Les semelles métamorphiques sont des unités d’origine océanique (≤ 500 m d’épaisseur) situées à la base des grandes ophiolites obductées (≤ 20 km d’épaisseur). Ces unités sont caractérisées par un gradient métamorphique inverse, où les conditions de pression (P) et de température (T) de cristallisation augmentent de la base vers le contact avec l’ophiolite sus-jacente : depuis 500±100˚C et 0.5±0.2 GPa jusqu'à 800±100˚C et 1.0±0.2 GPa. Formées et exhumées au cours des 2 Ma suivant l’initiation des subductions océaniques, les semelles sont des témoins directs de leur dynamique précoce. Les assemblages minéralogiques qu’elles portent et leur déformation fournissent des contraintes majeures, et rares, sur l’évolution de la structure thermique et sur le comportement mécanique de l’interface de subduction naissante. Au terme d'une étude pétrologique, (micro-) structurale et expérimentale sur les amphibolites naturelles de la semelle de Semail (Oman, UAE) et synthétisées en laboratoire, nous proposons un modèle où la semelle métamorphique résulte d’épisodes multiples d’accrétion d’unités homogènes en P–T (donc sans gradient métamorphique) au cours des premières étapes de subduction océanique. L’écaillage subséquent résulte de changements majeurs dans la distribution de la déformation, du fait des variations des propriétés mécaniques des roches à l’interface de subduction lors de son équilibration thermique et de l’augmentation au cours du temps de la proportion de sédiments entrant en subduction. Ce modèle rend compte d’une grande complexité thermique et mécanique à l’interface de subduction, encore insuffisamment examinée dans les études numériques actuelles. / Metamorphic soles are m to ~500 m thick tectonic slices welded beneath most large-scale ophiolites (usually ≤ 20 km thick). They typically show a steep inverted metamorphic structure where the pressure (P) and temperature (T) conditions of crystallization increase upward, from the base of the sole (500±100ºC at 0.5±0.2 GPa) to the contact with the overlying peridotite (800±100ºC at 1.0±0.2 GPa). Soles are interpreted as a result of heat transfer from the incipient mantle wedge toward the nascent slab during the first My of intra-oceanic subduction. Metamorphic soles are therefore direct witnesses of petrological processes during early subduction. Their mineralogical assemblage and deformation pattern provide major constraints on the evolution of the thermal structure, on the migration of fluids and on the effective rheology along the nascent slab interface. We present a detailed petrological, (micro-)structural and experimental study, with refined P–T estimates obtained with pseudosection modelling and EBSD measurements, on the garnet-bearing and garnet-free (natural and synthetized) amphibolite. We suggest a new tectonic–petrological model for the formation of metamorphic soles below ophiolites, which involves the stacking of several homogeneous slivers (without any T gradient) of oceanic crust to form the present-day structure of the sole. These successive thrusts are the result of rheological contrasts between the slab material and the peridotites of the upper plate as the plate interface progressively cools. This model outlines the thermal and mechanical complexity of the early subduction dynamics, and highlights the need for more refined numerical modelling studies.

Ophiolites - thrust sheets of oceanic lithosphere that have been emplaced onto the continental margin - provide the opportunity to explore the structure and genesis of oceanic crust. As many ophiolites formed above subduction zones, they also allow for the investigation of mantle wedge and subduction interface processes. This the- sis examines the Oman-United Arab Emirates (UAE) ophiolite, which is the largest and most intensely studied ophiolite on Earth. Three distinct problems are addressed. (1) Recent research has proposed that the architecture and tectonic evolution of the ophiolite in the UAE differs from in Oman. In Chapter 2, I test this hypothesis by integrating new geological mapping and field observations with previously published maps of the ophiolite in the UAE. My results indicate that the ophiolite is gently folded, but otherwise largely intact. I demonstrate that the architecture of the ophi- olite in the UAE is not significantly different from in Oman. Thus, there is no basis for a different tectonic evolution as recently proposed. (2) Observations from exper- iments and small-scale natural shear zones indicate that volumetrically-minor phases can control strain localization. In Chapter 3, I test the hypothesis that minor phases control strain-localisation at plate boundaries. To do so, I analyzed peridotites from the base of the ophiolite, a palaeosubduction interface. My results demonstrate that minor phases limited olivine grain growth, which led to rheological weakening. (3) The mechanisms by which metamorphic soles detached from the downgoing slab and accreted to the hanging-wall mantle is unclear. In Chapter 4, I examine a transect across the metamorphic sole in the UAE. My results reveal that granulite formation was more extensive than is typically considered. I propose that granulite formation resulted in rheological strengthening, which caused the subduction interface to migrate into the downgoing slab and accrete the metamorphic sole.