How deep do common basaltic magmas form and differentiate?

Abstract

This work investigates the potential of experiments on peridotite melting to provide quantitative estimates of the temperature and pressure at which basaltic melts were formed. As the primary chemical signal is systematically obscured by crystal fractionation, the first part of this study is devoted to the analysis of the main differentiation trends in the basaltic series from the major geodynamic sites: mid‐ocean ridges, hot spots (both intraplate and near the ridge axis), continental floods, and orogenic boundaries. Two main sorts of primary magmas can be identified: (1) the parent magmas of mid‐ocean ridge, continental flood, and orogenic tholeiites which are poor in Ti and Fe, rich in Al and, to a lesser extent, Na and (2) the parent magmas of within‐plate ocean island basalts (OIB) which have the opposite characteristics. The importance of plagioclase during postmelting fractionation increases in the following order: OIB, island arc tholeiites (IAT), low‐Ti continental flood basalts (CFB), high‐Ti CFB, and mid‐ocean ridge basalts (MORB) corresponding to differentiation pressure decreasing from >10 to ∼5 kbar. Ocean islands that formed near ridge axis have parent magmas and differentiation conditions intermediate between those of MORB and within‐plate OIB. Along‐ridge variations in TiO2 and Na2O contents in MORB are ascribed to mantle heterogeneities and suggest pervasive infiltration of an OIB component. The chemical compositions of synthetic melts in equilibrium with peridotite have been fitted not as a function of the degree of melting, which is neither available from the sandwich experiments nor independent of the fertility of the source, but as a function of temperature (mean error, 40°C) and pressure (mean error, 2.7 kbar). Although fractionation of multimineral assemblages cannot be accounted for with precision, the effect of individual minerals, plagioclase, clinopyroxene, and olivine can be calculated. Assuming that primary melts are in equilibrium with a mantle olvine Fo90–92, it is suggested that MORB segregated from their mantle source at ∼15 ± 5 kbar. Shield‐stage ocean island basalts (Hawaii, Gough, Réunion) and Solomon Island picrites segregated at 20–30 kbar, whereas even greater depths are suggested for Tahiti lavas. The estimated thickness of the molten layer from which the lavas evolve depends on the degree of melting assumed. It is ∼50 km under the ridges, and 20 km or even less at the base of the lithosphere under intraplate volcanoes.