Abstract : Ferritic steels possibly strengthened by oxide dispersion are candidates as structural materials for generation IV and fusion nuclear reactors. Their use is limited by incomplete knowledge of the iron-chromium phase diagram at low temperatures and of the phenomena inducing preferential segregation of one element at grain boundaries or at surfaces. In this context, this work contributes to the multi-scale study of the model iron-chromium alloy and their free surfaces by numerical simulations. This study begins with ab initio calculations of properties related to the mixture of atoms of iron and chromium. We highlight complex dependency of the magnetic moments of the chromium atoms on their local chemical environment. Surface properties are also proving sensitive to magnetism. This is the case of impurity segregation of chromium in iron and of their interactions near the surface. In a second step, we construct a simple energy model for high numerical efficiency. It is based on pair interactions on a rigid lattice to which are given local chemical environment and temperature dependencies. With this model, we reproduce the ab initio results at zero temperature and experimental results at high temperature. We also deduce the solubility limits at all intermediate temperatures with mean field approximations that we compare to Monte Carlo simulations. The last step of our work is to introduce free surfaces in our model. We then study the effect of ab initio calculated bulk and surface properties on surface segregation. Finally, we calculate segregation isotherms. We therefore propose an evolution model of surface composition of iron-chromium alloys as a function of bulk composition.