Re–Os isotope systematics and HSE abundances of the 3.5 Ga Schapenburg komatiites, South Africa: Hydrous melting or prolonged survival of primordial heterogeneities in the mantle?

International audienceWe report Re­Os isotope and highly siderophile element (HSE) abundance data for komatiites from the Schapenburg Greenstone Remnant (SGR), South Africa, an equivalent to the lowermost formations of the Onverwacht Group in the Barberton Greenstone Belt (BGB). The Re­Os isotopic data for 13 whole-rock samples define a regression line with an age of 3549 ± 99 Ma, consistent with the ~ 3.5 Ga age of the Onverwacht Group. The immobility of Os during alteration and the correct Re­Os age provide evidence that the initial ?187Os = + 3.7 ± 0.3 derived from the regression reflects a time-integrated suprachondritic 187Re/188Os in the source of the SGR komatiites. The HSE abundances in the emplaced SGR komatiite lavas are 2— to 3— lower than those in other well-studied komatiites. Compared to the Primitive Upper Mantle (PUM) estimate, the calculated mantle source of the SGR komatiites was moderately depleted in HSE and was characterized by a fractionated HSE pattern. The enrichment in 187Os, HSE depletion, and fractionated HSE pattern in the SGR komatiite source could be the result of fluid transport of radiogenic Os from the subducting slab, incorporation in the overlying mantle, and hydrous melting of the modified mantle to produce the komatiites. This would imply that plate tectonic processes operated as early as 3.5 Ga and that at least some komatiites formed via wet melting in island arc settings at relatively shallow depths and temperatures not exceeding 1400 °C. An alternative model would include dry melting of a chemically distinct, majorite-enriched mantle domain formed very early in Earth's history as a result of an initial stratification of the mantle during crystallization of a magma ocean. Preferential partitioning of Re into the majorite-rich domain would have resulted in its acquiring suprachondritic Re/Os. In order to grow the radiogenic Os isotopic composition, the majorite-rich domain would have to have remained isolated from the rest of the mantle for a billion years. If this model is correct, the data would require a deep plume origin for the SGR komatiites at temperatures ~ 300 °C higher than the ambient mantle temperatures. Both models have their shortcomings in reconciling the available geochemical and petrologic data for the SGR komatiites. Because BGB komatiites are geochemically closer to island arc tholeiites and boninites than the SGR komatiites, it will be especially important to determine the Re­Os isotope and HSE systematics of the BGB komatiites and of typical boninites