12 th International Workshop on Modeling of Mantle Convection and Lithospheric Dynamics August 20 th to 25 th 2011, D¨ollnsee Germany c Authors(s) 2011 Melt extraction at mid-ocean ridges: A story in three acts Laurent G.J. Mont´ esi 1 , Laura B. Hebert 1 1 Department of Geology, University of Maryland, College Park, USA montesi@umd.edu At mid-ocean ridges, lithosphere is created by melt extraction and metasomatism, and the oceanic crust forms as melt collects near the surface. As the presence of melt has both rheological and geochemical consequences for the lithosphere, it is of primary importance to understand the mech- anisms that control melt migration and extraction at mid-ocean ridges. Although melt migration is described rigorously by two-phase transport equations in porous or fractured media [1-3], scaling considerations and geological constraints lead to simplifications. It is possible to capture the essence of melt extraction by considering three principal stages: Stage 1) Melt rises vertically from the zone of melt production to a melt-impermeable boundary, or permeability barrier the base of the thermal lithosphere. At this stage, melt trajectories are sub-vertical. They are controlled by melt buoyancy and the high permeability of the partially molten mantle, in which melt remains connected even for low melt porosity [4]. Decompaction channels induced by melt-rock reaction and/or melt weakening may further increase the effective permeability of the mantle [5-8]. Stage 2) Melt travels long a permeability barrier that forms at a crystallization front [9]. As melt enters the lithosphere and cools, it crystallizes and possibly clogs the pore space [10-12]. A barrier is most likely to form where crystallization rate highest, which, in basaltic systems, occurs at the point of plagioclase ± pyroxene saturation [13]. This location follows approximately 1240 ◦ +1.9z, where z is depth in km [14,15]. As the depth of barrier depends on the thermal structure of the lithosphere, it is generally inclined so that melt, being buoyant, travels and is generally focused toward the ridge axis. If the thermal boundary layer is too thick, deep crystallization may be so slow that the lithosphere can decompact and accommodate the crystallization products leading to a metasomatized zone instead of a barrier [11,13]. This is most likely to occur at ultraslow spreading centers [15]. Stage 3) Melt is extracted to the surface, either because melt reaches a place where the barrier is horizontal and focusing stops, or because it enters a melt extraction zone (MEZ), which may be physically interpreted as the presence of faults and/or dikes leading to rapid lateral and vertical melt migration toward plate boundaries [16,17]. However, if focusing stops where the barrier is too deep, melt may instead crystallize at depth again metasomatizing the mantle at the level of the permeability barrier. Stages 2 and 3 are directly influenced by the structure of the thermal lithosphere, which is itself controlled by segmentation of the ridge axis and spreading rate [16-21]. Thus, it is possible to use along-strike variations in melt deliver at well-studied geological examples to constrain the various parameters controlling each of these stage. Recently, it has been shown that the crust along trans- form faults at fast-spreading ridges is anomalously thick [22], which suggest melt redistribution toward transforms, intra-crustal melt production, or efficient extraction of melt in the transform domain [20]. Specific models of the Siqueiros transform along the East Pacific Rise shows that con- sidering a melt extraction zone explains the presence of thickened crust in the transform domain 1