![]() |
|
Surface Termination and Interlayer Relaxations in Fe3O4(001)S. A. Chambers, S. Thevuthasan, S. A. Joyce, B. Stanka,(a,b) W. Hebenstreit,(b) and U. Diebold (a,b) Supported by OBER/OBES EMSP, OBES Division of Materials Science, DOE-EPSCoR, and NSF-CAREER. The termination of a compound material is among the most fundamental of all surface properties. Surface reactivity is strongly influenced by the elemental composition of the terminal layer, which typically changes as different layers are considered. The surface autocompensation principle, originally developed to explain surface reconstruction of compound semiconductors, provides a useful criterion for predicting the surface termination. This model states that surfaces with the greatest stability are those for which anion (cation)-derived dangling bonds are completely full (empty). However, this simple qualitative principle cannot predict interlayer relaxation, or the relative stabilities of structurally inequivalent surfaces that are autocompensated. Magnetite (Fe3O4) is an interesting and important material in catalysis, corrosion, geochemistry, and magnetism. Magnetite exhibits an inverse spinel structure, consisting of alternating layers of tetrahedrally coordinated Fe(III) and octahedrally coordinated Fe(II) and Fe(III) along with tetrahedrally coordinated oxygen. The ratio of Fe(II) to Fe(III) is 1:1 in each bulk octahedral Fe layer. The (001)-oriented surface of this material exhibits a (Ö 2xÖ2)R45° surface reconstruction. However, the surface termination and interlayer relaxations are not known. Based on previous successes of the autocompensation model as applied to oxides, we take as a working hypothesis that autocompensation is a necessary but insufficient condition for the correctness of any particular surface structural model. In this light, the most natural way to achieve an autocompensated Fe3O4(001) surface that possesses the observed (Ö2xÖ2)R45° symmetry is to terminate with a half monolayer (ML) of tetrahedral Fe(III). However, autocompensation can also be achieved with a terminal layer of octahedral Fe and tetrahedral oxygen by varying the number of oxygen vacancies and the Fe(III) to Fe(II) ratio per surface unit cell. It should be straightforward to discriminate between these two candidate structures based on the symmetry of scanning tunneling microscopy (STM) images. Indeed, atomically resolved STM images have been reported, but their interpretation has not been straightforward. We have shown by STM that Fe3O4(001), as grown by oxygen-plasma-assisted molecular beam epitaxy (OPA-MBE) on MgO(001), is terminated with 1/2 ML of tetrahedral Fe(III), rather than with a layer of octahedral Fe and tetrahedral oxygen. We have also determined the interlayer relaxations by x-ray photoelectron diffraction (XPD) for the as-grown surface. However, we have also shown that the octahedral Fe/tetrahedral O termination with an ordered array of oxygen vacancies is stable and tends to form after the surface has been exposed to the ambient atmosphere for ~1 hour, despite thorough cleaning once the specimen is back in an ultrahigh vacuum environment. In Figure 2.4 we show STM images of the two terminations of Fe3O4(001). The iron-derived density of states is much higher than that of oxygen near the Fermi level. Therefore, it is the Fe ions that are visible as bright spots in the STM images. A primitive unit cell of side length equal to ~8.5 Å is clearly visible in the Figure 2.4a. There are tetrahedral Fe(III) cations at the corners of the unit cell, and octahedral Fe cations in the layer below are not visible. In contrast, chains of octahedral Fe cations are clearly visible in Figure 2.4b, as are ordered arrays of O vacancies. It is even possible to image isolated tetrahedral Fe(III) cations, and one is seen in Figure 2.4b. A slight lateral displacement within the octahedral Fe chains, giving rise to a zig-zag pattern within the rows, is also visible in Figure 2.4b. We have determined the interlayer relaxations for the tetrahedral-Fe terminated surface by scanned-angle XPD. Detailed comparison of experiment with single scattering calculations via R-factor analysis reveals that the first four interlayer spacings are relaxed by -14%, -57%, -19%, and +29% of their respective bulk values.
This research highlight summarizes work described in the two articles listed below. Readers interested in more information are encouraged to read the source documents. ReferencesChambers, S. A., S. Thevuthasan, and S. A. Joyce, Surf. Sci. Lett., submitted (1999). Stanka, B., W. Hebenstreit, U. Diebold, and S. A. Chambers, Surf. Sci., to appear (1999).
William R. Wiley Environmental Molecular Sciences Laboratory Feedback: webmaster@emsl.pnl.gov Revised: June 12, 2001 Security & Privacy PNNL-13147 |