The interaction between silicate minerals and C-O-H bearing melts in the Earth's mantle

The experiments in this study were performed to examine the effect of volatile components on the melting of mantle rocks at conditions of the deep upper mantle (~180-400 km). Chemical compositions, including H2O contents, of hydrous melts, compatible with those that would form by small degree melting of an upper mantle peridotite at 180 km depth and near adiabatic temperatures, were determined using a series of iterative crystallization experiments. Experiments were performed in a multianvil apparatus at 6 GPa and 1400 ºC employing a hydrous, natural mantle peridotite, which was equilibrated with a large volume of hydrous melt. The melt composition was iterated in a series of experiments until it was in equilibrium with a complete peridotite assemblage with mineral compositions close to those encountered in a sub solidus experiment at the same conditions. The H2O content of the melt could be accurately determined through a mass balance calculation as a result of the large volume of melt employed. The hydrous melt compositions, when compared on a volatile free basis, are found to be similar to group II kimberlites. A model is proposed, whereby group II kimberlites form from cratonic lithosphere that has been previously melted but then metasomatised resulting in the addition of ~1.7 wt % phlogopite to the rock. Parameters in the model are constrained through the mineral melt partition coefficients of both H2O and K2O, which imply melting at near adiabatic temperatures. The H2O concentrations of mineral phases within the peridotite assemblages were also measured using an ion probe and elastic recoil detection analysis. Using the known melt H2O contents mineral-melt partition coefficients for olivine, clinopyroxene, orthopyroxene and garnet were determined. Using these results the on set and extent of melting at conditions equivalent to 180 km below a mid ocean ridge was examined as a function of bulk mantle H2O concentration. The results indicate that current estimates for the H2O content of the depleted mantle (50-200 ppm H2O) are insufficient to induce mantle melting at these conditions, which requires ~700 ppm wt H2O to produce 0.1 % melting and 1600 ppm for 1 % melting. However, melting can occur at these conditions within the mantle source of ocean island basalts, which is estimated to contain up to 900 ppm wt H2O. If adiabatic temperatures are 200 °C higher within such plume related sources, melt fractions of over 1 % can be reached at 180 km depth. In addition, based on inter-mineral H2O partitioning a model that predicts the distribution of H2O between peridotite mineral phases as a function of depth at H2O undersaturated conditions is presented. The model indicates that for a fixed mantle H2O content of 200 ppm wt, the olivine H2O content will increase with depth solely due to changes in inter-phase partitioning and mineral modes. The results of this model provide an explanation for the reduction in seismic anisotropy observed at depths >200 km. Further experiments were conducted at 6 and 13 GPa in the simplified systems Mg2SiO4-H2O and MgSiO3-H2O to determine the effect of H2O on melting point depression. Melting phase relations were examined using a simple thermodynamic model that supports the almost complete dissociation of H2O to OH- bonded to melt silicate components at these conditions, although results for forsterite at 13 GPa are inconclusive due to uncertainties in the dry melting temperature. This simple model is unsuccessful in predicting the depression of melting temperature due to H2O for peridotite melts, however, indicating that hydrous melting models in simple systems provide little guidance as to the behavior of H2O in natural melt compositions. In addition, H2O partition coefficients were determined as a function of temperature for end member forsterite and enstatite at 6 and 13 GPa. Very little temperature dependence to the partition coefficients was observed except for forsterite at 13 GPa. Experiments performed in the MgO-SiO2-H2O-CO2 system at 13 GPa were used to investigate the effect of CO2 on the mineral-melt H2O partition coefficient. It was found that a melt CO2/(CO2+H2O) (weight) ratio of 0.3 causes the forsterite H2O partition coefficient to decrease by 50 % compared to the CO2-free partition coefficient, due to a significant drop in the measured forsterite H2O content. This effect decreases with increasing temperature and must be caused by strong interaction between CO2 and H2O in hydrous melts. The results imply that at the base of the upper mantle H2O-CO2 bearing melts can form at adiabatic conditions if as little as 250 ppm wt of H2O is present in the bulk mantle. This dramatic effect of CO2 on the mineral-melt partitioning of H2O may explain seismic observations interpreted to be caused by the presence of melt at the base of the upper mantle.

Level thesis:degree_level

thesis.doctoral

Grantor dc:publisher

Universität Bayreuth

Year

2013

Author dc:creator

• Novella, Davide

Contributors dc:contributor

• Frost, D.J.

Repository record source_url

https://epub.uni-bayreuth.de/id/eprint/1799/

OAI identifier oai:identifier

oai:epub.uni-bayreuth.de:1799