Abstract
In very low-energy electron diffraction (VLEED), the fine structures appearing below the emergence thresholds in I00(E) are usually used to test surface barrier models. The authors reinvestigate this problem, in particular for the case of a loosely packed surface. Two surface barrier models are tested in the case of a very-low-energy intensity profile (E <or approximately=5 eV) obtained on the W(001) surface for theta =45 degrees along the (10) azimuth: the modified image barrier introduced by Jennings and Read (1974) and a new saturated image barrier. Neither of these models yields a good fit of experimental data as long as one assumes that the barrier joins the bulk inner potential at the top layer edge continuously or with a potential step. Without challenging the validity of the muffin-tin approximation inside the first layer, it can be suspected that the top-layer potential is less attractive than any bulk-layer one. it is found that for both barrier models, shifting the top-layer inner potential upwards leads to a significant improvement of the fit. It is also found that the modified image barrier tends to exhibit a non-physical region of constant potential potential near the top layer and as a consequence the saturated image barrier appears to be more reliable. Finally this work illustrates the fact, that, especially for loosely packed surfaces, the surface barrier concept must not be restricted to the vacuum region but has to be extended inside the crystal region where the bulk potential is modified by the presence of the surface.