Abstract
At slow-spreading mid-ocean ridges the lower oceanic crust is extremely heterogeneous, and its formation must be related to some extents of melt-rock interactions. To constrain the relative contribution of crystallization processes and melt-rock interactions on the geochemical budget and architecture of the slow-spread oceanic crust, I investigated two gabbroic sequences sampled in situ at the Atlantis Massif (AM, 30°N, Mid-Atlantic Ridge, MAR) and the Atlantis Bank (AB, 32°S, 57°E, Southwest Indian Ridge, SWIR), where gabbros are exposed by long-lived detachment faults. I performed (i) a multi-scale petro-structural, geochemical and numerical modeling study of primitive gabbroic rocks drilled at the AM, and (ii) a petrographic and geochemical study of olivine gabbros recovered at the AB. AM was drilled during IODP Expeditions 304/305. The heterogeneous lower oceanic crust recovered at Site U1309 presents discrete intervals of olivine-rich troctolites (Ol-T). They are distinguished by partially dissolved olivines with relatively high Fo (86) and Ni contents (>2000 ppm), and they are characterized by the co-precipitation of high Mg# (86-88) clinopyroxene and plagioclase. These characteristics suggest that Ol-T result from impregnation of an olivine-rich protolith by a melt undersaturated in olivine. The flat geochemical profiles across olivine and adjacent minerals suggest that the composition of the protolith was modified by this impregnating melt. Yet, Ni, Li and Co display extremely variable compositions at constant olivine Mg#, suggesting that they retain the signature of the precursor material. Modeling indicates that these chemical variations are likely inherited from the U1309D harzburgites. Experiments show that the melt distribution and paths in a porous media is controlled by the mineral modes of the host rock. The heterogeneous distribution of orthopyroxene in the precursor harzburgitic mantle locally drives the abundance of impregnating melt, leading to different extents of olivine dissolution, as evidenced by variations in mineral modes and chemistry of the AM Ol-T. Geochemical modeling indicates that the melt percolation and assimilation of about 5% of a mantle protolith can explain the formation of the Ol-T. One consequence of this reactive process in Hole U1309D is the shift of melt compositions toward apparent high pressure fractionation. However, no high pressure chemical signature is observed in MORBs from the AM, while it is recorded in MORBs from the nearly amagmatic region along the SWIR (61°-67°E). AB was drilled during IODP Expedition 360. The recovered lower oceanic crust is dominated by olivine gabbros (75%) and oxide gabbros (20%). The section is in places intensively deformed. Shipboard studies have documented textures of plagioclase assimilation by an invading melt crystallizing clinopyroxene. Compositions of olivine gabbros reach relatively evolved signature (Yb = 3-10 x C1-chondrite; MORB Yb = 19). Previous studies on olivine gabbros from AB showed that their formation can be ascribed to assimilation of an oceanic crust by clinopyroxene-saturated trace element enriched melts. The study of AM and AB reveals melt-rock interactions and mineral assimilation associated to melt transport through the accreting lower oceanic crust. Similar textural and chemical features are observed at Kane (24°N, MAR) and in ophiolite complexes (e.g., Alpine and Appennine ophiolites). These evidences indicate that melt-rock interactions are probably ubiquitous, and contribute to shaping the slow-spread lower oceanic crust. The characterization of gabbroic rocks drilled at Hess Deep suggests that melt-rock interactions may take place also at fast-spreading ridge. Melt transport and associated mineral assimilation processes likely play a major role in the building of the oceanic crust overall. Their contribution to the formation of MORB is likely controlled by melt productivity in the upwelling mantle.