The hydrothermal alteration of a natural peridotite leads to a strong mineralogical change, the socalled ―serpentinization process‖. During this process, the Fe2+ initially contained in olivines and/or pyroxenes is partially oxidized in Fe3+ incorporated in new mineral phases like serpentine or magnetite. This reaction conducts to the hydrolysis of water and therefore produces high amount of hydrogen. In the mid-ocean ridges, high concentrations of hydrogen and hydrocarbons have recently been measured on a new type of hydrothermal fields, always located on on ultramafic rocks. With the aim to better understand the processes controlling the hydrothermal alteration of peridodites, and in particular the related hydrogen production, numerous in-situ experiments of serpentinization were coupled with thermodynamic calculations. The altered products were characterized using different spectroscopic tools, particularly by X-ray absorption analysis in 3rd generation synchrotrons. Several experimental protocols were thus developed allowing to accurately quantify the phase proportions, the iron distribution, and the iron speciation in the altered samples. Experiments display that the serpentinization kinetic is highly dependent from the physico-chemical conditions (temperature, water-to-rock ratio, structural state and chemical composition of the protolith) inferred in the natural environment. In parallel, our work shows that the roles play by both serpentine and magnetite mineral phases evolve as function of temperature but as well change during the alteration process. These results were coupled with the numerical simulations to be replaced in a natural context determining the different processes encountered along a P-T hydrothermal pathway. As a result, the serpentinization reaction is mainly effective during the ―prograde‖ pathway and the maximum hydrogen production occurs around 325°C. Finally and based on the thermal gradients inferred in such contexts, the size and the geometry of the ―Rainbow‖ hydrothermal field were estimated.
