Search for a command to run...
Abstract We consider the rheological behaviour of felsic magma in the zone of partial melting and during subsequent crystallization. We also introduce and combine concepts (mushy zone, percolation theory, granular flow, shear localization) derived from the non-geological literature and apply them to field observations on migmatites and granites. Segregation and transportation of felsic magmas is commonly observed in association with non-coaxial deformation, suggesting that gravity forces have limited influence during magma segregation. Solid to liquid and liquid to solid transitions are shown to be rheologically different, which infirms the concept of a unique rheological critical melt percentage for both transitions. Four stages are examined, which depend on the melt fraction present. (1) A minimum of 8% melt by volume must first be produced to overcome the liquid percolation threshold (LPT) above which melt pockets can connect, thus allowing local magma displacement. Transport of the liquid phase is amplified by deformation toward dilatant sinks and is restricted to a very local scale. This corresponds to partially molten domains illustrated by incipient migmatites. (2) When more melt (20–25%) is present, a melt escape threshold (MET) allows segregation and transport of the melt and part of the residual solid phase, over large distances. This corresponds to segregation and transfer of magma towards the upper crust. (3) Segregation of magma also occurs during granite emplacement and crystallization. In a flowing magma containing few particles (≤20%), particles rotate independently within the flow, defining a fabric. As soon as sufficient crystals are formed, they interact to construct a rigid skeleton. Such a random loose packed framework involves ∼55% solids and corresponds to the rigid percolation threshold (RPT). Above the RPT, clusters of particles can sustain stress, and the liquid fraction can still flow. The only remaining possibilities for rearranging particles are local shear zones, often within the intrusion rim, which, as a consequence, develops dilatancy. This stage of segregation during crystallization is totally different from that of magma segregation during incipient melting. (4) Finally, the system becomes totally locked when random close packing is reached, at ∼72–75% solidification; this is the particle locking threshold (PLT). The introduction of four thresholds must be viewed in the context of a two-fold division of the cycle that generates igneous rocks, first involving a transition from solid to liquid (i.e. partial melting) and then a transition from liquid to solid (i.e. crystallization). Neither transition is simply the reverse of the other. In the case of melting, pockets of melt have to be connected to afford a path to escaping magma. This is a bond-percolation, in the sense of physical percoloation theory. In the case of crystallization, randomly distributed solid particles mechanically interact, and contacts between them can propagate forces. Building a crystal framework is a site-percolation, for which the threshold is higher than that of bond-percolation. For each transition two thresholds are applicable. The present approach, which basically differs from that based on a unique critical melt fraction, expands and clarifies the idea of a first and a second percolation threshold. One threshold in each transition (LPT and RPT, respectively) corresponds to a percolation threshold in the sense of physical percolation theory. Its value is independent of external forces, but relies on the type and abundance of minerals forming the matrix within which melt connectivity is developing. The exact value of the second threshold (MET or PLT) will vary according to external forces, such as deformation and the particle shape.