Spatial fractionation of rare earth elements in cratonic mantle: analysis of observations, model, and implications for kimberlite genesis
© Y.M. Khazan, O.V. Aryasova
REE abundances C in clinopyroxene and garnet from xenoliths entrained by kimberlites of Kaapvaal craton (South Africa) and Somerset Island (Arctic Canada), show strong and species dependent variation of abundances vs. pressure in a pressure ran-ge P of 2 to 6 GPa. Along the REE group the slopes of lnC vs. P for the REE abundances in clinopyroxene vary nearly monotonously from a strong decrease of La (in a factor of 50) to a weak increase of Lu. The slopes of the garnet REE abundances are species dependent as well although they are generally much smaller than those in clinopyroxene. As a result, the inter-element ratios of the REE vary across the upper mantle, an effect, which can be referred to as a REE large scale space fractionation. The slopes of the REE abundances strongly correlate with the activation energies for the REE diffusion in both clinopyroxene and garnet. Therefore one can speculate that the space fractionation had accompanied mantle melting events in the craton history and was due to a disequilibrium REE partition between the melt and solid residue. A model of disequilibrium partition during the mantle melting developed in the paper indicates that the observed strong spatial REE fractionation was due to a moderately high degree (~10-20%) partial melting of depleted peridotite (a few wt % of clinopyroxene content) implying that the mantle melting event responsible for the observed fractionation was preceded by another melting resulted in "washing out" of considerable part of clinopyroxene and garnet from the mantle. Dramatic distinction between the depth variation of the clinopyroxene and garnet REE abundances is due maintly to a large difference between the activation energies of the REE diffusion in clinopyroxene (~400 to 500 kJ/mol) and garnet (<300 kJ/mol). The model predicts that a large scale fractionation, comparable to that observed for the abundances of the LREE in clinopyroxene, should also be observed for incompatible species having the diffusion parameters in a mineral-concentrato similar to those for the LREE in clinopyroxene, therefore the model can be verified by observations.