Transportation of H 2O and melting in subduction zones

Abstract

Material recycling in subduction zones, including the generation and migration of aqueous fluids and melts, is key to understanding the origin of volcanism in subduction zones and is also important for understanding the global circulation of materials. Recent knowledge concerning the phase relationships of hydrous peridotitic and basaltic systems allows us to model the fluid generation and migration in subduction zones. Here I present a numerical model, in which the aqueous fluid migrates by permeable flow and interacts chemically with the convecting solid, including melting. The calculation results suggest that nearly all the H 2O expelled from the subducting slab will be hosted by serpentine and chlorite just above the slab, and is brought down by up to 150 km, depending on the temperature along the slab. Breakdown of serpentine and chlorite at these depths results in the formation of a fluid column through which H 2O is transported upwards. The fluid reaches a depth corresponding to a cusp of the H 2O-undersaturated solidus of peridotite (minimum at 2.5 GPa) and initiates extensive melting whose depth and the lateral extent toward the trench side is nearly fixed, irrespective of the age of the slab. This is because the flux of H 2O and the depression of the practical solidus temperature in the mantle wedge are similar for models with different slab ages. Exceptionally, for very young slabs (e.g. ≪10 Myr when the subduction velocity is ∼6 cm/y), different melting regimes occur, such as melting in the forearc region and slab melting. If the aqueous fluid released from the slab migrates upwards in disequilibrium (e.g. through fractures), significant melting occurs in the forearc region, since the serpentinite layer, an effective H 2O-absorber, is not formed. However, this is not the case in most subduction zones. Further studies with various subduction parameters, and melt segregation processes which are not included in this study, are required to compare the model results with the observed distribution and chemistry of arc magmas.