Inhomogeneous Distribution of Magmatic Volatiles in the Lunar Interior: Clues from the Mineral Apatite

Introduction: The magmatic volatiles water, fluorine, and chlorine play critical roles in the physicochemical processes that control thermal stabilities of minerals and melts, magma eruptive processes, and transport of economically important metals. While these volatiles are significant constituents in many terrestrial and martian magmas, lunar magmas are generally depleted in these three volatiles (i.e., 1, 2-5). Nevertheless, Cl, F and H2O likely became concentrated in late stage lunar melts because of their incompatible behavior with respect to high-temperature mineral phases. Importantly, the concentration of these volatiles in the hypothesized urKREEP residual melt could have played a significant role in all post-differentiation magmatism (e.g., mare basalts and magnesian suite rocks) that was generated by “urKREEP-heating” (due to the solidus-depressing abilities of the volatiles); however, little is currently known about the relative abundances of volatiles in late-stage lunar melts. Although these late-stage melts have not yet been found as quenched glasses and therefore, their volatile content cannot be studied directly, the volatiles in these liquids ought to have been incorporated into F-, Cl-, and OH-bearing mineral phases. As such, these minerals can be used to assess the relative amounts of dissolved volatiles in their parent magmas. Because lunar phosphates contain elevated amounts of both phosphorus and rare earth elements (REE), components that are generally enriched in latestage liquids due to their incompatible nature (they are also two of the main constituents in the hypothesized ur-KREEP liquid), they would be a potentially useful mineral group for understanding magmatic volatiles, which also typically behave incompatibly. Apatite [Ca5(PO4)3(F,Cl,OH)] and RE-merrillite [(Mg,Fe)2REE2Ca16P14O56] are the two main P-bearing minerals on the Moon (6) and they are found in a wide range of lunar rock types. Although merrillite is a volatile-free phosphate, fluorine, chlorine, and hydroxyl are essential structural constituents of apatite, making it the ideal phase for our study. Electron probe microanalysis (EPMA) of lunar apatite can be used to estimate the relative amounts of fluorine, chlorine, and potentially water in coexisting melt using known partitioning data between the apatite volatile site (X-site) and silicate melt and/or fluid (i.e., 7, 8-10). Importantly, partitioning studies show that Cl, F, and OH have melt/apatite partition coefficients that not only deviate from unity but also vary with anion. Relative to melt, F partitions strongly into apatite whereas Cl and OH are less compatible in apatite under a range of P-T conditions (i.e., ≤ 10 kbar & ≤ 1150 °C) . Complicating the issue further, apatite/melt partition coefficients are highly dependant upon melt composition (i.e., 8, 10); therefore, understanding the composition of the liquids from which apatite has crystallized is a crucial step in understanding the magmatic volatile content of that liquid. Hydroxyl cannot be measured directly by EPMA; however, it can be calculated from electron microprobe measurements of fluorine and chlorine if direct hydrogen analysis is not available. This requires the following assumptions: (i) only F, Cl, and OH populate the X-site, (ii) the X-site has minimal vacancies, and (iii) reliable analyses of F and Cl are available. Unfortunately, electron-microprobe analysis of fluorine in apatite is not straightforward because of apparent anisotropic diffusion of this anion during excitation by an electron beam when analyses are made parallel to the c-axis (11). This anisotropic diffusion can cause overcounting of F X-ray intensities, thereby resulting in spuriously high concentrations. If any OH were present, it would be masked in such analyses. Moreover, because this analytical problem directly affects the fluorine concentrations, use of partition coefficients is rendered useless if analyses are not monitored for such over-counting. Previous studies of this type did not take into account the potential analytical problems discussed above; however, we have devised a reliable method for testing the quality of apatite data collected by EPMA (Our analytical methods are identical to those described in the following abstracts: 12, 13; A plot illustrating count acceleration is presented in Figure 1.) and have employed this method to investigate the fluorine and chlorine contents of apatite in a variety of Apollo samples and lunar meteorites: 14161,7111, 14161,7233 14161,7269 14161,7264 14161,711