Magmatic evolution of Li–F, rare-metal granites: a case study of melt inclusions in the Khangilay complex, Eastern Transbaikalia (Russia)

We report compositions of homogenized quartz-hosted melt inclusions from a layered sequence of Li-, F-rich granites in the Khangilay complex that document the range of melt evolution from barren biotite granites to Ta-rich, lepidolite–amazonite–albite granites. The melt inclusions are crystalline at room temperature and were homogenized in a rapid-quench hydrothermal apparatus at 200 MPa before analysis. Homogenization runs determined solidus temperatures near 550 °C and full homogenization between 650 and 750 °C. The compositions of inclusions, determined by electron microprobe and Raman spectroscopy (for H₂O), show regular overall trends of increasing differentiation from the least-evolved Khangilay units to apical units in the Orlovka intrusion. Total volatile contents in the most-evolved melts reach over 11 wt.% (H₂O: 8.6 wt.%, F: 1.6 wt.%, B₂O₃: 1.5 wt.%). Concentrations of Rb range from about 1000 to 3600 ppm but other trace elements could not be measured reliably by electron microprobe. The resulting trends of melt evolution are similar to those described by the whole-rock samples, despite petrographic evidence for albite- and mica-rich segregations previously taken as evidence for post-magmatic metasomatism.

Melt variation trends in most samples are consistent with fractional crystallization as the main process of magma evolution and residual melt compositions plot at the granite minimum in the normative Qz–Ab–Or system. However, melts trapped in the highly evolved pegmatitic samples from Orlovka deviate from the minimum melt composition and show compositional variations in Al, Na and K that requires a different explanation. We suggest that unmixing of the late-stage residual melt into an aluminosilicate melt and a salt-rich dense aqueous fluid (hydrosaline melt) occurred. Experimental data show the effectiveness of this process to separate K (aluminosilicate melt) from Na (hydrosaline melt) and high mobility of the latter due to its low viscosity and relatively low density may explain local zones of albitization in the upper parts of the granite.