New constraints on archetypal South African kimberlite petrogenesis from quenched glass-rich melt inclusions in olivine megacrysts

Abstract

The evaluation of primary kimberlite compositions is hindered by significant melt modifications during ascent through the lithosphere by entrainment of xenolithic material, volatile degassing, and near surface alteration. Consequently, hypabyssal kimberlite emplaced in the upper crust may not provide a true reflection of the primary kimberlite magma. This contribution places new constraints on kimberlite melt composition by providing an assessment of quenched glass-rich polymineralic melt inclusions hosted in olivine megacrysts from the Monastery kimberlite, South Africa. Melt inclusions predominantly contain variable proportions of euhedral or skeletal grains of calcite, phlogopite, spinel, perovskite, serpentine, and fresh to devitrified glass. Estimates of the bulk compositions of melt inclusions, and the compositions of crystalline phases present therein, are compatible to those of hypabyssal kimberlites worldwide and show a volatile-rich (CO2 + H2O ~10–17 wt%) carbonated silicate (SiO2 ~27–41 wt%) composition. The glass component has a Si-Mg-Fe-rich and largely CaO-, K2O- and TiO2-free major element composition and is REE-depleted. It also contains approximately 10 wt% H2O but is CO2−free. The glass represents a residual melt that existed after crystallization of the observed mineral assemblage. From some, but not all melt inclusions, apophyses radiate outwards. These fractures typically contain partially devitrified glass that is compositionally identical to the fresh residual glass within the melt inclusions, indicating fracture formation during decompression of the hosting megacryst and at a stage after the melt had evolved. These features are consistent with a trapping of the melt inclusions at high pressure, prior to kimberlite ascent to the surface, in the SCLM at a depth corresponding to 4.5–6 GPa. Textures and compositions of phases within the melt inclusions represent stages of the kimberlite melt and magma evolution. They provide evidence in support of high-pressure experimental studies suggesting a carbonated silicate primary melt rather than a carbonatite. Furthermore, the composition of fresh glass in the melt inclusions, which is compositionally similar to serpentine, suggests that much groundmass serpentine in hypabyssal kimberlites may have formed from similar silicate melt or devitrified glass.