We applied electron-scattering quantum-chemistry theory to investigate the origin of contrast in scanning tunneling microscopy images of iron oxide grown on Pt͑111͒. We show that ͑a͒ image contrast and surface topography are not directly related; ͑b͒ the maxima in the images occur over O positions for Pt tips and over Fe positions for O-terminated tips; ͑c͒ the short-range corrugation is always due to O p z orbitals, with a small but important contribution of Fe; and ͑d͒ the state that carriers most of the tunneling current is located near the Fermi level and is made mostly of O p z and Fe 3d z2 orbitals. ͓S0163-1829͑96͒50840-4͔Oxide surfaces are technologically important and yet poorly understood in surface science. Complications arise from the existence of two different elements with possible variable stoichiometry due to loss of oxygen, and the presence of strong dipoles as a result of the ionic character. Due to its atomic-scale resolution, scanning tunneling microscopy ͑STM͒ is one of the techniques that can help most to advance this field. However, the interpretation of STM images is a nontrivial problem, augmented in the case of oxides by the desire to distinguish the two different elements. Among the most studied oxides using STM are rutile TiO 2 , 1-8 Fe 3 O 4 , and ␣-Fe 2 O 3 . 9-12 Confusion exists in the relationship between the images and the crystallography of the surface. Oversimplified views like the notion that O atoms should be imaged at negative sample bias and metal atoms at positive bias ͑on account of the valence-band containing states due to filled O 2p orbitals and the conduction-band empty metallic states͒ are wrong, as we will see.The agreement between theory and experiment obtained by the application of the electron-scattering quantumchemistry ͑ESQC͒ theory to interpret STM images for a large range of systems ͑graphite, adsorbates on metal surfaces, molybdenum sulfide MoS 2 ͑Refs. 13-17͒ prompted us to investigate the contrast mechanism in oxide surfaces. This is the first successful attempt at determining the origin of contrast and the relationship with atomic positions for an oxidelike surface.The FeO film was prepared by evaporating Fe from a hot filament onto the Pt substrate. After this, it was heated to 750°C in an oxygen pressure of ϳ10 Ϫ5 torr for ϳ30 s.Several favorable circumstances concur in the case of iron oxide films formed on single-crystal Pt substrates. One is that many important structural parameters of the monolayer have been recently determined by STM ͑Ref. 18͒ and x-ray photoelectron diffraction ͑XPD͒. 19 Second is the incommensurability between the oxide layer ͑3.

No abstract