Aluminum oxide (Al2O3, alumina, known as sapphire when in single-crystalline form) is a perfect electronic insulator, hence difficult to study with surface science techniques, which mostly rely on electrons either sent to the surface or emitted by it. Thus, several mysteries on alumina surfaces remain till today.

Very recently, we could solve one of these mysteries, the surface structure of the basal plane of Al2O3, i.e., the (0001) surface. The most stable structure is a so-called (√31 × √31)R±9° superstructure, which was previously believed to consist of an ultrathin layer of metallic aluminum on top of the oxide. This is very astonishing since this surface is obtained by heating to about 1200 °C, a temperature where an aluminum layer would evaporate in a millisecond! We could finally obtain high-quality non-contact AFM images of the surface, showing the exact position of each Al and O atom. DFT calculations then told us the structure of the layers below. Stay tuned, these results will be published soon!

Ultrathin alumina

Initial oxidation of pure or almost-pure aluminum leads to rather disordered oxides, with varying thickness, currently not accessible for rigorous examination of the surface structure. Aluminum alloys are different, especially these that liberate aluminum atoms rather slowly, allowing the oxide to take its time forming a well-ordered structure.

Ultra-thin alumina on NiAl(110)

© Michael Schmid/IAP

Ultra-thin alumina on NiAl(110)

The best-known well-ordered alumina (=aluminum oxide) film is that on NiAl(110), discovered in the early 1990ies. After numerous attempts by both experimental and computational groups, we finally obtained high-resolution STM images of this surface in 2004, which provided the basis for ab-initio calculations by Georg Kresse, opens an external URL in a new window, finally solving the old puzzle. It turned out that the building rules of the film are different from all bulk alumina phases, and even the stoichiometry is different from Al2O3. This could be understood after detailed analysis – then it all appeared so simple! The Al atoms forming the interface between oxide and substrate cannot supply all their electrons to the oxide, they need to bind to the metal underneath. So their formal charge is not Al3+ but Al2+. To keep the oxide charge-neutral, more Al atoms than in Al2O3 are needed.

We have also studied defects in these ultrathin alumina films, as well as an alumina film on a Cu-Al alloy that turned out to have exactly the same structure as the one on NiAl(110). Obviously, this structure is a very stable configuration.

  • G. Kresse, M. Schmid, E. Napetschnig, M. Shishkin, L. Köhler, P. Varga
    Structure of the ultrathin aluminum oxide film on NiAl(110)
    Science     308, 1440 (2005); doi: 10.1126/science.1107783

  • M. Schmid, M. Shishkin, G. Kresse, E. Napetschnig, P. Varga, M. Kulawik, N. Nilius, H.-P. Rust, H.-J. Freund
    Oxygen-deficient line defects in an ultrathin aluminum oxide film
    Physical Review Letters     97, 046101 (2006); doi: 10.1103/PhysRevLett.97.046101

Filling a hole

A surprise came when we looked at the surface oxide on another alloy surface, Ni3Al(111). Again, we obtained atomic resolution, and with the experience obtained in the meanwhile we thought that the structural model was quite clear. Georg Kresse again did ab-inito calculations, again stressing his computer cluster to the limit (the cell has >1200 atoms!). Our simple model did not show the correct symmetry seen in the STM images. So Georg suggested that there must be a hole in the oxide at the corner of the unit cell, and suddenly everything was fitting nicely!

Ultra-thin alumina on Ni3Al

© Michael Schmid/IAP

Ultra-thin alumina on Ni3Al

The most exciting aspect was yet to discover: With a diameter of about 0.4 nm (0.0000004 mm), these holes are just wide enough to fill in a few atoms, which sit in the hole roughly one on top of the other. We need 3 palladium atoms to fill the hole; then the slightly protruding uppermost Pd atom provides a nucleus for growing larger metal clusters. Thus, we have discovered a template for growing well-ordered metal clusters regularly spaced by 4.1 nm – one of the nicest self-organized template surfaces in nanotechnology so far!

  • M. Schmid, G. Kresse, A. Buchsbaum, E. Napetschnig, S. Gritschneder, M. Reichling, P. Varga
    Nanotemplate with holes: Ultrathin alumina on Ni3Al(111)
    Physical Review Letters 99, 196104 (2007); doi: 10.1103/PhysRevLett.99.196104
  • A. Buchsbaum, M. De Santis, H. C. N. Tolentino, M. Schmid, P. Varga
    Highly ordered Pd, Fe, and Co clusters on alumina on Ni3Al(111)
    Physical Review B 81, 115420 (2010); doi: 10.1103/PhysRevB.81.115420