Plagioclase Lherzolite Melting: Experimental Constraints on a Primary, High-Alumina MORB From the Southwest Indian Ridge

Abstract

We experimentally investigated the phase relations of a primary basaltic glass erupted on the 10–15°E ultraslow spreading (∼0.4 cm/yr) oblique amagmatic segment of the Southwest Indian Ridge (SWIR). Piston-cylinder experiments were conducted at 9–13 kbar and 1270–1310°C, in accordance with the predicted pressure and temperature (9.5 ± 0.8 kbar and 1290 ± 12°C) of saturation with a mantle lherzolite assemblage using the recently developed multiphase reverse fractional crystallization model ReversePetrogen (RevPet, Krein et al., 2021, https://doi.org/10.1029/2020JB021292). The experiments were multiply saturated with a plagioclase lherzolite assemblage (olivine + augite + orthopyroxene + plagioclase) at 10 kbar and 1290°C, within the uncertainty of the RevPet prediction. We use the forward model Petrogen (Krein et al., 2020, https://doi.org/10.1029/2020JB019612) to find the best fitting mantle composition and melting conditions that could produce glass KN162-9 48-21. This model independently finds a melting pressure and temperature of 9.5 kbar and 1285°C, which agree remarkably well with the RevPet predictions and our experiments. This sample is also significantly more depleted in incompatible trace elements than other glasses from the same ridge segment, suggesting its mantle source experienced prior episodes of depletion. This is consistent with previous work suggesting that ultraslow spreading at the SWIR preserves heterogeneities on a local scale. The variability of erupted compositions largely reflects aggregation and mixing of heterogeneous near-fractional melts produced in the spinel and plagioclase stability fields along similar melting paths and their fractional crystallization products. More generally, plagioclase-field mantle melting does occur at mid-ocean ridges and is preserved in some melting environments.