Global oceanic basalt sources and processes viewed through combined Fe and Mg stable isotopes

Oceanic basalts show variation in their iron and magnesium isotope compositions. One hypothesis for the origin of this is source variation: radiogenic isotope and trace element abundance studies have long argued that the Earth's upper mantle is geochemically heterogeneous and that subducted crust is a major contributor to this diversity. In contrast, a recent hypothesis posits that stable isotopes record disequilibrium during melt transport and so provide novel insight into the melting process. In this study we investigate the first of these hypotheses, that source heterogeneity explains global Fe-Mg isotope systematics. We compile a global dataset of oceanic basalt Fe and Mg isotopes and complement this with new Fe-Mg isotope data from locations possessing some of the most extreme radiogenic isotope ratios for their setting: ocean island basalts from the Cook-Austral and Society islands and a Mid-Atlantic Ridge basalt. Despite both Fe and Mg isotope systems having the ability to trace recycled crustal material in the mantle, their global systematics are very different in this dataset. The global compilation of primitive oceanic basalts records heavier Fe (higher δ57Fe) isotope compositions than bulk silicate earth (BSE), but a mixture of heavier and lighter Mg isotope compositions than BSE. By employing a coupled Fe-Mg equilibrium isotope fractionation model during mantle melting we show that much of this isotopic variability can be generated by the mixed melts produced by melting of peridotite mantle containing moderate amounts of recycled crust as a discrete lithology. The Fe isotope composition of the melts is controlled by the bulk isotope composition of the recycled crust (expected to be considerably heavier than BSE, but variable). In contrast, the Mg isotope composition is controlled by source mineralogy. Olivine-poor lithologies such as recycled crust are able to generate large Mg isotope fractionations during melting, both positive and negative (± 0.1‰) relative to the mantle source, depending on the presence of spinel, clinopyroxene or garnet. These melt Mg isotope fractionations are consistent with the Mg isotope compositions of mid-ocean ridge basalts generated by variable depths of mantle melting. Our equilibrium model provides a baseline to test hypotheses of Fe-Mg isotope variability in basalts: our results show that contributions from recycled crust-derived melts, generated in spinel-, pyroxene-, and garnet-bearing mineral assemblages in t