Controls of Hydrogen Partitioning on the Formation of Wet Reservoirs During Lunar Magma Ocean Crystallization

Kurzfassung

Recent studies indicate the presence of hydrogen (H) in lunar samples. H inherited from the proto-Earth and impactors, as well as added during the phase of accretion (contemporaneous to lunar magma ocean (LMO) crystallization) have been retained in spite of losses during the Moon-forming impact and magma ocean degassing [1]. The bulk H in the LMO or the Bulk Silicate Moon (BSM) is an important constraint to understand the dynamics of the Moon-forming impact as well as determine the origin of volatiles in the Earth-Moon system [2]. Recent analysis of Apollo samples indicates that the Moon has heterogeneous H reservoirs [3], which may be explained by the partitioning of H between nominally anhydrous cumulates and liquid as well as the entrapment of residual liquid during progressive crystallization of the LMO. The recent estimates of H2O in the BSM (5 to 1650 μg/g; [4]) rely heavily on the partition coefficients of H (DH) between minerals and melt used in the models (where DH = H concentration in mineral/ H concentration in the melt). Here we demonstrate the effect of DH between nominally anhydrous minerals (NAMs) and melt on mantle and crustal H contents by modeling the fractional crystallization of a 600 km deep LMO. We follow published crystallization sequences as well as use a combination of the codes SPICES and alphaMELTS. We use lower and upper limits of DH for each mineral-melt pair as published in the literature and vary the initial bulk H assuming 1% residual melt in the crystal mush after compaction. Using joint H2-H2O solubility, we further evaluate the extent of degassing in the LMO. We demonstrate that H in plagioclase may be explained by either DHmin or DHmax, if the initial H content of the LMO was 100 μg/g. However, with higher initial H content, i.e. 1000 μg/g, only DHmin would explain plagioclase chemistry. This demonstrates that the current range of published values of DH are not sufficient to fully capture the dynamics of LMO crystallization, and highlight the necessity in future studies to experimentally constrain the DH values between the NAMs and melt compositions, specific to LMO crystallization conditions. [1] Barnes et al., 2016. Nat. Comm. [2] Desch and Robinson, 2019. Geochem [3] Robinson et al., 2016. GCA [4] McCubbin et al., 2015. Am Min.