Library

Notes: Abstract  The use of ocean floor basalt chemistry as a tool to investigate mantle composition and processes requires that we work with basalts that have been modified little since leaving the mantle. One source of such basalts is melt inclusions trapped in primitive crystals. However, obtaining information from these melt inclusions is complicated by the fact that melt inclusions in natural basalts are essentially always altered by post-entrapment crystallization. This requires that we develop techniques for reconstructing the original trapped liquid compositions. We conducted a series of experiments to reverse the effects of post-entrapment crystallization by re-heating the host crystals to their crystallization temperature. For these experiments we used plagioclase crystals separated from a single Gorda Ridge lava. The crystallization temperature for these crystals was determined by a set of incremental re-heating experiments to be ∼1240–1260° C. The inclusions are primitive, high Ca-Al basaltic melts, saturated with plagioclase, olivine and Al-rich chromite at low pressure. The inclusion analyses can be linked to the host lava composition by low pressure fractionation. The major element composition of the re-homogenized melt inclusions within each crystal is relatively constant. However, the incompatible element analyses have extremely wide ranges. The range of La/Sm and Ti/Zr from inclusions analyzed from a single sample from the Gorda Ridge exceeds the range reported for lavas sampled from the entire ridge. The pyroxene compositions predicted to be in equilibrium with the melt inclusion trace element signature cover much of the range represented by pyroxenes from abyssal peridotites. The volumetric proportions of the magmas entering the base of the crust can be evaluated using frequency distribution of melt inclusion compositions. This distribution suggests that the array of magmas was skewed towards the more depleted compositions, with little evidence for an enriched component in this system. This pattern is more consistent with a dynamic flow model of the mantle or of a passive flow model where the melts produced in the peripheral areas of the melting regime were not focused to the ridge.