Li Isotope Variations in Mantle Derived Materials

The heavy Li isotope composition (high δ7Li) of subducting assemblages (1) should be evident in recycled material that returns to the surface in mantle derived melts. We have analysed a wide range of ocean island basalts (OIB) and document Li isotopic variations of ~8‰, well in excess of analytical uncertainty (~0.3‰). Our most extreme values are found in Samoan lavas (δ7Li 11‰). Strikingly, δ7Li values of all OIB head to values higher than fertile mantle nodules and N-MORB which define a mantle baseline around 3-3.5‰. No δ7Li values significantly lighter than this mantle baseline are found in whole rock OIB. This strongly implies that it is not the dehydrated, subduction zone processed oceanic crust that is recycled in many OIB, which Zack et al. (2) have shown to be isotopically light. Subduction-influenced mantle, dragged down with the slab is a more likely candidate for the heavy Li signatures. Despite the coherent picture for the widespread influence of recycled material in the mantle provided by Li isotope analyses in whole rocks, there is considerable complexity evident at the grain scale. Mantle xenoliths show variable olivine-clinopyroxene fractionations indicative of disequilibrium. Our bulk measurements show olivines with δ7Li varying from 3-5‰ but co-existing clinopyroxenes ranging from +7 to -9‰. Moreover, in-situ analyses of Li isotope ratios by ion probe reveal major isotopic zonations (up to 40‰ core-rim variations in a single cm sized crystal) in both lava phenocrysts and mantle xenoliths phases with major element equilibrium temperatures in excess of 900°. Such features appear to result from recent, magmatic redistribution of Li and isotopic fractionation as a result of the faster diffusion of 6Li relative to 7Li. Such diffusive processes provide a potential explanation for extremely light δ7Li (-17‰) of some clinopyroxenes found in mantle xenoliths (3). 1. L. H. Chan, et al., EPSL 108, 151-160 (1992); 2. T. Zack, et al., EPSL 208, 279-290 (2003); 3. Y. Nishio et al., EPSL 217, 245-261 (2004).