Silicate liquid immiscibility in impact melts

We have investigated silicate emulsions in impact glasses and impact melt rocks from the Wabar (Saudi Arabia), Kamil (Egypt), Barringer (USA), and Tenoumer (Mauritania) impact structures, and in experimentally generated impact glasses and laser-generated glasses (MEMIN research unit) by scanning electron microscopy, electron microprobe analysis, and transmission electron microscopy. Textural evidence of silicate liquid immiscibility includes droplets of one glass disseminated in a chemically distinct glassy matrix; sharp phase boundaries (menisci) between the two glasses; deformation and coalescence of droplets; and occurrence of secondary, nanometer-sized quench droplets in Si-rich glasses. The compositions of the conjugate immiscible liquids (Si-rich and Fe-rich) are consistent with phase separation in two-liquid fields in the general system Fe2SiO4- KAlSi3O(8)-SiO2-CaO-MgO-TiO2-P2O5. Major-element partition coefficients are well correlated with the degree of polymerization (NBO/T) of the Si-rich melt: Fe, Ca, Mg, and Ti are concentrated in the poorly polymerized, Fe-rich melt, whereas K, Na, and Si prefer the highly polymerized, Si-rich melt. Partitioning of Al is less pronounced and depends on bulk melt composition. Thus, major element partitioning between the conjugate liquids closely follows trends known from tholeiitic basalts, lunar basalts, and experimental analogs. The characteristics of impact melt inhomogeneity produced by melt unmixing in a miscibility gap are then compared to impact melt inhomogeneity caused by incomplete homogenization of different (miscible or immiscible) impact melts that result from shock melting of different target lithologies from the crater's melt zone, which do not fully homogenize and equilibrate due to rapid quenching. By taking previous reports on silicate emulsions in impact glasses into account, it follows that silicate impact melts of variable composition, cooling rate, and crystallization history might readily unmix during cooling, thereby rendering silicate liquid immiscibility a much more common process in the evolution of impact melts than previously recognized.