MULTICOMPONENT DIFFUSION IN TERNARY SILICATE MELTS IN THE SYSTEM K2O-AL2O3-SIO2 .2. MECHANISMS, SYSTEMATICS, AND GEOLOGICAL APPLICATIONS

Diffusion coefficient matrices (D) for ternary silicate melts have been analysed in terms of eigenvalues and eigenvectors of these matrices. We have used experimental data from our laboratory on melts in the system K2O-Al2O3-SiO2 (Chakraborty et al., 1995), as well as data from the literature. It is found that the eigenvectors, which determine the stoichometry of homogeneous melt reactions in the melts during diffusion, are relatively insensitive to composition in a given system and temperature. In contrast, the rates of these reactions, given by the eigenvalues of the [D] matrices are strong functions of both composition and temperature. Analysis of diffusion in terms of eigenvalues and eigenvectors helps us to explain observed diffusion paths in the ternary systems and rationalize the difference in behavior of alkalis vs. non alkalis in multicomponent systems, found earlier in the literature (Watson, 1982; Watson and Jurewicz, 1984). A method for predicting the occurrence of uphill diffusion during diffusion in multicomponent systems is discussed and tested experimentally. The concentration profiles obtained during interdiffusion experiments (Chakraborty et al., 1995) are fitted to theoretical models to retrieve tracer diffusion coefficients of Si and Al in these melts. For peraluminous melts in the silica rich part of the K2O-Al2O3-SiO2 system the activation energies for both Si and Al tracer diffusion are found to be 290-318 kJ/mol (70-76 kcal), for corresponding peralkaline compositions the energies are 125-146 kJ/mol (30-35 kcal). The retrieved tracer diffusivities are then used to constrain structural relaxation timescales and viscosities of these melts. Compositions that develop during mixing of multicomponent melts are determined by the nature of diffusion paths that develop during the mixing process. It is discussed that anomalous compositions may develop in the mixing zone due to both weak and strong coupling during diffusion which may make interpretation of chemical signatures of melt inclusions difficult. On the other hand, the peculiarities of diffusion paths in a multicomponent system may help understand complex geochemical mixing arrays and some features of anomalous alkali enrichment/depletion during magmatic evolution.