The importance of melt TiO2 in affecting major and trace element partitioning between Fe-Ti oxides and lunar picritic glass melts

Lunar mare basalts and picritic glasses have TiO2 abundances ranging from less than 1 wt.% to over 16%. Any high-Ti mare basalt or picritic glass petrogenetic model must include Fe-Ti oxides in the mantle source or invoke Fe-Ti oxide assimilation during magma ascent to the lunar surface. We conducted partitioning experiments to investigate high field strength element (HFSE), rare earth element (REE), and transition metal distribution between Fe-Ti oxides and lunar picritic glass melts over a range of melt compositions. Our results suggest that ilmenite-melt and armalcolite-melt HFSE, Cr, and V partition coefficients (DHFSE, DCr, DV) are strongly dependent on melt TiO2 content, whereas ilmenite-melt REE partition coefficients appear to be insensitive to melt composition. As TiO2 increases in picritic glass melts, HFSE, Cr, and V activities in melt also increase and Fe-Ti oxide-melt DHFSE, DCr and DV decrease. The effect of Ti on partitioning behavior can be attributed to the formation of Fe-O-Ti melt species in high-Ti melts. Ilmenite DHFSE range from compatible in oxides in equilibrium with low-Ti melts to incompatible in oxides in equilibrium with depolymerized high-Ti picritic glass melts. DHFSE are inversely correlated with TiO2 abundance in the melt and become nearly constant for melts with more than 6.8% TiO2. We present simple partitioning models that utilize the solubility of ilmenite and armalcolite in melt to effectively predict HFSE partition coefficients across a wide range of picritic glass melt compositions. The HFSE budget of ilmenite cumulates that crystallize from the lunar magma ocean strongly depends on the composition of the magma ocean. Low-Ti and high-Ti lunar basalts can be produced by an ilmenite or armalcolite bearing hybridized mantle source, or by assimilation of late-stage magma ocean cumulates. The dependence of DHFSE and DCr on melt TiO2 is consistent with the formation of lunar Type 1 armalcolite from high TiO2 picritic glass melt, however, lunar Type 2 and 3 armalcolite have a more complicated (yet undetermined) history of formation.