Effects of water, depth and temperature on partial melting of mantle-wedge fluxed by hydrous sediment-melt in subduction zones

This study investigates the partial melting of variable bulk H₂O-bearing parcels of mantle-wedge hybridized by partial melt derived from subducted metapelites, at pressure–temperature (P–T) conditions applicable to the hotter core of the mantle beneath volcanic arcs. Experiments are performed on mixtures of 25% sediment-melt and 75% fertile peridotite, from 1200 to 1300 °C, at 2 and 3 GPa, with bulk H₂O concentrations of 4 and 6 wt.%. Combining the results from these experiments with previous experiments containing 2 wt.% bulk H₂O (Mallik et al., 2015), it is observed that all melt compositions, except those produced in the lowest bulk H₂O experiments at 3 GPa, are saturated with olivine and orthopyroxene. Also, higher bulk H₂O concentration increases melt fraction at the same P–T condition, and causes exhaustion of garnet, phlogopite and clinopyroxene at lower temperatures, for a given pressure. The activity coefficient of silica (ϒ_SiO2) for olivine-orthopyroxene saturated melt compositions (where the activity of silica, a_SiO2, is buffered by the reaction olivine + SiO₂ = orthopyroxene) from this study and from mantle melting studies in the literature are calculated. In melt compositions generated at 2 GPa or shallower, with increasing H₂O concentration, ϒ_SiO2 increases from <1 to ∼1, indicating a transition from non-ideal mixing as OH⁻ in the melt (ϒ_SiO2 <1) to ideal mixing as molecular H₂O (ϒ_SiO2 ∼1). At pressures >2 GPa, ϒ_SiO2 >1 at higher H₂O concentrations in the melt, indicate requirement of excess energy to incorporate molecular H₂O in the silicate melt structure, along with a preference for bridging species and polyhedral edge decorations. With vapor saturation in the presence of melt, ϒ_SiO2 decreases indicating approach towards ideal mixing of H₂O in silicate melt. For similar H₂O concentrations in the melt, ϒ_SiO2 for olivine-orthopyroxene saturated melts at 3 GPa is higher than melts at 2 GPa or shallower. This results in melts generated at 3 GPa being more silica-poor than melts at 2 GPa. Thus, variable bulk H₂O and pressure of melt generation results in the partial melts from this study varying in composition from phonotephrite to basaltic andesite at 2 GPa and foidite/phonotephrite to basalt at 3 GPa, forming a spectrum of arc magmas. Modeling suggests that the trace element patterns of sediment-melt are unaffected by the process of hybridization within the hotter core of the mantle-wedge. K₂O/H₂O and H₂O/Ce ratios of the sediment-melts are unaffected, within error, by the process of hybridization of the mantle-wedge. This implies that thermometers based on K₂O/H₂O and H₂O/Ce ratios of arc lavas may be used to estimate slab-top temperatures when (a) sediment-melt from the slab reaches the hotter core of the mantle-wedge by focused flow (b) sediment-melt freezes in the overlying mantle at the slab-mantle interface and the hybridized package rises as a mélange diapir and partially melts at the hotter core of the mantle-wedge. Based on the results from this study and previous studies, both channelized and porous flow of sediment-melt/fluid through the sub-arc mantle can explain geochemical signatures of arc lavas under specific geodynamic scenarios of fluid/melt fluxing, hybridization, and subsequent mantle melting.