It is shown that whereas the spherical and spheroidal jellium models are inadequate to describe lithium clusters, only the ellipsoidal jellium model is adequate. The corresponding result, obtained by Yannouleas and Landman, was unpublished at the time of submission of our paper.

Some properties of small Lin clusters (n up to 20) are theoretically investigated, within the density functional theory formalism. The structural properties are examined at the so-called local level of approximation. For very small clusters (n<=8), the Lin conformations which are well known from ab initio calculations are found at very low computational cost. For n>8, optimal starting geometries are generated from two growth patterns, based on the increase of the number of pentagonal subunits in the clusters by adsorption of one or two Li atoms. Several new stable structures are proposed, for which the corresponding vibrational analysis is performed for n up to 18. The study of energetic properties and stability requires the use of gradient-approximated functionals. Such functionals are used for the determination of the relative stability of these clusters. For example, we show that the icosahedral structure is the most favorable geometry for Li13, whereas this is not the case for Na13. Ionization potentials and binding energies are also investigated in regard to the size and the geometry of the clusters. Comparison with experimental results and other theoretical approaches (such as nonspherical jellium model) suggests that some combinations of gradient-corrected functionals are more adapted than others to describe Lin energetic and structural properties.

Density functional calculations of ground and excited states of Li n (nle8) clusters have been performed, within two different approaches. Using a set of Kohn-Sham orbitals to construct wave functions, the calculation of the oscillator strengths of the electric dipole transitions is performed. Our results have been tested at two levels: first the necessary comparison with the experimental data, second the comparison with high level CI (MRD-CI) calculations. This last point is not a trivial challenge, because such anab initio method leads for small clusters to a highly accurate description of the electronic structure and optical absorption spectra. Finally, this is also a new test for the capability of using Kohn-Sham orbitals to construct physically meaningful wave functions. Transition energies, oscillator strengths and finally optical absorption spectra presented here are in general in reasonable agreement with the experimental data and the MRD-CI calculations. That is very promising for bigger systems, with regard to the modest computational effort (CPU time and memory size) of density functional calculations.