Water in a young Moon constrained from ferroan anorthosites

One of the most important conclusions resulting from the Apollo and Luna missions was the lack of water detected in the returned samples or at the surface of the Moon. This notion has been included in most geophysical and geochemical models of lunar formation and evolution. The view of an anhydrous lunar interior, however, has been challenged by the recent discoveries of hydrogen in picritic glass beads, apatites and olivine melt inclusions that were facilitated by the improvements of analytical detection limit of hydrogen. Indigenous water is now considered to be heterogeneously distributed in the lunar interior, some parts containing as much water as Earth's upper mantle. Hydrogen isotopic compositions of apatites in mare basalts have been used to identify the source of this water, with results indicating a combination of indigenous lunar water, water from comets, and solar wind protons. However, magma ocean crystallization modeling suggests that the water contents estimated for the lunar mantle from mare basalts may be overestimated. We present here measurements of water in the primary products of the lunar magma ocean (LMO) in order to potentially bypass the processes of later addition of water through impact event and/or solar wind implantation, and to estimate the water content of the Moon's interior at the time of the magma ocean crystallization, and that of the mare basalt source regions. Ferroan anorthosite is the only available lithology that is believed to be a primary product of the LMO. It is generally understood that plagioclase, after crystallization, floated in the LMO and formed ferroan anorthosite as the original permanent crust of the Moon. Therefore, any indigenous water preserved in pristine ferroan anorthosite was partitioned from the LMO. Using Fourier transform infrared spectroscopy, we have measured ~6 ppm of indigenous water as hydroxyl species in plagioclases from ferroan anorthosite 15415 and 60015. As plagioclase becomes a liquidus phase after 80% LMO crystallization, we calculated the initial water content of the latter to have been ~320 ppm H2O. Water accumulated in the final 2 vol.% LMO residuum as crystallization proceeded could reach amounts of ~1.4 wt.%. We therefore show that water has been present in the lunar interior since its earliest history, during LMO formation and crystallization. This water can affect magma ocean crystallization dynamics and the following mantle cumulate overturn process, and hence the genesis of the mare basalts. Furthermore, the amount of water calculated for the LMO is sufficient to explain those detected in lunar mare basalts, which are thought to be derived from more recently remelted magma ocean cumulates. Finally, ~6 ppm of indigenous water measured in ferroan anorthosite together with up to ~2.7 ppm that we have detected in plagioclase of troctolite 76535 may contribute a significant portion to the trace amount of water/hydroxyl recently detected on the surface of the lunar highland crust by spacecraft.