Diffusively driven Li isotope fractionation
Résumé
The cycling of Li at the Earth's surface results in marked isotopic fractionations (~40) that lead to the subduction of isotopically heavy Li. This scenario suggests that Li isotope ratios should provide a distinctive tracer of subducted components in oceanic basalts. Such promise is based on the tacit assumption of equilibrium, in which only low temperature processes can significantly fractionate stable isotopes. However, several recent experimental and observational studies have highlighted that Li isotopes are readily fractionated in magmatic processes by diffusion. In order to use Li isotopes as a tracer of recycled material it is thus critical to understand the role of diffusion on Li isotopes at different length scales. We have examined Li isotopic profiles in phenocrysts of lava flows. Except in glassy samples, we commonly find zoned crystals. Such isotopic zoning appears to be the natural consequence of cooling, during which the partition coefficient of Li in phenocrysts increases. We have reproduced the forms of isotopic profiles in a self-consistent model driven only by cooling. Deriving absolute cooling rates depends on several paramaters which are currently poorly constrained (e.g. the diffusivity of Li in olivine), but the data plausibly imply temperature drops of a few hundred degrees in hours. The influence of diffusion on the Li isotopic composition of a whole lava flow, rather than its redistribution during final cooling, is more difficult to determine. The general over-lap of (unzoned) olivines from peridotitic xenoliths with many mantle derived melts suggest that whole rock compositions are not ubiquitously fractionated by diffusion. As an empirical assessment of possible diffusively driven differences in bulk Li isotopic compositions, we have analysed a suite of samples from the East Pacific Rise (9-10°N). These samples show near homogeneity in long-lived radiogenic isotope tracers but contrasting U-series disequilibria, suggesting their derivation from different depths of the melting regime. This should give rise to differential chemical gradients between the melts and mantle through which they finally ascend, potentially driving different diffusional loss of Li. Although the sample suite shows a significant range in ..7Li related to disequilibrium, its magnitude is small (0.7), suggesting that the role of diffusion in influencing these lava compositions is minor.