Experimental settling, floatation and compaction of plagioclase in basaltic melt and a revision of melt density

Centrifuge-assisted piston cylinder experiments were conducted on plagioclase in basaltic melt at 1140–1250 °C, 0.42–0.84 GPa and mostly 1000 g. One set of experiments assesses the settling velocity of a dilute plagioclase suspension; a second sinks or floats plagioclase in a MORB-type melt exploring conditions of neutral buoyancy; and a third set examines floatation of plagioclase from an evolved lunar magma ocean composition. A compaction rate for plagioclase cumulates is established. The experiments demonstrate that neutral density of plagioclase An74 in a MOR-type tholeiitic basalt occurs at 0.59 ± 0.04 GPa (1200 °C), contrasting predictions by present models on melt density which yield a density inversion pressure at 0.10–0.15 GPa. In nature, the level of neutral buoyancy depends on melt composition; nevertheless, for the onset of plagioclase crystallization in dry tholeiitic basalts, our result is robust. As the molar volume of plagioclase is well known, the experimentally determined pressure of neutral buoyancy indicates a correction of -1.6% to previous density models for silicate melts. It follows that for (tholeiitic) layered mafic intrusions, plagioclase is negatively buoyant for early, relatively primitive, parent melts. In contrast, the extreme Fe enrichment of a fractionating lunar magma ocean leads to melt densities that let anorthite always float. Compaction φ/φ0 of experimental plagioclase cumulates is quantified to φ/φ0 = − 0.0582 log (Δρ·h·a·t) + 1.284, where φ0 is the porosity after settling (67 ± 2%), h the cumulate pile height, a acceleration and φ porosity as a function of time t. Gravitational-driven compaction in tens of m-thick plagioclase cumulate in basaltic magmas reaches down to ~ 40% porosity within hundreds of years, a timescales competing with characteristic cooling times of cumulate layers of mafic intrusions. To achieve plagioclase modes > 80% due to compaction, an additional overload of ~ 100 m (layers) of mafic minerals would be required. Compaction of a lunar anorthosite crust of 35 km to 20% porosity (i.e. ~ 90% plagioclase after crystallization of the interstitial melt) would require 30 kyrs.