Mid-Ocean Ridges and Ocean Islands: Comparing the constraints on melt segregation from U-series data

U. Faul

The isotopes in the Uranium decay series have half-lives from 1600 years (²²⁶Ra), 33,000 years (²³¹Pa) to 75,000 years (²³⁰Th). This range in half-lives is assumed to cover the time scale from the first generation of melt at depth beneath mid-ocean ridges or ocean islands to the eruption at the surface. U series data from mid-ocean ridges and ocean islands shows ubiquitous Th excess (activity ratio (²³⁰Th/²³⁸U) >1), indicating that the parent ²³⁸U is slightly more compatible and therefore has a slightly longer residence time in the melting column. For mid-ocean ridge basalts (MORBs), plots of the activity ratio of (²³⁰Th/²³⁸U) versus the ratio of the absolute abundances of Th and U indicate a correlation with more enriched basalts (higher Th/U wt ratio) having larger Th excess. This correlation can be explained with more enriched source material having a lower solidus temperature and beginning melting deeper in the garnet stability field. Since garnet most strongly fractionates Th from U, the more melt is produced in the garnet stability filed the larger the excess. For the much shorter lived ²²⁶Ra the opposite correlation is observed: the most depleted basalts have the highest Ra excesses. ²³¹Pa excess appears uncorrelated with enrichment. Since the partition coefficients between melt and matrix of all U-series isotopes, but particularly Ra, are small this is taken to imply that porous flow up to the last stages of melting takes place at very low porosities (of order of 0.1%).In comparison with MORBs, ocean island basalts (OIBs) have much smaller Ra and Pa excesses, even though the source material is generally more enriched and fast transport through the lithosphere is indicated by the presence of xenoliths in the erupted lavas. The available data from OIBs show a better correlation between enrichment and ²³⁰Th excess than MORBs and a good correlation between Pa excess and enrichment is apparent whereas there is none for MORBs.OIB disequilibrium data can be modelled by porous flow of the melt with reasonable permeabilities and melt properties, whereas modelling the MORB data by porous flow requires unreasonably high permeability or low melt viscosity. This and the scatter of the data for individual ridge segments possibly indicates that processes in the shallow mantle, such as flow through veined, previously depleted and cooled mantle could play an important role in determining some of the U-series characteristics of MORBs.