Stress-driven melt segregation and reactive melt infiltration in partially molten rocks deformed in torsion with applications to melt extraction from Earth's mantle

Melt extraction from Earth's upper mantle requires transport of magma from regions of partial melting at depth to the Earth's surface. During its ascent, melt interacts chemically and mechanically with the rock matrix. Melt reduces the viscosity of the partially molten rock compared to that of a melt-free rock. This weakening is a potential mechanism of strain localization that could have significant geodynamical implications. Magma interacts chemically with mineral phases during its ascent, dissolving phases in which it is undersaturated and precipitating phases in which it is oversaturated. Such melt-rock reaction can be a driving force for melt migration. Water and other volatiles also partition into the melt from minerals and are then expelled to Earth's oceans or atmosphere. This process leaves behind stronger dehydrated rocks, and it could be the mechanism by which the oceanic lithosphere (mechanical boundary layer) is formed.;The work presented here is an experimental investigation of several mechanisms that influence the distribution of melt within a viscously deformable partially molten rock. The experiments make use of the torsion geometry, which is a relatively recent addition to techniques used in studies of high-pressure, high-temperature deformation of Earth materials. Three mechanisms are considered, either alone or in various combinations. (1) An applied shear stress causes melt to align and segregate into melt-rich bands with a consistent geometrical relationship to the shear geometry. In Chapter 2, we investigate possible means of scaling the bands that form in experimental samples to Earth's mantle and explore the evolution of melt-rich bands at high shear strain. (2) Interfacial tension driven flow acts to homogenize the distribution of melt within a partially molten sample. Flow of melt, driven by the minimization of interfacial energy within the system, must be coupled with corresponding flow of the solid through compaction/decompaction of the matrix or diffusive transport of matter through the liquid phase (dissolution/precipitation). In Chapter 3, we investigate the evolution of melt distribution during static annealing of a sample that was previously deformed and in which melt-rich bands developed in response to the shear deformation. We compare the experimental results with models of interfacial tension driven flow to determine which mechanisms control the rate of melt redistribution. (3) A melt source that is undersaturated in some component, when coupled with a sink that is rich in that component, will infiltrate into the sink through reactive flow. This reactive flow can develop into an instability in which fingers of high melt fraction propagate into the sink. In Chapter 4 we investigate this process under static conditions, and we investigate the interactions of this process in combination with stress-driven melt segregation.