Previous page     Contents     Next page

Although the Archimedean method provides a good tool to measure one bar densities of melts at high temperature, little is known about the density of silicate melts at high pressures. Although there have been estimates of silicate melt density at high pressure, based on molecular dynamics simulations and elastic property measurements at low pressure, few direct measurements on silicate melts at high pressure have been conducted until recently. Dynamic determinations using shock wave techniques and static determinations using the "falling sphere" technique (density + viscosity) and the "sink/float" method (density) have been carried out on silicate melts by previous investigators to determine the variation of density as a function of pressure.

Extension of the high pressure density and viscosity data bases is the goal of this project using the "falling sphere" technique. The density is determined by measuring the distance of sinking (or floating) of a crystalline or metal sphere as a function of time and is calculated by applying Stokes´ law.

First attempts were made on a model melt with a basaltic composition (36 mol% anorthite and 64 mol% diopside). Such a composition was chosen because some of the properties of this melt have already been investigated at high pressure. For this study, diamond, graphite, forsterite and ruby spheres were used. Spheres with different diameters were also used in order to optimize the run duration. For example, a long duration run could result in contamination of the melt by the crystal. The difficulty of applying this method to any silicate liquid is that the time/distance relationship needs to be bracketed. A overly long run duration will lead to the sphere completely traversing the sample chamber so that the travel distance in a given time cannot be measured. A run with too short a duration, on the other hand, will not allow sufficient movement of the sphere for an accurate distance measurement to be made. In addition, very precise density data for the marker minerals is required at high P/T, which is the case only for few minerals. Finally, the viscosity of a melt as function of pressure can be determined by using two different mineral markers of contrasting densities.