Prediction of New Multiplet Structure in Photoemission Experiments

Abstract

Correlation effects are included in a detailed theoretical study of the multiplet structure observed by x-ray photoelectron spectroscopy. This correlated multiplet hole theory resolves the difficulties of earlier multiplet hole theory studies: It correctly determines the magnitude of the observed 2s and 3s (doublet) splittings and their intensity ratios, and predicts the existence of new structure for the 3s binding energy splittingsstructure which does not exist in one-electron theory and which has now been confirmed by Kowalczyk et al, Photoelectron spectroscopy (PS) has become an important tool for studying the electronic and magnetic properties of solids in both the ultraviolet 1 (UPS) and x-ray 2 (XPS) energy regions. As with the application of any new technique, new observations 1 ' 3,4 have revealed serious flaws in our theoretical understanding. 5 We present here the results of a detailed theoretical study of the multiplet structure of the core-electron bindingenergy splittings observed by XPS in transitionmetal ions such as Mn 2+ and Fe 3+ in insulating compounds. We show that correlation effects are of prime importance for understanding the magnitude of the splittings and their intensity ratios. We predict new structure for the 3s multiplet structure, which does not exist in any rigorous one-electron theory. In addition, we correctly determine the magnitudes of the 6 2s and 4 3s (doublet) splittings in the XPS spectra of these ions, thus eliminating the factor of 2 discrepancy between multiplet hole theory 7 (MHT) and experiment, and obtain the correct intensity ratio of the 3s splittings (2:1 rather than the MHT ratio of 7:5). The recent confirmation by Kowalczyk et al. 8 of the predicted structure demonstrates a unique success of theory and indicates the possible importance of correlation effects on other photoelectron-spectroscopy studies of solids. 1 " 3,5 XPS multiplet splittings, first observed in the if-shell peaks of the paramagnetic molecules NO and 0 2 , 9 have now been observed in a series of transition-metal 4 ' 6,10 and rare-earth metal 11 compounds. To understand the origin of these splittings, consider the case of the free Mn 2+ ion whose 6 S ground state has the configuration [(core) 3s 2 3^63d 5 ] 6 S. If a 3s electron (or equivalently a 2s electron) is removed, two states of the Mn 3+ ion may be formed which are eigenfunctions of L 2 and S 2 : [(core) 3s 1 ( 2 S)3p 6 3d 5 ( 6 S)] 7 S and [(core) 3s l ( 2 S)3p*3d 5 ( 6 S)] 5 S. These two states differ in the angular-momentum coupling of the 3d 5 ( 6 S) shell. For the 7 S state, the 3s electron can be thought of as having its spin parallel to those of the five (parallel) 3d electrons; for the 5 S state, the 3s spin may be thought of as antiparallel to the 3d spins. In terms of the Slater F k (Coulomb) and G k (exchange) integrals, the energy difference between these two states is f G 2 (3s, 3d). Now if the reorganization (relaxation) effects are small, the multiplet splitting may be obtained by evaluating the G 2 integral using the Har...

Hyperfine Interactions and ESCA Data

“…These satellite peaks 2+ may be observable. Such peaks were predicted [15] in Mn and observed [16] in MnF (Fig. 5).…”

The interfacial and surface properties of thin Fe and Gd films grown on W(110) as studied by scanning tunneling microscopy, site-resolved photoelectron diffraction, and spin polarized photoelectron diffraction

“…11 for the Gd 4s spectrum is thus most likely due to enhanced conelation effects arising from the high degree of 4s-4f overlap. However, the apparent influence of these correlation effects in 4s emission from Gd is opposite in sign to that studied previously in the analogous 3s emission from ionic MI?+, for which the high-spin 7s peak in relative intensity compared to the closest-lying low-spin 5s peak [45]. Thus, the Gd 4s spectrum exhibits an anomalous intensity ratio that warrants fbrther theoretical study.…”

“…The experimental and theoretical Gd 5s splittings are found to agree excellently (3.91 eV versus 3.62 eV, respectively), whereas the same splittings for Gd 4s are very far apart however, we still expect on the basis of prior theory [45] that the two members of the doublet will be highly spin polarized, and thus that they can be used to derive the clear SPPD effects that are discernible in the 4s spin asymmetry, as shown in the previous section.…”

“…The fact that this anomolous ratio is seen with both low and high photon energies rules out an explanation in terms of deviations from the high-energy sudden approximation limit that has been used successfblly to explain core-level multiplet structures in both Mn [45] and Gd [29]. This violation of Eq.…”

“…With the Gd 4s multiplet, the situation becomes more complicated due to the weaker valence electron coupling to begin with, as well as to the stronger correlation effects giving rise to a manifold of final states as compared to the relatively simple 7S and 9s states found in the 5s. Hence, variations with temperature among the different elements of the 4s configuration interaction (CI) manifold [41,45] due to spin polarization may produce changes in the multiplet peak splitting that cannot be explained simply by including valence effects.…”