In Situ Falling Sphere Measurements of the Viscosity of Silicate Melts at High Pressures

Introduction Experimental measurements of viscosity are critical to understanding igneous processes in the Earth’s mantle that control the rate of extraction of melts from partially molten source regions and the time scales for magma mixing. Traditionally, the settling velocity of a highdensity sphere through a molten sample is determined from the initial and final positions of the sphere and elapsed settling time in a quench experiment. This velocity is used to solve Stokes’ law for viscosity. These fall-and-quench experiments have generally been limited to 2.5 GPa in the piston cylinder device (e.g., Refs. 1 and 2). Kushiro [1] measured the viscosity of fully polymerized albite melt between 0.5 and 2 GPa at 1673K and found that the viscosity decreased by a factor of four over this pressure range. Brearley et al. [2] measured viscosities of diopside-albite melts and found that the viscosities of polymerized albite-rich melts decreased between 1 atm and 2.5 GPa, whereas the viscosities of depolymerized diopside melts increased between 0.5 and 1.5 GPa. Brearley et al. also reported a minimum viscosity in Ab25Di75 liquid at 1.2 GPa and 1873K. Mori et al. [3] extended fall-and-quench measurements of albite melt viscosity to 7 GPa in a multianvil apparatus and reported a negative pressure dependence generally consistent with that reported by Kushiro [1]. However, because of uncertainties in quench experiments, it is not feasible to extend these traditional falling sphere studies to higher pressures. The in situ falling sphere technique, in which x-ray radiograph images record the movement of an x-ray opaque marker sphere through a molten sample, is the most promising approach to accurately measuring silicate melt viscosities at high pressures. Recent in situ studies of silicate melt viscosity have produced data on both polymerized [4] and depolymerized [5] compositions that are grossly consistent with the results of fall-and-quench experiments. The main goal of the present study is to use the in situ falling sphere technique to determine the pressure dependence of the viscosity of dacite (67 wt.% SiO2) melt. We use measured viscosities to evaluate the relationship between viscosity and self-diffusion. Methods and Materials We conducted in situ falling sphere viscosity measurements from 1.6 to 7 GPa between 1730 and 1950K, by using the 1000-ton press and the T-25 multianvil apparatus at GSECARS beamline station 13ID-D at the APS [6]. The dacite glass powder used in these experiments was synthesized from laboratory reagents. Following Hazen and Sharpe [7], x-ray-opaque marker spheres were formed by explosively melting 0.1mm-diameter Pt wire with an arc welder. Assemblies were pressurized while cold. The temperature was increased slowly to ~50K below the melting point, then increased rapidly at ~800K per minute to the run temperature. The settling velocity of a Pt sphere was measured in radiographic images of the melt, which were recorded throughout the heating procedure to identify all movement of the sphere in the sample. Figure 1 shows a series of time-lapse images from one experiment. An energy-dispersive x-ray spectrum for the MgO-to-BN pressure standard was collected immediately following the fall of the marker, and the run pressure was determined by using the equation of state for MgO [8].