Wiggling its way out of surface polarity: Fe3O4(100) (A Perspectives

Spintronics is the catch phrase for novel kinds of electronics that exploit the spin of charge carriers – electrons or holes – in addition to their charge. Massive efforts by scientists and engineers are underway to bring many exciting concepts in this field to fruition. It would be good if one could find a practical material where all the conduction electrons have the same spin orientation. Such a material could deliver electrical current that would be 100% spin-polarized, which could, for example, be used to flip magnetization domains in a device by simply sending a current through it. Magnetite, Fe3O4, is a presently considered one of the ‘hot’ magnetic materials. (It is, at the same time, a very old one. In fact, loadstone, as it is commonly known, was part of the earliest compasses. And bacteria synthesize magnetite and use it as a navigational tool to guide themselves into their preferred habitat [1].) Density-functional theory (DFT) calculations for Fe3O4 predict that the minority spin band in Fe3O4 is metallic, while the majority band has a band gap [2]. This is exactly the property one is seeking; commonly called ferromagnetic halfmetallic behavior. In addition, Fe3O4 has a high Curie temperature of over 800 K, which makes it particularly attractive for spintronics applications. It is still an issue of ongoing discussion whether or not this 100% spin polarization is really present in Fe3O4, see for example two articles in a recent focus issue on this topic [3,4]. (This matter is complicated by the fact that the material becomes insulating around 120 K, which makes it impossible to use Andreev reflection measurements as a litmus test for a ferromagnetic half-metal.) In any event, if one wants to use Fe3O4 as a spin source, the spin polarization would need to be preserved across its surface. Herein lies the crux of the problem: the most important lowindex surface, Fe3O4(100), turns out to be polar. Polar surfaces have a perpendicular, non-vanishing dipole moment, which in the established view leads to a high-energy, unstable situation [5]. In order to lift polarity, surfaces often reconstruct [6]. So does Fe3O4(100). A ( p 2 p2)R45 reconstruction has consistently been reported [7–10]. The models for this reconstruction vary, and so do the predicted consequences for ferromagnetic half-metallicity on the surface [11]. Based on DFT calculations, Pentcheva et al. [12] have previously made an interesting suggestion. They have proposed that the surface is fully stoichiometric and terminates with rows of octahedrally-coordinated Fe atoms, separated by oxygens (Fig. 1). (Fe3O4 contains both, octahedrally and tetrahedrally-coordinated Fe atoms in the bulk.) Their model contains a small, yet important modification to a bulk-like termination. Atoms are not aligned in straight rows, which is what one would get by just cutting the crystal. Instead, they form a wave with a periodicity that causes the ( p 2 p2)R45 symmetry. This structure is in very nice agreement with early STM studies [8] and an XRD analysis [12]. According to Pentcheva et al.’s previous DFT calculations, these ‘wiggles’ are sufficient to provide the lowest-energy surface for a wide range of oxygen chemical potentials. A Jahn–Teller-like distortion of the surface atoms was invoked as an explanation. This is a novel idea, and much different from other reconstructions of polar surfaces that typically involve non-stoichiometries and/or a re-arrangement of charges. Moreover, in Ref.