Abstract:
We present (super 235) U- (super 231) Pa measurements for a suite of well-characterised lavas from the Mariana Islands. There is limited variability in ( (super 231) Pa/ (super 235) U), 1.18-1.55, with the lowest values typically found in the incompatible element depleted samples. We also report the complementary (super 238) U- (super 234) U- (super 230) Th- (super 232) Th analyses, which are in good agreement previous measurements on these same samples. In keeping with this earlier work, we interpret the range of ( (super 230) Th/ (super 238) U) from 0.65 to 1.06 as the result of variable contributions of an enriched, sediment melt component coupled with a near constant flux of a relatively U-rich aqueous fluid, derived from the subducted mafic oceanic crust to the sub-arc mantle. We reassess the composition of the subducting sedimentary assemblage and find that its ( (super 230) Th/ (super 232) Th) was previously under-estimated. We performed quantitative models to reproduce the coupled (super 238) U- (super 230) Th- (super 232) Th and (super 235) U- (super 231) Pa data starting from a mantle source modified by subduction-related components according to traditional assumptions of a Pa (and Th)-free fluid and a sedimentary component in (super 231) Pa- (super 235) U and (super 238) U- (super 230) Th equilibrium. This scenario does not allow the U-series isotope compositions of the Mariana lavas to be reproduced. Successful solutions are instead possible with a sediment melt component with relatively high ( (super 230) Th/ (super 232) Th) approximately 0.9 and a modest (super 230) Th- and (super 231) Pa excess, ( approximately 30%). The need for a (super 230) Th- and (super 231) Pa-excess in the sedimentary component implies that the sediment melted <150 ka before eruption of the lavas that carry this signature and that it melted to sufficiently high degree to exhaust monazite. In contrast, the low ( (super 230) Th/ (super 238) U) of the aqueous fluid component is explained by its derivation from lower temperature, mafic crust that contained residual allanite. The maximum ( (super 230) Th/ (super 232) Th) of this fluid is set by the well-defined array of lavas on an equiline plot and this argues for a fresh, rather than significantly altered MORB protolith. Moreover, in-growth is insufficient to account for the (super 231) Pa excesses observed in the Marianas lavas given a Pa-free fluid. Thus we infer that the combined U-Th-Pa signatures of the Marianas are more strongly controlled by the initial compositions of the slab components than by mantle melting. Creation of U-series disequilibria is strongly favoured by the presence of key accessory phases, such as allanite, monazite and zircon, that are likely to be stable in the slab and we suggest that the variable U-Th-Pa systematics of arc lavas globally may well reflect the contrasting stabilities of accessory phases in subduction zones with different thermal and compositional controls. Abstract Copyright (2012) Elsevier, B.V.