Consequences of realistic embedding for the L 2,3 edge XAS of α-Fe 2 O 3 .

Cluster models of condensed systems are often used to simulate the core-level spectra obtained with X-ray Photoelectron Spectroscopy, XPS, or with X-ray Absorption Spectroscopy, XAS, especially for near edge features. The main objective of this paper is to examine the dependence of the predicted L2,3 edge XAS of α-Fe2 O3 , an example of a high spin ionic crystal, on increasingly realistic models of the condensed system. It is shown that an FeO6 cluster model possessing the appropriate local site symmetry describes most features of the XAS and is a major improvement over the isolated Fe3+ cation. In contrast, replacing next nearest neighbor positive point charges with Sc3+ , a closed shell cation of similar spatial extent to Fe3+ , only marginally improves the match to experiment. This work suggests that second nearest neighbor effects are negligible. Rather, major improvements to the predicted L2,3 edge XAS likely requires additional many body effects that go beyond the present study in which the multiplets are restricted to arise from angular momentum coupling within a single open shell configuration.