Based on ab initio and density-functional theory calculations, an empirical potential is proposed to model the interaction between a fullerene molecule and many sodium atoms. This model predicts homogeneous coverage of C60 below 8 Na atoms, and a progressive droplet formation above this size. The effects of ionization, temperature, and external electric field indicate that the various, and apparently contradictory, experimental results can indeed be put into agreement. PACS numbers: 61.48.+c,36.40.Qv,34.70.+e Fullerene molecules are now commonly used as building blocks of complex materials having unusual physical and chemical properties [1]. The possibilities offered by molecules such as C 60 or C 70 in terms of electric or optical devices have laid ground for the rapidly expanding nanotechnology area. The interaction between fullerenes and alkali atoms has attracted a lot of attention after the discovery of superconductivity in the K 3 C 60 solid [2]. Sodium-C 60 compounds have also received much attention [3,4]. More recently, several groups investigated experimentally the gas phase properties of exohedral alkali-C 60 compounds [5,6,7], to further predict the solid state of alkali-doped fullerite.Using mass spectrometry, Martin and coworkers [6] inferred from the most stable "magic numbers" that sodium covers the C 60 molecule in a continuous and homogeneous fashion, and that metallic bonding starts above 6 metal atoms. Palpant and coworkers [7] deduced from photoelectron spectroscopy measurements that coating proceeds by trimers units rather than single atoms. They also estimated that metallic bonding only appears at n = 13 sodium atoms. Very recently, Dugourd and coworkers [8] measured the electric polarizability of Na n C 60 clusters in the range 1 ≤ n ≤ 34. They concluded that a sodium droplet is formed on the surface of the fullerene. These apparent contradictory results are surprising, as far as they rely on similar experimental conditions but different interpretations.The situation is in fact even more intricate due to the variety of theoretical conclusions on the very same systems [9,10,11,12]. With the exception of the study by Rubio and coworkers [9], which assumes complete wetting of C 60 by sodium in a continuous two-shell jellium description, the electronic structure calculations by Hira and Ray [10,11], and by Hamamoto et al.[12] reach different conclusions as to whether the Na 2 molecule remains as a dimer or dissociates into atoms, which locate on opposite sides of C 60 . At the unrestricted HartreeFock level, Hira and Ray find that charge transfer is negligible in NaC 60 , in apparent contradiction with experiments [7,8], and also that Na 2 remains as a dimer loosely bound to C 60 . On the other hand, the more realistic density-functional theory (DFT) calculations performed by Hamamoto and coworkers [12] at the local density approximation level tend to favor regular coating, preferentially by trimers, in agreement with the picture of Palpant et al. [7]. However, all these ab initio investig...