Melt flow and hypersolidus deformation in the lower ocean crust: Preliminary observations from IODP Leg 345

Models for the construction of the fast-spreading lower ocean crust include the gabbro glacier model (GGM), in which most crystallization occurs within a shallow melt lens and the resulting crystal mush subsides downwards and outwards by crystal sliding. Second, the Sheeted Sill Model (SSM) predicts magmatic injection at many levels in the crust, and requires rapid cooling of the lithosphere. A second set of models seeks to reconcile the relatively unevolved nature of most MORB with the existence of an extensive lower crust with both layering (in the lower crust) and highly evolved gabbros (in the upper plutonic sequence). The mechanisms involved here are melt aggregation during vertical porous flow in the lower crust as opposed to lateral sill injection and in-situ crystallization. Here we report new observations from IODP Expedition 345 to the Hess Deep Rift, where propagation of the Cocos Nazca Ridge (CNR) into young, fast-spreading East Pacific Rise (EPR) crust exposes a dismembered lower crustal section. Drilling in ~4850 m water depth produced 3 holes of 35 to 100 mbsf with ~30% recovery of primitive (Mg# 79-87) plutonic lithologies including troctolite, olivine gabbro, and olivine gabbronorite, showing cumulate textures found in layered mafic intrusions and some ophiolite complexes including: 1. Spectacular modal layering 2. Orthopyroxene very early on the liquidus compared to canonical MORB. 3. Delicate large (2-5 cm) skeletal and hopper structures in olivine. 4. Oikocrystic clinopyroxene enclosing chadacrysts different from the host assemblage. These complex relationships are only hinted at in the existing observations from the ocean floor, and will require significant lab study, however some preliminary inferences can be drawn from the petrographic observations. First, the textures observed in olivine throughout the cores are consistent with rapid crystallization, possibly due to steep thermal gradients in the lower crust. They occur early in the crystallization sequence and survive until crystallization is completed, suggesting that limited hypersolidus deformation occurred during their crystallization. This inference is essentially incompatible with the prediction of substantial hypersolidus flow implicit in the GGM Second, while some authors (Korenaga and Kelemen, 1998) suggest that widespread layering is incompatible with pervasive melt flow models, some of the banding observed is exactly consistent with predictions of such models (Sanfilippo and Dick, 2013). Third, orthopyroxene early on the liquidus cannot be achieved simply by lower crustal processes. High level melt-rock reaction in the mantle combined with dry crystallization in the lower crust may do so, however this requires that some melts crossing the MOHO are not aggregated.