An experimental study of the microstructural evolution of an immiscible basaltic emulsion shows that the Fe-rich liquid forms homogeneously nucleated droplets dispersed in a continuous Si-rich liquid, together with droplets heterogeneously nucleated on plagioclase, magnetite, and pyroxene. Heterogeneous nucleation is likely promoted by localised compositional heterogeneities around growing crystals. The wetting angle of Fe-rich droplets on both plagioclase and magnetite increases with decreasing temperature. Droplet coarsening occurs by a combination of diffusion-controlled growth and Ostwald ripening, with an insignificant contribution from coalescence. Characteristic microstructures resulting from the interaction of immiscible Fe-rich liquid with crystal grains during crystal growth can potentially be used as an indicator of liquid unmixing in fully crystallised natural samples. In magma bodies < ~ 10 m in size, gravitationally driven segregation of immiscible Fe-rich droplets is unlikely to be significant.

It is common practice to estimate a mean diameter for spherical or sub-spherical particles or vesicles in a rock by multiplying the average diameter of the approximately circular cross-sections visible in thin section by a factor of 1.273. This number-weighted average may be dominatedby the hard-to-measure fine tail of the size distribution, and is unlikely to be representative of the average particle diameter of greatest interest for a wide range of geological problems or processes. Average particle size can be quantified in a variety of ways, based on the mass or surfacearea of the particles, and here we provide exact relations of these different average measures to straightforward measurements possible in thin section, including an analysis of how many particles to measure to achieve a desired level of uncertainty. The use of average particle diameter isillustrated firstly with a consideration of the accumulation of olivine phenocrysts on the floor of the 135 m thick picrodolerite/crinanite unit of the Shiant Isles Main Sill. We show that the 45 m thick crystal pile on the sill floor could have formed by crystal settling within about a year.The second geological example is provided by an analysis of the sizes of exsolved Fe-rich droplets during unmixing of a basaltic melt in a suite of experimental charges. We show that the size distribution cannot be explained by sudden nucleation, followed by either Ostwald ripening or Browniancoalescence. We deduce that a continuous process of droplet nucleation during cooling is likely to have occurred.

The separation of immiscible liquids has significant implications for magma evolution and the formation of magmatic ore deposits. We combine high-resolution imaging and electron probe microanalysis with the first use of atom probe tomography on tholeiitic basaltic glass from Hawaii, the Snake River Plain, and Iceland, to investigate the onset of unmixing of basaltic liquids into Fe-rich and Si-rich conjugates. We examine the relationships between unmixing and crystal growth, and the evolution of a nanoemulsion in a crystal mush. We identify the previously unrecognised role played by compositional boundary layers in promoting unmixing around growing crystals at melt-crystal interfaces. Our findings have important implications for the formation of immiscible liquid in a crystal mush, the interpretations of compositional zoning in crystals, and the role of liquid immiscibility in controlling magma physical properties.

No abstract