Fe oxyhydroxides in riverbanks and their high binding capacity can be used to hypothesize that riverbanks may act as a "biogeochemical filter" between wetlands and rivers and may constitute a major mechanism in the trapping and flux regulation of chemical elements. Until now, the properties of Fe minerals have been very poorly described in riverbanks. The goals of the present work are to identify Fe speciation in riverbanks where ferric deposits are observed and to determine their impact on the metal behavior (As, Co, Cu, Ni, Pb, Zn, etc.). At the surface, Fe speciation is mainly composed of small poorly crystalline Fe phases, i.e. ferrihydrite (~30%), Fe-OM associations (~40%) as well as crystalline Fe phases, i.e. goethite (~35%). At the subsurface, the Fe distribution is dominated by goethite (~35%) and Fe-mica (~35%), the proportion of which increases at the expense of ferrihydrite and the Fe-OM associations. At the riverbank surface, ferrihydrite and the Fe-OM associations are therefore the main Fe hosting phases in response to (i) the fast Fe(II) oxidation induced by the presence of O and (ii) the high amount of OM favoring the formation of nano-phases bound to OM (Fe monomers, polymers and nanoparticles) and preventing mineralogical transformation (ferrihydrite into goethite). During the high-water level period (high flow), a strong erosion of the riverbank transfers these ferric deposits into the river. However, the physicochemical parameters of the river (pH 6.6-7.6 and continuous oxic conditions) do not promote the dissolution of Fe oxyhydroxides and OM. Ferric deposits and the associated trace metals are therefore maintained as colloids/particles and are exported to the outlet. All of the results presented here demonstrate that the ferric deposits trap metals on a seasonal basis and are therefore a key factor in the mobilization of metals during riverbank erosion by river flow.