Evidence for Slab-derived Silicate Melt in the Sub-Arc Mantle

Abstract

The sources of arc-related lavas are commonly modified by metasomatic fluids and/or melts believed to be extracted from parts of the subducting slab (e.g., the basaltic crust and/or overlying sediments). Evidence from phase equilibria, trace-element geochemistry, and volatile contents arc-related lavas and melt inclusions strongly suggest that aqueous fluid derived from subducted basalt is one of these phases. In contrast, the properties and even presence of slab-derived melts in the sub-arc mantle are relatively uncertain. Commonly, a role for slab-derived melt is implicated where arc lavas are enriched in sediment-born and fluid-insoluble trace elements such as Th (e.g., Elliott et al., 1997). However, this evidence is contradicted by or fails to address other constraints; for instance: Solvi between water-rich silicate melts and solute-rich aqueous fluids 'close' at T-P conditions within the range of those in the slab and mantle wedge, suggesting that the dichotomy between 'slab fluid' and 'slab melt' may be poorly posed (e.g., Bureau and Keppler, 1999); furthermore, thermal models of subduction zones suggest that no section of the slab should melt except under special circumstances or through generally unexpected physical effects (Peacock, 1991); finally, experimental and empirical constraints on the expected geochemical properties of slab fluids and melts are inconsistent with one another in several respects, adding ambiguity to trace-element arguments for or against the role of slab melt (e.g., Keppler, 1996; Johnson and Plank, 1999).