Extensive LEED analysis of Ni(110). II. R-factor analysis of I(E) data

Abstract

For pt.I see ibid., vol.15, p.3223 (1982). The authors report on the r-factor analysis of the I(E) experimental part presented in the first paper. A critical discussion of the two more commonly used r-factors (Zanazzi-Jona (1974) and Pendry (1974)) is given, particular attention being paid to the size and to the nature of the data based on the final result. It is shown that small data bases may lead to different structural parameters depending on the choice of the r-factor. In this connection the normal incidence subset is a striking example of disagreement. On the other hand both types of r-factor analysis lead to similar models when the data base is diversified and large enough. Among the results obtained for the best model, the top layer is found contracted by 7%, much more than in all previous studies either by LEED or by medium-energy ion scattering. The energy dependence of the refractive potential is not only confirmed but is really a necessary ingredient for the quality of the final result. In addition, the surface refractive potential is found smaller than the bulk value by 2 eV. No evidence of more than a very few per cent random steps is found on the studied surface.