The origin of compositional variations in kimberlites based on comparative petrology and geochemistry of samples from four cratons

Abstract The term ‘kimberlite’ describes rocks that span a large mineralogical variety including enrichments in mica, carbonates, perovskite, spinel and/or ilmenite. The origin of these compositional variations is addressed here by comparing the petrography, mineral chemistry and bulk-rock as well as groundmass geochemistry of seven representative kimberlite samples (from Wesselton in South Africa; Karowe in Botswana; Diavik and Gahcho Kué in Canada; Majuagaa in Greenland, and Letšeng in Lesotho). These samples exhibit a broad range of mineral and bulk geochemistry covering the whole kimberlite spectrum. Bulk-groundmass compositions are variously enriched in Si, K, Ti, CO 2 and H 2 O depending on the dominant groundmass mineralogy – e.g., high K in mica-rich samples. Interaction with mica and ilmenite-bearing lithospheric mantle appears to be the driving factor of K (± Al) and Ti enrichment, respectively. Degassing controls CO 2 , and higher SiO 2 in the melt derived from assimilation of lithospheric pyroxenes leads to a decrease in CO 2 solubility. Serpentinization by deuteric and/or crustal fluids governs H 2 O concentrations, generally exceeding the H 2 O solubility in kimberlitic melts at upper crustal conditions. Even where the groundmass composition closely approximates predicted kimberlitic melts such as at Majuagaa, the low contents of Na require substantial loss of alkalis via fluids during ascent and emplacement. Thus, compositional variations in erupted kimberlites reflect the combination of asthenospheric source variability, lithospheric assimilation, crystallization, degassing and interaction with deuteric and crustal fluids.