MELT GEOMETRY, MOVEMENT AND CRYSTALLIZATION, IN RELATION TO MANTLE DYKES, VEINS AND METASOMATISM

Consideration of theoretical, experimental and natural rock data show that basic-ultrabasic melt will disperse along mineral grain edges in olivine-rich mantle rock and thereby form a connected three-dimensional network throughout the rock even when present in only small (less than 1 %) volumes. The viscosity of such melts will also allow small (less than 1-5 %) volumes to move on appropriate geological timescales as a result of gravity-driven compaction. These features mean that small volume basic-ultrabasic melts are capable of infiltrating and metasomatizing mantle peridotites. Modally metasomatized mantle xenoliths are commonly closely associated with an array of dyke-like and vein injection phenomena. Textural, structural and modal characteristics of a wide array of mantle dykes, veins and metasomatic rocks suggest that such rocks have certain features in common with cumulates, and might usefully be distinguished as dyke cumulates and metasomatic infill cumulates. They represent partial crystal precipitates from melt flowing along channelways or pervasively through peridotite, and their bulk rock compositions provide poor guides to actual mantle melt compositions. The crystallization of the minerals in dykes/veins/metasomites causes differentiation of the melt by crystal fractionation processes, but at the same time the melt may maintain equilibrium with host rock phases (e.g. olivine) and chromatographic column or percolation effects will control the range of transport of different chemical components by the melt. These combined processes are referred to as percolative fractional crystallization. Data on the actual trace element compositions of melt in equilibrium with the minerals of mantle dykes/veins/metasomites are calculated from trace element analyses of the minerals by using partition coefficients. For a wide variety of metasomatic suites, the calculated melt compositions show a progression of trace element abundances from ones similar to primitive asthenospheric OIB-like compositions towards more incompatible element enriched compositions. Thus they support the hypothesis that fractional crystallization and percolative fractional crystallization processes operating upon initial primitive asthenospheric melts may yield melt compositions matching those necessary for wide varieties of mantle metasomatism. The differentiation of the melts and evolution of the metasomatic rocks proceed together. No evidence for the involvement of volatile-rich fluids distinct from melts has been found. The trace element compositions of many kimberlitic and lamproitic melts may also arise by processes of percolative fractional crystallization of initially primitive melts with OIB-like trace element compositions, as a result of flow through mantle peridotite.