Magnetic dichroism in photoemission as a spin-resolving probe for electronic correlations

Abstract

Magnetic dichroism with unpolarized x rays in angle-resolved photoemission has been systematically investigated for the 2p core levels of Fe, Co, and Ni. In contrast to Fe 2p, we find the dichroism in the Ni 2p spectra to be mainly governed by electronic correlations. The fine structure observed in the Co 2p spectrum indicates a distinct influence of many-body interactions also in this material. ͓S0163-1829͑96͒50446-7͔Electronic interactions govern many phenomena in solidstate physics. Ferromagnetism, for instance, is a direct consequence of spin-dependent electronic correlations. Besides their influence in the ground state, correlations are very important for understanding electron spectroscopy results. These techniques always probe the many-electron system, even if the results may often be interpreted in the singleparticle limit. A famous example for the inadequacy of the single-particle picture is photoemission from Ni. Although a density functional approach can reproduce the average exchange splitting found in the experiment by a proper choice of the exchange correlation potential, 1 it fails to explain the photoemission satellite structures found, 2 yet another manifestation of electronic correlations. Their interpretation requires suitable many-electron formalisms, for example, a Hubbard-type approach as has been successfully applied to Ni. 3 A narrow d band in Ni-as compared to Fe and Co-is held responsible for the formation of correlation-induced spectral features. This bandwidth argument leads to an exciting aspect if surfaces and thin films are considered. It is well-known that the smaller number of nearest-neighbor atoms at the surface causes a dehybridization of the wave functions and thereby a band narrowing. A similar effect happens in monolayer films as a consequence of the reduced dimensionality, as long as hybridization with the substrate is negligible. As pointed out recently by Chen, 4,5 many-particle effects may thus be enhanced in systems of reduced dimensions.Magnetodichroic phenomena in angle-resolved photoemission appear for various experimental situations. 6-9 Previous studies have centered on the role of experimental geometry, light polarization, and excitation energy. The question of how the presence of electronic correlations will affect the magnetodichroic spectra remained unanswered. In this paper we demonstrate the strong influence of spindependent electronic correlations on the magnetic linear dichroism in 2 p photoemission. In order to make contact to explicitly spin-resolved results, the dichroism was measured with unpolarized x rays. On the one hand, the Fe 2 p spectra are similar to those observed earlier with circularly polarized light 6 and can be understood within a single-particle picture. On the other hand, we find the highest magnetodichroic signal for Ni 2p in the 6 eV satellite. In this case the magnetic dichroism is clearly dominated by correlation effects. The fine structure in the Co 2 p spectra from a 5 monolayer ͑ML͒ thick film shows evidence for a many-body satell...

Growth of cobalt on a VO(111) surface: Template, surfactant or encapsulant role of the oxide nanolayer?

“…An alternative interpretation of the high binding energy emission in the Co 2p 3/2 line shape might be in terms of a correlation-induced shake-up satellite, which is caused by the screening of a two-d-hole c À1 3d 8 4s 2 final state (c À1 is the core hole) by the conduction electrons [20]. This satellite structure is located at about 4.5 eV from the main emission line (c À1 3d 9 4s 1 ) and its spectral weight was shown to be crucially dependent on the Co dimensions: whereas the satellite intensity is hardly distinguished in Co 2p spectra of bulk samples, it may become significantly enhanced in systems with reduced dimensions, as recently demonstrated for Co nanolayers on Cu(1 0 0) and OCu(1 0 0) surfaces [21][22][23]. Theoretical results have predicted that the spectral weight of the shake-up satellite becomes significant in the case of strong d-d correlation and dehybridisation of the Co 3d bands [24,25].…”

Magnetic dichroisms in absorption and photoemission for magnetic characterization in x-ray photoelectron emission microscopy

“…The most prominent one is the sensitivity of MLDAD on the transverse component of the magnetization, well known from numerous photoemission experiments. [11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28] This is even true for excitation by circularly polarized light, 7 where the intensity asymmetry for different kinetic energies depends on both the longitudinal and the transverse components. This is due to the fact that for circular polarization a mixture of two effects contributes to this asymmetry, 28 -32 one being the linear dichroism displayed in Fig.…”

“…5 If energy filtering of photoemitted electrons is possible, magnetic dichroisms which occur in photoelectron spectroscopy may be employed as magnetic contrast mechanisms. Such dichroisms occur for various experimental geometries that involve circularly, 6-10 linearly, [10][11][12][13][14][15][16][17][18][19] or even unpolarized light, [20][21][22][23][24][25][26][27] and they can be observed in valence band 6,7,[17][18][19][20]28 or core level photoemission spectra. 8 -16,21-27 For highly symmetric geometries, circular dichroism in photoemission can be described in terms of relativistic selection rules for transitions between exchange and spin-orbit split states.…”

Electron correlation and charge transfer at the Ni/Co interface

“…In addition, satellite features at 3.7, 27.2, and 42.7 eV above the 2p 3/2 peak are indicated by downward pointing arrows. These satellites have been discussed previously, 15,16 with the 3.7 eV feature being ascribed to a two-hole final state in an analogous manner to the case for Ni XPS satellites. 14 The MCDPE spectrum, which is the difference between the two displayed XPS spectra, is also shown in Fig.…”

“…This line shape is similar to that reported by Dürr et al 17 for Fe thin films grown on Cu͑001͒ which could not be reproduced by one-electron calculations. From the similarity between the Fe MCDPE 15,17 and the Co MCDPE presented in this work it can be expected that a one-electron approach will also fail to reproduce the Co MCDPE. For the latter this is in part due to the electron correlation features in the MCDPE which are not considered in a one-electron approach.…”