Experimental Determination of the H2O-undersaturated Peridotite Solidus

Abstract:

Knowledge of the H₂O-undersaturated lherzolite solidus places important constraints on the process of melt generation beneath oceanic spreading centers. While it is generally accepted that the small concentration of H₂O (~50-200 ug/g) dissolved in the oceanic upper mantle has a strong influence on the peridotite solidus, but this effect has not been directly determined through experiments. This is because (1) precisely controlling low concentrations of H₂O in high-pressure melting experiments is thought to be difficult, (2) small amounts of melt are difficult to identify, and (3) the size of mineral grains that grow in near-solidus experiments is too small to be analyzed for H₂O by either Fourier transform infrared (FTIR) spectroscopy or secondary ion mass spectrometry (SIMS). We have developed an experimental approach for determining the peridotite solidus as a function of H₂O content that overcomes these difficulties. Our approach utilizes large (~300 um diameter) spheres of San Carlos olivine to monitor the concentration and behavior of H₂O in our experiments.. The spheres are mixed in 5:95 proportions with a synthetic peridotite that has the composition of the depleted MORB mantle of Workman and Hart (2005). Partial melting experiments are conducted in is a piston cylinder device using pre-conditioned Au₈₀Pd₂₀ capsules. During an experiment, the H₂O content of the San Carlos olivine spheres diffusively equilibrates with the peridotite matrix. After each experiment, the concentration of H₂O dissolved in the olivine spheres is determined by secondary ion mass spectrometry. By analyzing the H₂O content of the San Carlos olivine spheres and performing a simple mass balance, we can then calculate the amount of H₂O in the capsule. The spheres also provides a means to determine the solidus temperature due to the strong partitioning of H₂O into silicate melt compared to olivine, pyroxene, and spinel. When a small amount of melt is present the H₂O partitions into the melt and the H₂O content of the spheres drops in a predictable fashion. The H₂O-undersaturated solidus indicated by our experimental results is in agreement with previous experimental determinations of the nominally anhydrous solidus. This suggests that the potential temperature of the oceanic upper mantle is higher than existing petrologic estimates.