Determination of small melt fraction peridotite partial melts using re foil melt traps

Abstract

The global record of trace element and isotopic compositions of ocean island basalts (OIB) suggest that they are derived from several distinct reservoirs in the mantle. Major and minor element compositions of OIB also span a relatively large range of compositions when compared with the more nearly uniform basaltic compositions erupted at mid-ocean ridges. This may be due to distinct melting lithologies or conditions of melting beneath different oceanic islands. The upper mantle is widely believed to be composed primarily of peridotite, which likely contributes to the formation of OIB. Melting at most OIB localities is confined to depths beneath the rigid oceanic lithosphere (~100 km) at pressures ~3 GPa where garnet is the stable aluminous phase, and the high incompatible element concentrations in these basalts suggest that melting is limited to low melt fractions. Thus, identifying the composition of low melt fraction partial melts of garnet peridotite at 3 GPa is key to understanding the genesis of OIB. Low melt fraction partial melts of garnet peridotite were produced experimentally at 3 GPa in a piston cylinder apparatus using a new Re-foil melt trap. Melt traps allow partial melts to be segregated from crystalline residues, preventing changes in the compositions of small melt pools that can result from crystallization onto nearby mineral grains while the experiment is quenching. Re-foil melt traps are similar to diamond aggregate and vitreous carbon sphere traps, but rely upon the wetting properties of Re as well as pressure gradients to attract the melt phase. Re traps are easier to polish than diamond traps and, unlike vitreous carbon spheres, are capable of withstanding high temperatures required for peridotite partial melting at 3 GPa Partial melting experiments were performed in Pt/C capsules containing natural peridotite powder, KLB-1, or an oxide mixture approaching the composition of KLB-1. Experiments were performed from 1470-1530 °C and produced melt fractions between 0.059 and 0.235. In contrast with previous peridotite melting studies at 3 GPa, melts in the lowest melt fraction experiments coexisted with garnet. Resulting melt compositions are similar to those found by other peridotite partial melting studies for many components, including: TiO2, Cr2O3, MnO, MgO, and Na2O; however, there are relatively large differences in FeO* and CaO from the melt compositions of Walter (1998). FeO* in Re trap experiments varies from 6.6-7.9%, considerably lower than the Walter (1998) melts which are never lower than 8.8% FeO* at 3 GPa, and CaO varies from 11.4-14.3%, significantly higher than melts in Walter (1998) that vary from 7.7- 10.9%. The Re trap method successfully produces large melt pools that can be analyzed by electron microprobe at melt fractions as low as 0.05, although this may be about the lower limit of its effectiveness. The disagreements with previous experiments in CaO and FeO* contents are likely the result of the heterogeneous character of the melt pools and cast doubt on whether the measured compositions of melts produced in these experiments are true low-degree peridotite melt compositions.