A melting model for variably depleted and enriched lherzolite in the plagioclase and spinel stability fields

Abstract

Here we develop a lherzolite melting model and explore the effects of variations in mantle composition, pressure, temperature, and H[subscript 2]O content on melt composition. New experiments and a compilation of experimental liquids saturated with all of the mantle minerals (olivine, orthopyroxene, clinopyroxene, plagioclase and/or spinel) are used to calibrate a model that predicts the temperature and major element composition of a broad spectrum of primary basalt types produced under anhydrous to low H[subscript 2]O-content conditions at upper mantle pressures. The model can also be used to calculate the temperature and pressure at which primary magmas were produced in the mantle, as well as to model both near-fractional adiabatic decompression and batch melting. Our experimental compilation locates the pressure interval of the plagioclase to spinel transition on the solidus and shows that it is narrow (∼0.1 GPa) for melting of natural peridotite compositions. The multiple saturation boundaries determined by our model provide a method for assessing the appropriate mineral assemblage, as well as the extent of the fractional crystallization correction required to return a relatively primitive liquid to equilibrium with the mantle source. We demonstrate that an inaccurate fractionation correction can overestimate temperature and depths of melting by hundreds of degrees and tens of kilometers, respectively. This model is particularly well suited to examining the temperature and pressure of origin for intraplate basaltic volcanism and is used to examine the petrogenesis of a suite of Holocene basaltic lavas from Diamond Crater in Oregon's High Lava Plains (HLP).