Magnetite Fe3O4(111): surface structure by LEED crystallography and energetics

The atomic structure of the Fe3O4(111) surface was determined by means of dynamical low-energy electron diffraction (LEED) after being prepared in two different ways. In a first experiment up to 10 monolayers of well-ordered iron oxide films were grown epitaxially onto Pt(111) substrates. A 1 monolayer thick film forms a hexagonal lattice with a lateral repeat distance of 3.2 Å, 15% larger than the lateral periodicity of Pt(111). Above 1 monolayer coverage the LEED pattern reveals a lateral repeat distance of 3.0 Å, indicating a contraction of the oxide lattice with respect to the first monolayer. This new LEED pattern shows half-order spots and is compatible with (2 × 2) reconstructed FeO(111) and bulk terminated Fe3O4(111) surfaces. By applying automated tensor LEED to many possible surface structures of these two iron oxides, 8 monolayer thick films were identified to be magnetite, Fe3O4. Auger electron spectroscopy (AES) measurements on these films also reveal a stoichiometry close to that of Fe3O4. In a second experiment the (111) surface of an α-Fe2O3 single crystal was prepared by Ar+ ion bombardment and subsequent annealing. Brief annealing to 900, 1000 and 1200 K in 10−10 and 10−6 mbar oxygen creates three different LEED patterns indicating structural transformations occurring in the surface region of this crystal. Prolonged annealing to temperatures between 900 and 1200 K stabilizes the same LEED pattern and gives identical intensity-voltage curves as obtained on the 8 monolayer thick films. Therefore the crystal surface region has been reduced to Fe3O4 and has the same surface structure as the epitaxially grown films. X-ray photoelectron spectroscopy (XPS) measurements on this surface also reveal a stoichiometry dose to that of Fe3O4. The best fit structure for both preparations corresponds to an unreconstructed, but strongly relaxed, polar (111) surface termination of magnetite that exposes 14 monolayer of Fe ions over a distorted hexagonal close-packed oxygen layer and minimizes the number of dangling bonds. The surface relaxations are probably driven by electrostatic forces. Our results indicate that minimization of both the number of dangling bonds and the electrostatic surface energy are important in determining the termination and relaxations of this polar metal oxide surface. The electrostatic surface energetics is qualitatively discussed within general, simple concepts applicable to all ionic crystals.