Natural evidence of kimberlite-garnet reactions in the upper mantle

Percolation of melts in the Earth’s upper mantle leaves geochemical and mineralogical imprints on both the melts and the mantle rocks that need to be interpreted using the available natural samples. Kimberlites, which have the deepest origin of all terrestrial magmas, provide the unique opportunity to study the melt-rock reactions down to the deepest portions of the cratonic lithospheric mantle. The imprints of these reactions are potentially recorded in the kimberlite-hosted mantle xenoliths and in the kimberlitic melts themselves, but they are often obscured by the processes occurring during and after the kimberlite eruption. Garnet and clinopyroxenes, which are two important constituents of the peridotitic mantle, are more resistant to these processes and are in fact commonly found in kimberlites as xenocrysts derived from the disaggregation of the former mantle rocks. Here we focus on the garnet xenocrysts suite of the Zagadochnaya kimberlite (Yakutia, Russia). This is a good case study to investigate the melt percolation processes in the upper mantle, because these samples preserve the clear evidence of garnet-melt reactions occurred at mantle depths shortly before the kimberlite eruption. All analysed garnet grains are of peridotitic origin and most of them are strongly zoned, showing domains where the original (Ca, Cr)-rich garnet has been replaced by (Ca, Cr)-poor garnet + clinopyroxene + spinel (± phlogopite ± amphibole ± Ti oxides). These domains are often located at the edges of the grains and in some cases extend pervasively throughout the entire grain. The trace element contents of the secondary garnets, which are systematically higher with respect to the primary hosts, are compatible with equilibrium with the Zagadochnaya kimberlite. This observation, combined with other supporting textural and mineralogical evidence (Nimis et al., 2009; Ziberna et al., 2013), indicate that these replacement reactions were driven by melts closely related to the host Zagadochnaya kimberlite and that they occurred at mantle depths shortly before the eruption. The wide spectrum of textural, mineralogical, and major and trace element features of these microxenoliths makes this case study suitable for the application of forward modeling of petrogenetic processes. Therefore numerical simulations of trace element transfer will be tested, which can potentially help to identify the melt-rock reactions that occurred in the mantle shortly before the eruption of the kimberlite.