Experimental constraints on element mobility from subducted sediments using high-P synthetic fluid/melt inclusions

A series of hydrothermal piston-cylinder experiments have been performed to determine the composition of representative fluids and fluid/melt/rock interaction in subduction zones. Experiments were conducted under H2O saturated conditions at 2.2 GPa over a temperature range from 600–750 °C. The experiments contained synthetic, trace-element-doped pelitic starting material and fractured quartz chips to trap and preserve synthetic fluid/melt inclusions. Pelite residues from the subsolidus experiments (600–650 °C) consist of an eclogite-facies mineral assemblage including quartz, phengite, epidote, rutile, garnet, amphibole, apatite, and zircon. Coexisting hydrous fluids are expected to be completely buffered for trace elements by this mineral assemblage. At 2.2 GPa the wet solidus for the pelitic starting material is located at approximately 675 °C and hydrous fluid and melt coexist as immiscible phases at least up to 750 °C. Residue phases in the supersolidus experiments (700–750 °C) are garnet, rutile, and zircon, which suggest that HREE and HFSE are largely retained in slab residues during very high degrees of H2O saturated melting.Laser ablation ICPMS analysis and quantification of trapped fluid inclusions from the experiments indicate that subsolidus hydrous fluids released from subducted sediments have relatively high LILE contents compared to REE and HFSE, but overall are remarkably dilute. Total solute contents are approximately 5 wt.%, of which > 75% is SiO2 and around 15% is Na2O + Al2O3. The experimental results are used to show that subducting sedimentary rocks do not undergo significant element loss during metamorphic dehydration up to eclogite facies. If these fluids are representative of aqueous fluids released at sub-arc depths then simple slab dehydration models may be unable to account for element transfer from the slab to arc magmas. Instead, element recycling through subduction zones may be a product of complex fluid-melt-rock interaction processes involving multiple slab components.