Deep fluids in subduction zones

The fluid inclusions preserved in high and ultrahigh pressure rocks provide direct information on the compositions of fluid phases evolved during subduction zone metamorphism, and on fluid–rock interactions occurring in such deep environments. Recent experiments and petrologic studies of eclogite–facies rocks demonstrate that stability of a number of hydrous phases in all rock systems allows fluid transport into the mantle sources of arc magmas, as well as into much deeper levels of the Earth's mantle. In eclogite–facies rocks, the presence of large ion lithophile elements (LILE) and light rare earths (LREE)-bearing hydrous phases such as epidote and lawsonite, together with HFSE repositories as rutile and other Ti-rich minerals, controls the trace element budget of evolved fluids and fluid-mediated cycling of slab components into the overlying mantle. Studies of fluid inclusions in eclogite–facies terrains suggest that subduction mainly evolves aqueous solutions, melts being produced only locally. Eclogite-facies rocks diffusely record processes of fluid–melt–rock interactions that exerted considerable control on the element and volatile budget of subduction fluids. Trace element fractionation during such interactions needs to be tested and quantified in more detail to achieve the ultimate compositions actually attained by fluids leaving off the slab. Variably saline inclusions with minor CO 2 and N 2 are trapped in rock-forming high pressure minerals; brines with up to 50% by weight dissolved solute are diffusely found in veins. The latter inclusions are residues after fluid–rock interactions and deposition of complex vein mineralogies: this evidence suggests increased mineral solubility into the fluid and formation, at a certain stage, of silicate-rich aqueous solutions whose geochemical behaviour and transport capacity can approach that of a melt phase. This is supported by experimental work showing high solubility of silicate components in fluids at high pressures. However, natural examples of inclusions trapping such a fluid and quantitative analyses of its major and trace element composition are not yet available. Fluids in high and very high pressure rocks do not move over large scales and the channelways of fluid escape from the slab are not yet identified. This suggests that only part of the slab fluid is lost and returned to the surface via magmatism; the remaining trapped fraction being subducted into deeper levels of the upper mantle, to renew