An experimental study of pyroxenite partial melts at 1 and 1.5GPa: Implications for the major-element composition of Mid-Ocean Ridge Basalts

To better assess the potential role of pyroxenites in basalt generation at mid-ocean ridges, we performed partial melting experiments on two natural websterites and one clinopyroxenite representative of worldwide pyroxenites. The experiments were conducted at 1 and 1.5 GPa in a piston-cylinder apparatus; the microdike technique was used to separate the liquid from the solid phases and to obtain reliable glass analyses even at low degrees of melting. Contrasted melting behaviors were observed depending on the phase proportions at the solidus, especially the abundance of orthopyroxene. (1) If orthopyroxene is abundant, the main melting reaction is similar to the melting reaction in peridotites (clinopyroxene + orthopyroxene ± spinel = liquid + olivine), and the liquids are similar to peridotite-derived melts for most major elements. (2) In the absence of orthopyroxene, the main melting reaction is clinopyroxene + spinel = liquid + olivine, yielding liquids that are strongly depleted in SiO2 in comparison to peridotite-derived melts. This low-SiO2 content can be associated with a high FeO content, a combination usually ascribed to a high average pressure of melting (of a peridotitic source). Because of their higher melt productivities and lower solidus temperatures, 5 wt.% of pyroxenites in a heterogeneous mantle may contribute up to 40 wt.% of the total melt production. (1) In some cases, pyroxenite-derived melts differ strongly from peridotite partial melts, leading to a distinct pyroxenite signature in the average melt (lower alkali and TiO2 contents, lower SiO2, higher FeO and/or lower Mg#). The classical criteria used to select primitive mantle-derived magmas (melt inclusions hosted into high Mg# olivine or MORB glasses with Mg# ≥67) or to track down enriched mantle sources (MORB glasses with high incompatible element contents) must be considered with caution, otherwise melts carrying a pyroxenite signature may be eliminated. (2) In general, however, the major-element signature of pyroxenites should be hardly detectable in the average melt because of the similarity of most pyroxenite-derived melts with peridotite partial melts. This similarity may explain why MORB have relatively uniform major-element compositions, but may have variable trace element and/or isotopic compositions