HR: 10:40h
AN: T51H-02    [PDF]
TI: Wet Melting in the Oceanic Mantle
AU: * Plank, T
EM: tplank@bu.edu
AF: Boston University, Dept Earth Sciences, Boston, MA 02215
AU: Kelley, K
EM: kelleyk@bu.edu
AF: Boston University, Dept Earth Sciences, Boston, MA 02215
AB: Mantle melting beneath spreading centers is driven by adiabatic decompression, but the amount of melt generated is a function of both mantle potential temperature (T$_{p}$) and water content. Separating these two effects, while a petrological challenge, is of fundamental consequence to the structure and rheology of the oceanic plate and upper mantle. Back-arc basin spreading centers provide a natural setting to quantify the effects of water and T$_{p}$ on mantle melting. In one view, back-arc basins are like small mid-ocean ridges, where the volume and composition of mantle melts is well explained by T$_{p}$ variations. Viewed another way, back-arc basins tap regions of mantle that have been recently hydrated by subduction, and so should reflect water-fluxed melting systematics. Here we try to reconcile these two views, and quantify the melting systematics of the oceanic mantle. Almost ten years ago, Stolper \& Newman (EPSL, 1994) illustrated a linear relationship between the amount of water (H$_{2}$O$_{o}$) and the fraction of melting (F) in the mantle beneath the Mariana back-arc. Here we extend their approach to several back-arc basins where recent studies have determined water contents in submarine basaltic glasses. We use Ti as a proxy for F after correcting for crystal fractionation, and account for Ti source composition with a model based on Ti/Y variations in mid-ocean ridge basalts (MORBs). We use F then to calculate H$_{2}$O$_{o}$, which varies from low values typical of average MORB mantle (100's ppm) to 0.5 wt% H$_{2}$O. Each back-arc basin forms a distinct, nominally linear trend in F vs. H$_{2}$O$_{o}$, where the F intercept at zero H$_{2}$O$_{o}$ reflects dry, decompression melting driven by T$_{p}$ variations similar to global MORB (T$_{p}$ = 1300 - 1500$\deg$C, in the order of Scotia, Marianas, Manus and Lau). The slopes of the trends also appear to vary with T$_{p}$, with more productive wet melting at higher T$_{p}$ (dF/dH$_{2}$O =25-75 wt% melting/ wt% H$_{2}$O). These systematics should also apply to melting beneath mid-ocean ridges, although melt pooling at low water contents may lead to regional arrays with inverse slopes (Asimow \& Langmuir, Nature, 2003). We have also observed positive linear F-H$_{2}$O$_{o}$ correlations for several volcanic arcs (see Kelley et al., this meeting). For example, Mariana arc basalts broadly extend the back-arc trend, consistent with decompression of mantle with similar T$_{p}$ to the back-arc, but with higher water contents (up to 1 wt% in the mantle source). Thus wet melting beneath ridges, back-arcs and arcs can be approximated with a linear dependence on H$_{2}$O, combined with dry, decompression melting that varies globally as a function of mantle temperature.
DE: 1025 Composition of the mantle
DE: 3640 Igneous petrology
DE: 8120 Dynamics of lithosphere and mantle--general
DE: 8121 Dynamics, convection currents and mantle plumes
SC: Tectonophysics [T]
MN: 2003 Fall Meeting