Correlation effects in the low-energy region of nickel photoemission spectra

Abstract

The role of on-site correlation in the low-energy excitations of nickel is studied by comparing the results of high-angular and high-energy resolution photoemission spectroscopy with quasiparticle states calculated as a three-body scattering solution of a multiorbital Hubbard model. It is found that correlation effects modify the energy dispersion and spin polarization of electron states and are essential in order to get a quantitative agreement with experimental data.It is well established that the photoemission spectra of narrow-band materials, such as the elements of the d transition-metal series and their compounds, cannot be entirely explained within a one-electron picture, due to the presence of local correlations between electrons in the partially filled d band.1 Band mapping, i.e., the reduction of the measured spectra to a band structure and its comparison with theoretical results can be a powerful tool to directly investigate correlation effects, provided that some important requirements are fulfilled: on the experimental side, highangular and high-energy resolution photoemission techniques are necessary in order to identify quasiparticle energy dispersions and lifetime broadenings associated with the many-body character of the electronic excitations; on the theoretical side, a realistic description of the band structure must be combined with an accurate treatment of many-body electron-electron interactions to account for the mixed itinerant and localized behavior of the valence states.In this paper we present direct evidence of correlation effects in the low binding-energy region of the valence band of nickel. While these effects dominate the high-energy region of the Ni photoemission spectrum with the presence of the well-known 6 eV satellite peak, 2 the electron states near the Fermi energy are commonly believed to be less influenced by many-body interactions; this has to do with the very general properties of low-energy excitations of Fermi liquids, 3 but also with the observation that the Fermi surface of ferromagnetic nickel is nicely reproduced by a singleparticle band structure. 4 All the same, significant discrepancies are known to exist between the observed energy dispersions of some bands and the results of standard singleparticle band calculations; 5 these discrepancies are related to the energy renormalization due to electron-electron interaction. Here we want to investigate these effects in detail in order to get a band mapping of low-energy quasiparticle excitations.A high-resolution photoemission data set was measured at room temperature on a Ni͑110͒ surface. Using He I radiation for excitation (hϭ21.21 eV), energy spectra were taken in 1°steps for a continuous range of polar angles along the ͑001͒ plane, from normal emission ( m ϭ0°) to a polar angle of m ϭ70°off normal. Each spectrum spans a bindingenergy range from Ϫ300 meV ͑above E F ) to 1200 meV ͑below E F ) and was measured with an energy resolution of 35 meV. The angular resolution was set to Ϯ0.5°using an iris aperture in front...

Spectral Function of Ferromagnetic3dMetals: A Self-ConsistentLSDA+DMFTApproach Combined with the One-Step Model of Photoemission

“…A comparison of our results for Ni(011) with a former theoretical study and the corresponding experimental spec tra [21] is presented in Fig. 3.…”

“…For the complete variety of emission angles the energetic peak positions coincide with the experiment within about 0.1 eV. Upper row: comparison between LSDA-based calculation and experiment [21], middle row: comparison between experiment and non-self-consistent quasiparticle calculations neglecting ma trix element and surface effects [21], lower row: spin-integrated LSDA + DMFT spectra including photoemission matrix ele ments (this work). Theory: solid red line, experiment: black dots.…”

Angle resolved photoemission from Cu single crystals: Known facts and a few surprises about the photoemission process

“…This however means that now the relative peak intensities and line shapes have to be used in the interpretation, rather than the simple peak positions. Theory has advanced meanwhile to a state, where these spectral functions can be calculated under certain approximations and then compared with the actual photoemission data (17,(37)(38)(39). The agreement between theory and experiment for correlated magnetic solids however is far from perfect at present (40).…”

Transmission through correlatedCunCoCunheterostructures

“…On the contrary, in the minority-spin channel one finds a high density of electrons which can be immediately excited, leaving back holes. Such an occupation asymmetry has consequences concerning possible interaction channels in the multiorbital Hubbard model: A majority-spin hole can only scatter with opposite-spin particles, which would cost an effective interaction U , while a minority-spin hole may also scatter with parallel-spin particles with the effective interaction U − J < U [68]. Therefore correlation effects are expected to manifest themselves differently for majority-and minority-spin electrons.…”