DIFFUSION-REACTION AS A MEANS OF LARGE-SCALE MAGMA DIFFERENTIATION

Can chemical transport by diffusion through a fluid/melt differentiate magmas over the km scales needed to explain compositional variations in igneous plutonic suites? The answer to this question depends on constraining several variables at the relevant pressure-temperature conditions, namely: the rates of diffusion and mineral reaction, the length scales for diffusive transport, and the duration of the process. I will present piston cylinder experiments juxtaposing two partially molten materials (gabbro-basaltic andesite or peridotite-basanite couples) illustrating the speed over which diffusion-reaction (DR) takes place. Effective binary diffusion coefficients for elements in the bulk material are ~10^-6 to 10^-7cm2/s, indicating that solid-state diffusion through minerals does not limit the rate of chemical exchange. Plagioclase and olivine, two common solid-solution minerals, both react with a melt that is changing its composition by DR through establishment of rim-melt equilibrium while undergoing internal component exchange. This process serves to buffer changes in the melt composition and allows differentiation without mechanical separation of melt and crystals. Plagioclase at the interface between partially molten gabbro and basaltic andesite develops highly anorthitic cores, explaining a petrologic anomaly found in nature. In all of these experiments, melt-solid equilibrium is rapidly established with transport between materials continuously differentiating the two materials. The length scale over which the DR process occurs can be approximated by (4Dtφ)^1/2 where D is the diffusion coefficient, t is time, and φ is the melt porosity. Diffusion coefficients in fluids/melts could range from 10^-7cm2/s up to the diffusivity of silica in supercritical H₂O at high T, ~10^-3cm2/s (Watson and Wark, 1997). If the duration of the DR process were Myrs, km diffusion scales may be possible. A two-porosity regime of melt conduits spaced at 0.5 km connected to intergranular melt would shorten the needed time scale further. Given current constraints, differentiation by DR at <5% melt porosity could produce observed km-scale rock suites without requiring large, high porosity magma chambers or multiple sheet intrusions.