A NUMERICAL APPROACH TO BOUNDARY-LAYER FRACTIONATION - APPLICATION TO DIFFERENTIATION IN NATURAL MAGMA SYSTEMS

With few exceptions, geochemical models of igneous differentiation have assumed that crystals form homogeneously throughout the cooling magma chamber. This is in spite of field evidence and physical models suggesting for natural systems, that heat loss, and thus most crystallization, occurs primarily along the walls and roof of magma chambers. Such systems would produce aphyric, evolved composition lavas. Existing models of boundary layer crystal fractionation suggest that the products of such a process are significantly different from the products of homogeneous crystallization. To investigate this process further, we have developed a model using a numerical approach to boundary layer crystal fractionation based on new phase equilibria and trace element constraints. In addition to olivine, augite and plagioclase, this new model includes the ability to calculate the effects of oxide (magnetite and ilmenite), apatite and low-Ca pyroxene fractionation. This is critical if we are to effectively evaluate the geochemical signature of boundary layer fractionation. This is because the products of boundary layer fractionation are most different from homogeneous fractionation for high degrees of crystallization in the solidification zone. These are the same conditions where oxides, low-Ca pyroxene and apatite are saturated in most natural, differentiated mafic systems. The results of this model suggest that boundary layer fractionation can help to explain such phenomena as "phantom crystallization", variation in incompatible element ratios, and the decoupling of major and trace element systematics. In addition, many of the geochemical patterns that we use to distinguish between the products of low pressure and high pressure differentiation are obscured by the fact that the liquid line of descent can be controlled by the removal of phases not in equilibrium with the entire magma chamber.