Due to the fast development of microelectronics, quite an important number of the chemical aspects of component formation has been occulted. In particular just a few studies are related to thin film formation in the scope of chemical reactions and modelisation. Thermodynamics itself has been implemented rather ponctually and in an incomplete way. One of the Mayn reasons for this is that thin films are considered as standingly non equilibrium systems and that. thermodynamics, as a macroscopic science, has Maynly been considered as inadequate. Thus, we have decided to reconsider the problem of thin films formation from a strictly chemical point of view.Using a simple system (formation of silicon nitride on monocrystalline silicon), the chemical reactions have been investigated. Classical thermodynamics, though inaccurate and approximate for such systems is necessary to develop the first steps of a thermodynamical model describing the deposit formation. Using a computer program helping a global treatment of several chemical reactions confirms the conclusions obtained by classical thermodynamics. We have discussed the insufficiency of the classical model, and shown that, with respect to what was experimentally observed, it was nevertheless globally valid, and that the nitride formation should be regarded as a superposition of classical chemical reactions that occurred in "close to the equilibrium" states and globally following the classical laws of thermodynamics. Thus, we propose a model of formation for the first mono layer of Si3N4 on monocrystalline silicon which is in total agreement to experimental observations.