HR: 13:40h
AN: V33C-01 INVITED     [Abstracts]
TI: Melt - Fluid Dichotomy vs. a Regime Involving Liquids Supercritical With Respect to the Endpoint of the Solidus and Geochemical Consequences for Subduction Zones (or: the end of the ''Melting the Slab'' - Myth)
AU: * Schmidt, M W
EM: max.schmidt@erdw.ethz.ch
AF: ETH Zürich, Inst.Min.Pet., Zürich, 8092 Switzerland
AU: Kessel, R
EM: kessel@vms.huji.ac.il
AF: Hebrew University, Inst.Earth Sci., Jerusalem, 91904 Israel
AB: At crustal pressures, phase relations in natural rock-H₂O systems involve low density aqueous fluids (supercritical with respect to the endpoint of the H₂O melt-gas phase) and/or high density hydrous melts. The wide miscibility gap between these two liquid phases leads to a dichotomy of mobile phases with quite distinct element solubilites and geochemical signatures. As pressure increases, the fluid-melt miscibility gap closes at ever lower temperatures, until the crest of the miscibility gap intersects the ''wet'' solidus at it's endpoint, leaving a single liquid, that has chemical and physical properties continuously evolving with temperature, and which is supercritical with respect to the endpoint of the solidus. These facts are well known and the principal necessary phase diagrams have been completed at latest with Ricci (1951). The question is then, at what conditions would the endpoint of the solidus be relevant for natural rock compositions. We have determined this endpoint by two different methods. In a potassium-enriched MORB, a greywacke, and a metapelite (Schmidt et al, 2004, EPSL) a discontinuous melting reaction where phengite disappears in favor of a quenchable hydrous melt exists to 5 GPa. Continuous dissolution of phengite in all three compositions is observed at 6 GPa, a quenchable K-bearing phase was then not formed. Secondly, in a K-free MORB, we measured the composition of the liquid phase from a diamond trap including H₂O-contents (Kessel et al, 2005, EPSL), again observing classical melting at 5 GPa but a continuously evolving liquid composition at 6 GPa. The subsolidus assemblage consisted always of cpx+gar+coes±ky(+phengite in K₂O bearing bulks), the main differences being omphacitic instead of jadeitic cpx and grossular enriched garnets in MORBs, and of course phase abundances. At 3-5 GPa, melting reactions and initial granitic melt compositions are similar in clastic metasediments and K-bearing MORB, a quartz poor greywacke would have the highest melt productivity. Locating the endpoint of the solidus between 5 and 6 GPa, 900-1000 °C indicates, that (i) at higher pressures, the dichotomy of fluid vs. melt ceases to exist in the oceanic crust, and (ii) that relatively small amounts of CaO, MgO and FeO (in the metapelite) shifts the solidus' endpoint to quite high pressures, i.e. from 1.0 and 1.5 GPa in the SiO₂- and albite-H₂O systems, respectively. The consequences of the solidus' endpoint were investigated by measuring trace element partitioning between cpx-gar-liquid, the latter either an aqueous fluid, hydrous melt, or supercritical liquid. Hydrous melts and supercritical liquids (the latter down to at least 200 °C below the hypothetical extension of the solidus) are almost undistinguishable in their trace element pattern, in particular, both have bulk D^solid/liquid_Th > D^solid/liquid_u, the mobility of Th and Be is even increased in the supercritical liquid (Kessel et al, 2005, Nature). Thus, recycling rates of these elements are not indicative of melting, and in the fast and steep circum-pacific subduction zones, they most likely testify for production of a mobile phase from the subducting crust in the supercritical liquid regime (beyond the endpoint of the solidus).
DE: 1012 Reactions and phase equilibria (3612, 8412)
DE: 1037 Magma genesis and partial melting (3619)
DE: 1065 Major and trace element geochemistry
DE: 3630 Experimental mineralogy and petrology
DE: 8413 Subduction zone processes (1031, 3060, 3613, 8170)
SC: Volcanology, Geochemistry, Petrology [V]
MN: Fall Meeting 2005