TRACER DIFFUSION OF NETWORK FORMERS AND MULTICOMPONENT DIFFUSION IN DACITIC AND RHYOLITIC MELTS

Tracer diffusivities of Si, B, and Ga were measured in anhydrous dacitic and rhyolitic melts at 1.0 GPa and 1300 to 1600-degrees-C. Diffusivities were fit to the following Arrhenius equations: [GRAPHICS] where diffusivities and pre-exponential factors are in m2/s, and activation energies are in kJ/mol. In each melt silicon is the slowest diffusing cation followed by B and then Ga in the rhyolitic melt; in the dacitic melt B and Ga diffusivities are virtually identical. The low activation energy for Si in the rhyolitic melt (139 kJ/mol) is proposed to be due to reduction of the Si-O bond strength by clustering of alkalies. Clustering of network modifiers near Si cations is not as great in the dacitic melt because fewer of the network modifiers are alkalies, but instead are alkaline earths and transition metals. Measured diffusivities of B and Ga are lower than diffusivities previously measured for nonnetwork modifiers, even the modifiers that have a similar charge and a larger ionic radius, i.e., Eu3+. Differences between B and Ga diffusion in rhyolite are related to the large proportion of B in trigonal coordination in these melts whereas Ga is only in tetrahedral coordination. Activation energies for Ga diffusion are presumed similar to those necessary for aluminum tracer diffusion. The gallium activation energies are also similar to the activation energies for viscous transport in these two melts and the activation energy for Si-Al interdiffusion between these two melts. This is suggestive evidence that aluminum transport is the rate controlling step in both viscous transport and interdiffusion in these melts. Tracer diffusivities of these network formers were used to model interdiffusion between dacitic and rhyolitic melts previously measured at similar conditions of pressure and temperature (BAKER, 1990). Using the model of LASAGA (1979) with diffusivities measured in this study for network formers and diffusivities of network modifiers from other studies, diffusion profiles were calculated at 1500, 1400, and 1300-degrees-C. Additionally, the "Oxide-Molecule" model of OISHI et al. (1982) was also used at 1300-degrees-C. Calculated and measured profiles agree well at 1500-degrees-C, but as temperature decreases the mismatch between experiment and model increases. Even the inclusion of nonideal activity terms (GHIORSO et al., 1983) into the 1300-degrees-C model does not resolve discrepancies between calculations and experiments. The LASAGA (1979) model provides a better fit to experiments than the OISHI et al. (1982) model at the conditions studied. Despite discrepancies between modelled and measured diffusion profiles at 1300-degrees-C, the differences between experiment and the LASAGA (1979) model are small enough that the model may be used to place constraints on the possible effects of diffusion in dacitic to rhyolitic melts.