The compositional diversity and the growth of eruptible melt-rich horizons in magma reservoirs is regulated by the processes that control melt-crystal separation. Melt-crystal separation is generally driven by relatively small density differences (in silicic magmas) and the phase separation process must be taking place at a faster or comparable rate to the cooling and crystallization of the magma body to leave an imprint in the rock record. At high melt fraction, crystal settling plays an important role on melt-crystal separation, yet conventional models of settling disregard some important aspects of melt-crystal interactions which affect both the estimated rate of settling and the dynamics at play during settling. In the first part of the talk I will briefly introduce a corrected settling model and describe some of the notable differences when solutions are compared to traditional models. In the second part of the talk I will focus on melt-crystal separation in crystal mushes, as this is the state under which magmas spend most of their lifespan. Compaction in crystal mushes has been invoked to explain the formation of eruptible horizons and chemical differentiation in these systems. However, the rock record provides a contrasting view, careful study of samples suggests that the deformation of the  mush framework does involve little to no intra-crystalline deformation and simple modeling effort looking at compaction rates suggest that these are too slow to affect magma reservoir before their thermal death. Through a combination of laboratory experiments, numerical modeling and field samples analyses (from the Spirit Mountain Batholith, Nevada, see picture) I posit that most of the compaction and melt loss in these mushes result from crystal repacking, i.e. the reorganization of crystals by rotation and translation. Repacking involves a combination of hydrodynamic resistance (crystal-melt interactions) as well as frictional resistance (crystal-crystal contacts) that is presently parameterized by laws borrowed from granular mechanics. We find that repacking is consistent (in terms of melt loss and development of fabric) with the data analyzed in the field and that it reconciles the abrupt change in compaction rate inferred from different types of experiments on partially molten rocks in the literature. Repacking is several orders of magnitude faster than grain-boundary diffusion in these aggregates which supports the fact that repacking plays a role in the evolution of crustal magma reservoirs.

Click here for more information