Viscosity of Carbonate-Silicate Melts Using Ultra-High Resolution Falling Sphere Viscometry

Carbonatites are carbon-rich magmas that participate in the deep-Earth carbon cycle, and may be precursors to many types of terrestrial magmatism. [1] Viscosity is a crucial parameter in determining migration rates and behavior of melts from the upper mantle to Earth's surface, but very little is known about the viscosity of carbonate-silicate liquids at conditions relevant to Earth's interior. To examine the viscosity of carbonate-silicate liquids as a function of composition, we performed in situ falling sphere viscosity experiments using a high speed X-ray camera and Paris-Edinburgh press at the HPCAT beamline (Advanced Photon Source, Argonne National Laboratory). Mixtures from the CaCO3-CaSiO3 (calcite-wollastonite) binary system were used to simulate mantle silicate carbonatites. Samples were loaded using the experimental setup of Yamada et al [2], held at 3 GPa, and heated until the sample was fully molten (between 1350-1650 oC). The high speed camera recorded the falling rate of a platinum sphere placed near the top of the sample chamber, enabling the calculation of terminal velocity and hence viscosity. Results indicate that pure CaCO3 at upper mantle conditions has a very low viscosity of ~0.006 Pa-s, only a little higher than that of water. This viscosity is 2-6 times lower than that of potassium carbonates at similar pressures (2.5-4.0 GPa) but at somewhat lower temperatures (800-1200 oC). [3] Our measured viscosity as a function of increasing silicate content increases along a log-linear trend, reaching 0.256 Pa-s for CaSiO3 liquid. This heavy dependence of viscosity on composition has implications for melt migration processes at different depths, suggesting either viscosity-driven or porosity-driven migration depending on both extent of melting and carbonate content. [1] Dasgupta, R. et al. (2013) Nature 493, 211-215. [2] Yamada, A. et al. (2011) Rev. Sci. Instr. 82, 015103. [3] Dobson, D. et al. (1996) Earth Plan. Sci. Lett. 143, 207-215.