^1,2Yishen Zhang, ^3,4,5Bernard Charlier, ⁴Stephanie B. Krein, ⁴Timothy L. Grove, ^1,5Olivier Namur, ⁵Francois Holtz
Earth and Planetary Science Letters 646, 118989 Link to Article [https://doi.org/10.1016/j.epsl.2024.118989]
¹Department of Earth and Environmental Sciences, KU Leuven, 3001 Leuven, Belgium
²Department of Earth, Environmental and Planetary Sciences, Rice University, 6100 Main Street, MS 126, Houston, TX 77005, USA
³Department of Geology, University of Liège, 4000 Sart Tilman, Belgium
⁴Massachusetts Institute of Technology, Department of Earth, Atmospheric, and Planetary Sciences, Cambridge, MA 02139 USA
⁵Institut für Erdsystemwissenschaften, IESW, Abteilung Mineralogie, Leibniz Universität Hannover, 30167 Hannover, Germany
Copyright Elsevier

The latest stages of the lunar magma ocean (LMO) crystallization led to the formation of ilmenite-bearing cumulates and urKREEP, residual melts enriched in K, rare earth elements (REEs), P, and other incompatible elements. Those highly evolved lithologies had major impacts on the petrogenesis of lunar volcanic rocks and the compositional diversity of post-LMO magmatism resulting from mantle remelting. Here, we present new experimental results constraining the composition of the very last liquids produced during LMO crystallization. To test the potential role of silicate liquid immiscibility in the formation of urKREEP, synthetic samples representative of residual melts of bulk Moon compositions were placed in double platinum-graphite capsules at 1020–980 °C and 0.08–0.10 GPa in an internally-heated pressure vessel. The produced silicate liquids are multiply saturated with plagioclase, augite, silica phases, and ilmenite (± fayalitic olivine ± pigeonite). Our experiments show that the liquid line of descent reaches a two-liquid field at 1000 °C and >97% crystallization for a range of whole-Moon compositions. Under these conditions, a small proportion of silica-rich melt (70.0–71.4 wt.% SiO2, 6.4–7.3 wt.% FeO, 5.4–6.1 wt.% K2O, 0.2–0.3 wt.% P2O5) coexists within an abundant Fe-rich melt (42.6–44.1 wt.% SiO2, 27.6–28.8 wt.% FeO, 0.9–1.0 wt.% K2O, 2.8–3.2 wt.% P2O5) with sharp two-liquid interfaces. Our experimental results also constrain the relative onset of ilmenite crystallization compared to the development of immiscibility and indicate that an ilmenite-bearing layer formed in the lunar interior before immiscibility was attained. Using a self-consistent physicochemical LMO model, we constrain the thickness and depth of the ilmenite-bearing layer during LMO differentiation. The immiscible K-Si-rich and P-Fe-rich melts together also produced an immiscible urKREEP layer ∼2–6 km thick and ∼30–50 km deep depending on the trapped liquid fraction in the cumulate column (≤10%) and the thickness of the buoyant anorthosite crust (30–50 km). We provide constraints on the relationship between the compositions of immiscible urKREEP melts and those of KREEPy rocks. By modeling the mixing of KREEP-poor basalt and the immiscible melt pairs, we reproduce the K and P enrichments and apparent decoupling of K from P in KREEPy rocks. Our results highlight that processes such as the assimilation of evolved heterogeneous mantle lithologies may be involved in hybridization during post-LMO magmatism. The immiscible K-Si-rich lithology may also have contributed to lunar silicic magmatism.