Comparative Geothermometry in High-Mg Magmas from the Etendeka Province and Constraints on their Mantle Source

There is still debate whether Large Igneous Provinces (LIPs) are caused by high mantle temperatures induced by thermal plumes or by other factors that enhance melt production from the mantle. A prerequisite for assessing the thermal plume model is a reliable estimate of liquidus temperatures of the magmas produced, preferably based on more than one method of geothermometry. The study reported here compares multiple geothermometers for the Etendeka LIP, which is among the largest Phanerozoic examples and one that shows several features suggestive of a plume origin (continental flood basalt province linked via an age-progressive volcanic ridge to an active hotspot). Magnesium (Mg)-rich magmas emplaced as dikes in NW Namibia are the most primitive rocks known from this province and are thus best suited to determine the composition and melting conditions of their mantle source. Earlier studies of the Etendeka Mg-rich dikes reported high liquidus temperatures based on olivine-melt Mg-Fe equilibria. We extend that work to a larger set of samples and compare the results of olivine-melt Mg-Fe thermometry with other methods based on spinel-melt and spinel-olivine equilibria (Al-in-olivine thermometry), as well as olivine-melt trace-element exchange (Sc/Y thermometry and V oxybarometry). All methods used the same starting assumptions of nominally anhydrous melts and a crystallization pressure of 0.5 GPa. Only mineral-melt or mineral-mineral pairs consistent with compositional equilibrium were used for calculating temperatures. The trace-element compositions of olivine are also used to discuss the relative proportion of peridotite and pyroxenite in the mantle source for these magmas. Twelve dike samples were studied, with whole-rock MgO concentrations ranging from 8.4 to 19.4 wt %. Diagnostic element ratios of transition metals in olivine (e.g., Mn/Fe, Mn/Zn, Zn/Fe) indicate a peridotite-dominated mantle source for the magmas, which is consistent with the other indicators based on whole-rock data e.g., 10 000xZn/Fe, CaO-MgO trend, FeO/MnO and FC3MS (FeO/CaO-3xMgO/SiO2). The temperature variations show a positive correlation with the Fo-content of host olivines, and values from high-Fo olivine agree well with olivine and spinel liquidus temperatures calculated from thermodynamic models of bulk-rock composition. All methods and most samples yielded a temperature range between 1300 degrees C and 1400 degrees C. An exceptional few samples returned temperatures below 1300 degrees C, the minimum being 1193 degrees C, whereas several samples yielded temperatures above 1400 degrees C, the upper range being 1420-1440 degrees C, which we consider to be a robust estimate of the maximum liquidus temperatures for the high-Mg magmas studied. The conversion to mantle potential temperatures is complicated by uncertain depth and degree of melting, but the functional relationship between T-p and primary melt MgO contents, using melt inclusions from olivine phenocrysts with of Fo > 90, indicate a T-p range from 1414 to 1525 degrees C ( 42 degrees C), which is 100-150 degrees C higher than estimates of ambient upper mantle T-p in the South Atlantic today.