromagnetic ordering (T N = T MIT = 410 K) that can create a periodic potential. Furthermore in NaOsO 3 the MIT is continuous and no structural symmetry change occurs. However several important questions have so far remained experimentally inaccessible hindering the development of further insight into the mechanism of this unusual MIT and prohibiting a quantitative description beyond the mean-field approach invoked by Slater for a magnetic MIT.Principally, since the MIT is driven by magnetic ordering the m...

Crossover from itinerant to localized magnetic excitations through the metal-insulator transition inNaOsO3

“…This is consistent with a Slater picture in which interactions are mean-field like [20]. Meanwhile previous RIXS measurements showed well-defined and strongly gapped (∼50 meV) dispersive spin-wave excitations at 300 K [22]. This was found to be consistent with an anisotropic nearest-neighbor Heisenberg picture for the magnetic Hamiltonian and is suggestive of localized magnetic moments.…”

“…The nature of the low-energy excitations at 300 K-deep in the antiferromagnetic insulating phase-has already been described in Ref. [22]. Later in this manuscript (Sec.…”

“…Just as in Ref. [22], four peaks are evident in the RIXS spectra. The peak centered at zero energy loss comprises the elastic line and other low energy features such as phonons, magnons, etc.…”

“…[22]) assumes that the magnons are underdamped, whereas it has already been shown that the behavior at 450 K appears consistent with overdamped spin excitations (Sec. V).…”

“…First, the magnitude of the anisotropy presented in Ref. [22] is likely to be an overestimate. This is a direct result of the finite momentum resolution of the RIXS spectrometer, which most prominently affects the magnetic excitations at the Brillouin zone center, where the dispersion is steepest [58].…”