An Experimental Study of Trace Element Fluxes from Subducted Oceanic Crust

We have determined experimentally the hydrous phase relations and trace element partitioning behaviour of ocean floor basalt protoliths at pressures and temperatures (3 GPa, 750-1000°C) relevant to melting in subduction zones. To avoid potential complexities associated with trace element doping of starting materials we have used natural, pristine mid-ocean ridge basalt (MORB from Kolbeinsey Ridge) and altered oceanic crust (AOC from Deep Sea Drilling Project leg 46, ∼20°N Atlantic). Approximately 15 wt % water was added to starting materials to simulate fluid fluxing from dehydrating serpentinite underlying the oceanic crust. The vapour-saturated solidus is sensitive to basalt K2O content, decreasing from 825 ± 25°C in MORB (∼0·04 wt % K2O) to ≈750°C in AOC (∼0·25 wt % K2O). Textural evidence indicates that near-solidus fluids are sub-critical in nature. The residual solid assemblage in both MORB and AOC experiments is dominated by garnet and clinopyroxene, with accessory kyanite, epidote, Fe-Ti oxide and rutile (plus quartz-coesite, phengite and apatite below the solidus). Trace element analyses of quenched silica-rich melts show a strong temperature dependence of key trace elements. In contrast to the trace element-doped starting materials of previous studies, we do not observe residual allanite. Instead, abundant residual epidote provides the host for thorium and light rare earth elements (LREE), preventing LREE from being released (ΣLREE 1500 and La/SmPUM (where PUM indicates primitive upper mantle) ∼1, most closely matching the geochemical signal of arc lavas worldwide, were generated from AOC at 800-850°