| The crust of the Earth (and other terrestrial planets) forms as the product of partial melting of the silicate mantle. Most melting occurs at plate boundaries, which are deforming intensely and must be weak in order to develop the style of plate tectonics observed on Earth. The aim of this thesis is to understand the interactions of melt segregation, strain localization, and large-scale weakening. In shear deformation experiments at uppermost mantle pressures and temperatures (300 MPa and 1573 K) on partially molten mantle materials, melt segregates spontaneously into melt-rich bands. Differential stress drives melt segregation and controls the dynamics of self-organization of melt into channel networks. In these experiments, it is shown that (1) the band spacing is controlled by the compaction length of the crystal-melt system, a length scale that describes coupling between solid and fluid flow, (2) melt-band networks have consistent band angles of ∼20° with respect to the shear plane, (3) melt segregation is accompanied by an initial strengthening followed by weakening as strain focuses into the band network, (4) the average orientation of bands corresponds to that which minimizes the pressure difference between the band and non-band regions, and (5) when strain is partitioned between bands and non-band regions the flow fields are modified as recorded in the orientations of crystals. This observation implies that the seismic fast axis (olivine a-axis) may rotate 90° to the shear direction in partially molten regions, contrary to the basic assumption made in interpreting mantle flow from seismic anisotropy. This rotation explains anomalous mantle anisotropy of the Reykjanes Ridge south of Iceland. The configuration of melt-rich networks is scaled to natural conditions to first approximation using the compaction length. However, the effects of experimental boundary conditions complicate estimating effective viscosity reductions. Furthermore, for conditions at which viscoelasticity of the matrix is important, we develop a hypothetical fracture process in which deformation and melt transport are coupled in the brittle-elastic regime. These discoveries provide fundamental constraints for understanding the interactions of deformation and melt migration, and will lead to new insights into the dynamics of partially molten regions of the Earth. |
