Understanding the variability in the titanium contents of lunar basalts using high-pressure, high-temperature layered partial melting experiments and thermodynamic modelling

New high-pressure, high-temperature layered partial melting experiments have been performed to simulate the interaction between the last stage, dense Fe-Ti-rich cumulate layer that remained after about 99.5% crystallization of the lunar magma ocean (LMO), and the underlying solidified Mg-rich mantle. For this, a synthetic assemblage of an upper Fe-Ti-rich cumulate layer (6.1 wt.% TiO2 and 39.7 wt.% FeO) and a lower forsteritic olivine layer (Mg# = 92.8) has been taken in 1:4 weight ratio, respectively, and subjected to experiments at pressures ranging between 1 to 3 GPa and temperatures varying between 1100 C and 1525 C using a piston cylinder apparatus. In the top cumulate layer, phases such as Fe-rich clinopyroxene, Fe-poor clinopyroxene, pigeonite, orthopyroxene, rutile, ilmenite, quartz and melt were formed, depending upon different P-T conditions. The Fe-Ti-rich basaltic melts (5-18.5 wt.% TiO2, 13-28.6 wt.% FeO, and 35-59 wt.% SiO2) produced in this cumulate layer at different degrees of partial melting approach the lunar mare basalts in their compositions and can be used to explain the huge variation in TiO2 enrichment that is observed in lunar basalts (between 0 to ~17 wt. %). Following LMO crystallization, the last stage dense mineral-melt cumulate layer is expected to undergo a gravitational overturn due to density instability. This work aims to simulate the partial melting of the cumulates in this layer as a result of the overturn. The consequent compositional heterogeneity of the lunar mantle is used to justify the observed variation in Ti-rich basalt compositions on the lunar surface. The basaltic melts produced in these experiments are mostly Al, Mg-poor compared to available lunar basalt samples. However, this deficiency may be addressed by assimilation of these melts into low-Ti, Mg-rich basalt magmas that would have subsequently erupted from the underlying mantle. Simulations of such assimilation using thermodynamic modelling have also been done and the results support this theory. The possible fate of the last stage melt of the LMO has thus been studied and used to understand the variable compositions of lunar basalts.