HR: 17:45h
AN: V34A-08 INVITED     [Abstracts]
TI: Melt of subducting sediment and basalt is a component in arc magmas worldwide
AU: * Kelemen, P B
EM: peterk@ldeo.columbia.edu
AF: Lamont Doherty Earth Observatory, Geochemistry P.O. Box 1000, Palisades, NY 10964 United States
AU: Hanghoj, K
EM: khanghoj@ldeo.coulumbia.edu
AF: Lamont Doherty Earth Observatory, Geochemistry P.O. Box 1000, Palisades, NY 10964 United States
AU: Hacker, B
EM: hacker@geol.ucsb.edu
AF: University of California, Santa Barbara, Geological Sciences & Institute for Crustal Studies Webb 2120, Santa Barbara, CA 93106-9630
AB: As part of a recent review of arc geochemistry [1] we performed a simple mass balance to calculate the composition of subduction zone aqueous fluids, using recent high pressure fluid/rock partition coefficients and estimates for H2O content in the subducting plate. We compared these fluid compositions to a global compilation of primitive arc magmas (molar Mg/(Mg+Fe), or Mg#, greater than 0.6), and to calculated compositions of partial melts of metasediment and metabasalt in eclogite facies. Like previous workers, we found that arc enrichments relative to MORB, for concentrations of trace elements such as Th and light REE that are relatively `immobile' in aqueous fluids, cannot be easily explained by addition of aqueous fluids to arc mantle, but are consistent with addition of a partial melt of subducting material. This mass balance approach is limited, because Th and light REE could be gradually leached from subducting material in abundant, focused fluids, transported to some locus in the mantle wedge, and gradually accumulated along an interface where fluids are consumed by reaction. A more compelling result, again anticipated by previous workers, is that trace element RATIOS in the subduction component, including `mobile' Ba, Sr and Pb as well as `immobile' Th and light REE, (e.g. Ba/Th) can be explained by addition of a partial melt from subducting material in eclogite facies, and are inconsistent with addition of these elements via aqueous fluid transport. These results may seem somewhat at odds with influential work separating a basalt fluid component with high Ba/La and low Th/La from a sediment melt component with low Ba/La and high Th/La. However, we examine covariation of, e.g., Ba, Th and La concentration in primitive lavas, rather than trace element ratios in evolved lavas. The high Ba/La component in arcs has relatively low Ba and La concentrations. A high Ba/La fluid component may be present, but it does not explain the main vector of global enrichment of Ba or La in primitive arc lavas compared to MORB. Recent research shows that thermal models permit, but do not require, fluid-saturated melting in most subduction zones [2]. Ultra-high P (UHP) rocks subducted to > 2 GPa record peak T at, or above, fluid-saturated solidii for metasediment and metabasalt. Their transfer from footwall to hanging wall limits the utility of exhumed UHP samples for tracking Benioff zone conditions, but illustrates possibilities for isobaric heating, partial melting and/or mantle wedge diapirism of subducted metasediment and metabasalt [1,3]. A complication is that there may be a continuous gradation, rather than a solvus, between aqueous fluids and hydrous melts in equilibrium with metasediment and metabasalt in subduction zones beneath arcs [e.g., 4]. However, as outlined above, the subduction component in arc magmas must have partitioning properties similar to those of hydrous melt, and different from those of aqueous fluid. For these reasons, and to avoid confusion with supercritical NaCl-H2O fluids at lower pressure and temperature, if there is supercritical liquid in subduction zones beneath arcs, following [4] we favor calling it `supercritical melt', rather than `supercritical fluid'. 1. Kelemen et al TOG03 2. Van Keken et al G302, Kelemen et al AGU03, Conder PEPI04, Kincaid & Griffiths G304 3. Gerya & Yuen EPSL03, Gerya et al Geol04 4. Schmidt et al EPSL04
DE: 8410 Geochemical modeling (1009, 3610)
DE: 8412 Reactions and phase equilibria (1012, 3612)
DE: 8413 Subduction zone processes (1031, 3060, 3613, 8170)
SC: Volcanology, Geochemistry, Petrology [V]
MN: Fall Meeting 2005