Three basically different models of CSN clusters are studied, partly in ord
er to explore the limitations of each model and partly in order to study ge
neral ground-state properties of the clusters. One model is based on the sp
herical-jellium model within the density-functional formalism, another is a
semi-empirical tight-binding model obtained by parametrizing band structur
es for an infinite crystal, and the third model is a spherical-well model f
or non-interacting particles. Particularly stable clusters are found for sy
stems with only completely filled electronic shells, although this result i
s somewhat obscured by surface effects for the tight-binding model. For the
density of states as a function of N the tight-binding model is the one pr
oviding the most accurate information, especially for the features closest
to the Fermi level. Only this model gives the proper description of those i
n the limit N --> infinity. Finally, we examine the electron density for di
fferent clusters and explore how Friedel oscillations occur. In particular
the jellium model predicts very regular density oscillations, which can be
ascribed to electron-electron interactions. We study both the pure clusters
and ones with a void at the centre, where the latter represents a simple m
odel for Cs-covered C-60 molecules. The two systems show many similarities
- in particular it is demonstrated that the stable clusters occur with the
same spacing Delta R of the radius of the system. The cluster sizes range u
p to values of N of about 10 000 for the jellium and the tight-binding mode
ls and to over 30 000 for the spherical-well model. In total the study show
s that although many properties are well described by all of the models, it
is important to be aware of their limitations, and it would be desirable t
o incorporate more experimental information in order to be able to evaluate
the quality of the different models. To this end the 'magic numbers' are l
ess convenient.