Phosphorus, mainly in the form of P 2, has been evaporated on a InP(100) sample in order to compensate P vacancies and to remove metallic In clusters from the surface. These defects were induced by argon ion bombardment or vacuum annealing during the “in situ” cleaning procedure of the substrate. The damaged InP surface, characterized by EELS and AES, may be considered as a (In, InP) two-phases system in which the content and the distribution of metallic indium is dependent on the experimental parameters used in the treatments. Phosphorus evaporation on this sample at room temperature induces a typical change of the loss spectra. The intensity of the lines characteristic of metallic indium (surface and volume plasmons at 8,6 eV and 11,5 eV, respectively) are decreased while the volume plasmon of indium phosphide at 15 eV is enhanced. A saturation of the interaction between phosphorus and the (In, InP) system occurs after metallic indium clusters are changed to indium phosphide. The transformation of metallic In to covalent InP bonds is supported by Auger line shape analysis of the In-MNN and P-LMM spectra. The removal of the stoichiometry defects of the InP substrate leads to a surface more stable (against heat treatments and electron stimulated oxydation) than the initial disordered one. For comparison, a pure polycrystalline metallic In sample has been investigated simultaneously to the InP substrate, and subjected to phosphorus evaporation under similar conditions. Such a treatment results in the formation on the In metal surface of a thin InP layer of a thickness larger than the probed depth (about 2 nm).