The Effect of Water on Partial Melting in the Upper Mantle.

This thesis presents experimental constraints on incipient melting of mantle peridotite under hydrated conditions. High P-T experiments were performed at pressures of 3 to 13 GPa, and at temperatures of 1200--1450C. These experiments measure mineral/melt H2O partitioning and storage capacity of peridotite components, as well as determine melting phase relations and the compositions of partial melts and residues of hydrated peridotite. Incipient melt H2O concentrations are estimated by peridotite/melt H2O partitioning ( Dperidotite/meltH2O ). To parameterize Dperidotite/meltH2O, mineral/melt H2O partition coefficients were determined for all crystalline phases of the peridotite solidus assemblage (Chapter 2). Combining these Dmineral/meltH2O values with corresponding modal abundances along the solidus yields a Dperidotite/meltH2O of 0.005--0.010 from 1 to 5 GPa, which is dependent on pressure due to varying garnet and pyroxene modal abundances, and to variable pyroxene Al content. This Dperidotite/meltH2O range predicts that incipient melts of MORB source (50--200 ppm H2Obulk) and OIB source (300--1000 ppm H2Obulk) upper mantle contain 0.5--3.8 wt.% and 3--20 wt.% dissolved H2O, respectively. The amount of dissolved H2O in incipient melt dictates hydrous solidus depression, Delta T, which ultimately controls the stability of hydrous melts at P and T. This DeltaT-H2 Omelt relationship was investigated at 3.5 GPa by partially melting hydrated peridotite from 1200--1450C (Chapter 3). Mass balance of phases allows for determination of melt fractions (F) from experiments, as well as estimation of H2Omelt. Delta T values are quantified as the difference in melting temperature between dry and wet peridotite at a particular F. Parameterization of DeltaT as a function of H2Omelt predicts that solidus melts with 1.5, 5, 10, and 15 wt.% dissolved H 2O generate DeltaT values of 50, 150, 250, and 300C, respectively. Combination of this paramterization with Dperidotite/meltH2O (Chapter 2) insinuates that 500 ppm H2Obulk is necessary to stabilize melt across the observed seismic low velocity zone (LVZ) beneath oceanic lithosphere, which is significantly greater than the MORB source upper mantle H2Obulk of 50--200 ppm. This observation argues against suggestions that hydrous melting is solely responsible for the LVZ. At higher pressures the aforementioned parameterizations are difficult to constrain experimentally, but the onset of hydrous melting can be determined by the peridotite H2O storage capacity, defined as the maximum H2O concentration that peridotite can store without stabilizing a hydrous fluid or melt. A new method of determining a minerals H2O storage capacity is employed, in which a hydrated monomineralic layer is equilibrated with a layer of hydrated peridotite and a small amount of melt (Chapter 4). Experiments were carried out at conditions near the 410 km transition zone (TZ) depth to investigate hydrous melting due to the H 2O storage capacity contrast between the TZ and upper mantle. Measured olivine and orthopyroxene H2O storage capacities, combined with estimates of garnet H2O storage capacity, and P-dependent lherzolite modes, yields a peridotite H2O storage capacity of 700--1100 ppm directly above 410 km. This is not consistent with pervasive melting above 410 km, as this range is several times greater than MORB source upper mantle H2Obulk. However, regional melting in areas such as H2O-rich OIB source, or areas of recent subduction may likely occur, leaving residues with ∼1000 ppm H2Obulk.