NB-TA-RICH MANTLE AMPHIBOLES AND MICAS - IMPLICATIONS FOR SUBDUCTION-RELATED METASOMATIC TRACE-ELEMENT FRACTIONATIONS

We report the trace element compositions of amphibole and mica separated from mantle xenoliths in alkali basalts and analyzed by ICP-MS. Nb and Ta are highly (50-200-fold) enriched in vein amphibole and mica relative to primitive mantle compositions, whereas Th and U are depleted. Some disseminated amphiboles do not have such extreme Nb-Ta enrichments, but Nb-Ta partition coefficients between amphibole and clinopyroxene are remarkably high, ranging from 10 to 85. In apparent contrast with the results on natural mantle amphiboles, recently reported Nb and Ta partition coefficients between amphibole and melts are very low [1,2]. The reason for the apparent contradiction may lie in either the composition of the amphibole or the fluid phase (silica-rich aqueous fluid rather than silicate or carbonate melt). In either case, our observations show that amphibole and mica can be important hosts for Nb and Ta and cannot be ignored in identifying the underlying cause of the nearly universal relative Nb-Ta depletion of subduction-related volcanic rocks. We propose a metasomatic model for creating source regions that are depleted in Nb and Ta relative to Th, U and the LREE. Fluids generated by dehydration of the subducted slab ascend through the overlying mantle wedge and precipitate amphiboles. Highly incompatible elements including Nb and Ta are transferred with the fluid into the wedge where the 'open-system' precipitation of amphibole fractionates the trace elements and thus generates low (Nb,Ta)/(Th,U,LREE) ratios in the residual fluid. As this fluid travels further it either directly induces partial melting in hot regions of the wedge or is consumed through 'closed-system' crystallization of disseminated amphibole in host peridotite that can later undergo partial melting. In either case the resulting source regions of subduction-related magmas are enriched in highly incompatible trace elements but not in Nb and Ta. This model may be considered either as a complement or as an alternative to published models explaining the chemistry of are magmas.