DISSOLUTION ORIGIN AND SYN-ENTRAPMENT COMPOSITIONAL CHANGE OF MELT INCLUSION IN PLAGIOCLASE

Abstract

In the system Di–Ab–An–H2O, melt inclusions in plagioclase were reproducibly formed in a series of partial dissolution experiments, and the compositional differences from the surrounding matrix melt were evaluated. At 200 MPa and 1050−−1150°C, a seed crystal of An59 was partially dissolved into the matrix melt, which was in equilibrium with more calcic plagioclase than the seed. The reaction proceeded inward from the crystal surface by a dissolution and recrystallization process to form a reaction zone composed of melt channels and recrystallized calcic plagioclase. The melt channels were choked (necked) off by the recrystallized plagioclase at the crystal surface to create melt inclusions. The experimentally formed inclusions, having square and zigzag shapes up to a few tens of micrometers in size, are similar in texture to some common types of natural melt inclusion. Since the partial dissolution of phenocrysts can occur during various magma mixing processes, it should be taken into consideration as one of the inclusion formation mechanisms when trying to interpret information from natural melt inclusions, especially for those in hydrous arc magmas. The chemical composition of the reproduced melt inclusions deviates from the matrix melt basically along the plagioclase liquidus isotherm. The molar ratio of Ca/(Ca + Na) and the MgO concentration in the synthesized melt inclusion are up to 17 and 61% (relative) lower than those in the matrix melt, respectively. This compositional difference is caused by a combination of factors: (1) supersaturation with dissolved plagioclase component; (2) insufficient diffusive homogenization in the melt channels due to rapid dissolution of plagioclase; and (3) the choking-off effect. On the basis of the experimental results, a model for diffusion in a growing melt channel in dissolving plagioclase was constructed which revealed that the concentration of incompatible elements with low diffusivities can be significantly lower in the melt inclusions than that in the matrix melt. These results demonstrate the importance of petrography for interpretation of inclusion data and of more experimental studies on melt inclusion origin.