The elusive Congo craton margin during Gondwana breakup: Insights from lithospheric mantle structure and heat-flow beneath the Xaudum kimberlite province, NW Botswana

The continental lithospheric mantle (CLM) beneath the southern margin of the Congo craton has remained elusive mainly because of thick Phanerozoic sedimentary cover concealing possible kimberlite and lamproite diatremes. In this study, we explore this lithospheric mantle section by using major and trace element compositions of mantle-derived clinopyroxene and garnet xenocrysts from kimberlites of the ca. 84 Ma Nxau Nxau cluster in NW Botswana, which is part of the poorly known Xaudum kimberlite province extending into northern Namibia. We utilize these data to better understand the thermal and compositional evolution of the lithospheric mantle at the southern margin of the Congo craton. The clinopyroxene population (83 individual grains) comprises Cr-rich and Cr-poor diopsides with variable major (Al2O3, Na2O, Mg#) and incompatible trace element (U, Th, Zr, Hf, Nb, Ta, REEs) compositions. The large garnet population studied (496 individual grains) is dominated by lherzolitic G9 (38%) and `megacrystic´ G1 (41%) compositions, with minor contributions from Ti-metasomatized G11 (7%) and eclogitic G3 (6%) cratonic mantle sources. Harzburgitic G10 garnet is very rare (two grains only), consistent with a lherzolite-dominated CLM section in a craton margin position. The eclogitic garnet population has compositions akin to garnet from high-Mg cratonic mantle eclogite xenoliths, and such compositions have recently been interpreted as metasomatic in origin within the mantle xenoliths literature. Pressure–temperature calculations using the single-grain clinopyroxene technique reveal a relatively cold cratonic geotherm of 37-38 mW/m2 for the study region during the Late Mesozoic. For peridotitic garnets, projections of calculated Ni-in-garnet temperatures onto the independently constrained regional conductive geotherm suggest that lherzolite dominates at <145 km depths, whereas high-Ti lherzolitic G11 garnets and `megacrystic´ G1 garnets originate mostly from greater depths, down to the lithosphere base at 150 to 210 km depth. The apparent confinement of ´megacrystic´ G1 garnet to the bottom of the lithosphere suggests formation from infiltrating asthenosphere-derived proto-kimberlite liquids during melt–rock interactions. In general, the data suggest that the CLM beneath NW Botswana is depleted to about 145 km depth, and between 145-210 km depths a thick metasomatized layer is identified, representing the transition into the underlying asthenosphere. A relatively thin lithosphere beneath NW Botswana is consistent with the proposed craton margin setting, especially when compared to the thicker cratonic roots beneath the central regions of the Congo and Kalahari cratons in Angola and South Africa, respectively, reaching down to 250 km depth and possibly even deeper. The compositional dissimilarity between the deepest-derived garnets from kimberlites in NW Botswana (i.e., from the diamond stability field) and garnets that occur as inclusions in diamond from cratons worldwide suggests extensive overprinting of the lowermost cratonic lithosphere by oxidative melt-related metasomatism. This finding, together with the very low diamond grades of the Xaudum kimberlites, points to a diminished diamond potential of the large and mostly unexposed ‘cratonic’ region (e.g., covered by thick desert sand) located between the major diamond mining districts of the Congo craton to the north (e.g., Catoca) and the Kalahari craton to the south (e.g., Orapa and Jwaneng).