U-Series Isotope Studies of Atlantic OIB: Links Between Inferred Rates and Degrees of Partial Melting, and the Influence of (Lithospheric) Source Composition

Abstract

Partial melting processes beneath ocean islands are arguably less well understood than those beneath mid-ocean ridges (MOR), not least because the source composition variations are demonstrably greater, and the thickness of the lithospheric lid influences the depth at which melting ceases. For MORB the average degrees and depths of melting increase with increasing temperature and decreasing ridge depth (Klein and Langmuir, 1987). The degree of 23~ disequilibrium also appears to be greater in MORB from shallower ridge depths, and this has been attributed to more of the melt zone being within the garnet stability field in areas of higher mantle temperatures (Bourdon et al, 1996). By contrast, in OIB, the degree of 23~ disequilibrium tends to be less, and the melt fraction lower, in those OIB associated with high buoyancy flux, i.e. high mantle potential temperature (Chabaux and All~gre, 1994; Sims et al., 1995). Thus, the controls on 23~ disequilibrium appear to be different in MORB and OIB. In intraplate settings partial melting should start deeper in regions of high buoyancy (i.e. high Tp), and the height of the melt zone is controlled by the thickness of the overlying lithosphere, which in turn limits the total amounts of melting in any area. However, in general links between the inferred rates and degrees of partial melting are not well established, and this contribution reviews the results of a number of detailed studies on Atlantic Ocean OIB. U-Th isotope analyses of young volcanic rocks from the Azores (Turner et al., 1997), the Canary Islands, and the Cape Verde Islands, indicate that the majority are out of 23~ equilibrium with (23~ in the range 1.08?1.35. All are associated with relatively low buoyancy mantle plumes, and in contrast to the results from Hawaii (Sims et al., 1995) there is no simple link between the degree of 23~ disequilibria and degree of silica saturation (Fig. 1), although the range in (23~ is greater in the silica undersaturated rocks. The Hawaiian data suggests that there may be a negative correlation between the degree of partial melting and the amount of U-Th disequilibria (Sims et al., 1995). However, there are no obvious negative trends within OIB from the central Atlantic region, despite variations in lithospheric thickness from 20 to 125 km and lithospheric age from 7 to 175 Ma. In detail, the inferred degrees of melting on Lanzarote, for example, range from 1 to 4% and yet there is no correlated change in the degree of 23~ disequilibria. Overall, the Canary Islands, Cape Verdes and Azores plumes have similar buoyancy fluxes (1.0, 1.6 and 1.1 Mg s -1 respectively) and exhibit large 23~ isotope disequilibria consistent with the argument that low buoyancy plumes result in magmas with significant U-Th disequilibria. These plumes also appear to have similar low melt production rates of 0.02, 0.03 and 0.01 km 3 y-1 respectively, despite the location of the Azores close to the axis of the mid-Atlantic ridge. Plumes may rise beneath lithosphere of any age, and so the rates of melt production, and hence 23~ disequilibria, appear largely independent of lithosphere thickness. In contrast, there is a broad negative correlation between lithosphere thickness and the integrated degrees of melting (Haase, 1996). It follows that there should be no link between the 23~ disequilibria, and hence the inferred rates of melt generation, and the integrated degrees of melting, and that is what is can be observed in the Atlantic OIB