DIFFUSIONAL GRADIENTS AT THE CRYSTAL/MELT INTERFACE AND THEIR EFFECT ON THE COMPOSITION OF MELT INCLUSIONS

Melt inclusions are widely believed to represent melts from which host crystals grew. Melt inclusions represent melt adjacent to growing crystals, where compositional gradients exist due to preferential incorporation or exclusion of components by the crystallizing mineral. The possibility arises, therefore, that melt in inclusions may differ significantly from melt which was more remote from growing crystals at the time the crystals grew. We have tested this possibility by analyzing 45 major, minor, and trace components in 50 to 400 mu m diameter melt inclusions in phenocrysts from the rhyolitic Bishop Tuff, California. The following observations indicate that the effects of compositional gradients on chemical compositions of melt inclusions are negligible: (1) melt inclusions in quartz and sanidine phenocrysts have indistinguishable compositions; (2) no correlation is observed between sizes of melt inclusions and their chemical compositions; (3) ten melt inclusions in four quartz phenocrysts from a single pumice clast display well-defined negative correlations between the concentrations of U and La, Ce, Ca, Mg; and (4) melt inclusions have the same major element compositions as the matrix glass and whole rocks. Melt inclusions >50 mu m can, therefore, be used to represent the melt from which their host crystals grew. The most likely explanation for the negligible compositional gradients is that a modest gradient of a major component would suppress local supercooling and constrain the crystal growth rate. For components that have diffusivities greater than dr similar to the major component which controls the rate of crystal growth, the effect of chemical gradients on compositions of trapped melt inclusions is likely to be negligible.