REVIEW ARTICLE PUBLISHED ONLINE: 25 MAY 2014 | DOI: 10.1038/NGEO2173 Heterogeneous distribution of water in the Moon Katharine L. Robinson* and G. Jeffrey Taylor Initial analyses of lunar samples returned by the Apollo missions indicated that the Moon was essentially devoid of water. However, improved analytical techniques have revealed that pyroclastic glass beads in Apollo samples contain measurable amounts of water. Taking into account volatile loss during eruption of the glass beads onto the surface, the pre-eruption magma could have contained water on the order of 100 ppm by weight, concentrations that are similar to the mantle sources of mid- ocean ridge basalts on Earth. Lava flows from vast basaltic plains — the lunar maria — also contain appreciable amounts of water, as shown by analyses of apatite in mare basalt samples. In contrast, apatite in most non-mare rocks contains much less water than the mare basalts and glass beads. The hydrogen isotopic composition of lunar samples is relatively similar to that of the Earth’s interior, but the deuterium to hydrogen ratios obtained from lunar samples seem to have a larger range than found in Earth’s mantle. Thus, measurements of water concentration and hydrogen isotopic composition suggest that water is heterogeneously distributed in the Moon and varies in isotopic composition. The variability in the Moon’s water may reflect heterogeneity in accretion processes, redistribution during differentiation or later additions by volatile-rich impactors. eginning with the first glimpses of lunar basalts returned by the measurement techniques honed for measuring volatile contents in Apollo 11 mission in 1969, the conventional wisdom was that volcanic glasses from the sea floor12–15. Lunar volcanic glass depos- the Moon was essentially anhydrous. Although this view was its (Fig. 1) were formed by volatile-driven fire fountains and are B 16 based on sound reasoning (Box 1), it turned out to be incorrect, as distinguishable from glass particles produced by impacts . They shown by discoveries of water in volcanic glass beads1 and in apatite contain a huge amount of information about the lunar interior, in in lunar basalts2. part because of the wide range in their chemical compositions. This These discoveries provide a new tool to unravel the processes compositional difference shows up in the colours of the glasses: red involved in lunar origin, differentiation and bombardment. The have the highest titanium content (~15 wt% TiO2) and green the present consensus is that the Moon formed as the result of a giant lowest (<1 wt% TiO2). impact of an approximately Mars-sized planetesimal with the proto- The glass beads contain up to 45 ppm by weight (ppmw) H2O Earth3–5. A chief geochemical virtue of this model is that the hot (ref. 1) with most in the range 10–30 ppmw (Fig. 2), comfortably conditions led to loss of volatile elements, explaining the strong above detection limits and the <1 ppmw expected from conven- depletion of volatile elements in the Moon compared with Earth. tional wisdom. These low water contents represent lower limits, as One might assume that all water would be lost during such an event, significant amounts of 2H O would have been lost by degassing dur- 6,7 1 but this is not correct . The water in the Moon is a tracer of the ing eruption . Consistent with loss, the concentrations of H2O, Cl, processes that operated in the hot, partly silicate gas, partly magma F and S decrease from the interiors to the surfaces of glass beads. 1 disk surrounding Earth after the impact (see Box 2 for a discussion Diffusion calculations show that the initial magma had a H2O con- of what ‘water’ means). centration of 260­–745 ppmw, not significantly different from those Water could have been redistributed during lunar differentiation. measured in nondegassed mid-ocean ridge basalt (MORB) glasses. The Moon’s differentiation began with a global magma ocean at least In other words, the water contents were Earth-like. This does not hundreds of kilometres deep. Crystallization produced a cumulate mean that the Moon contains as much water as Earth, because the crust a few tens of kilometres deep, dominated by anorthosite. MORBs originate in the driest part of the terrestrial mantle. For Overturn of the magnesian, low-density early cumulates (domi- example, pre-eruptive magmas produced in terrestrial subduction nated by olivine and orthopyroxene) produced hybrid sources that zones have percent levels of H2O (ref. 17). Nevertheless, water con- when heated by long-lived radioisotopes produced other igne- tents similar to MORBs are highly significant. ous lithologies, including the mare basalts that make up the dark The water contents of melt inclusions trapped inside olivine regions of the lunar nearside. Any of these processes could alter the microphenocrysts in orange glass beads (Fig. 1) confirm the dif- distribution of water in the Moon8,9. fusion calculations. The melt inclusions contain 615–1,200 ppmw The Moon was also bombarded by numerous large planetesi- H2O (ref. 18; Fig. 2). These data clearly show that the mantle source mals, making the large impact basins and craters that decorate its regions for the volcanic glasses contained on the order of 100 ppmw 10 surface. It is conceivable that water was added by such impactors . H2O (assuming 10% partial melting and that H2O strongly con- If so, then at least some of Earth’s water might also have been added centrates in the melt), which is similar to the mantle sources after it essentially finished accreting. Water in the lunar interior for MORB14. might thus be a tracer for addition of water to Earth. The isotopic composition of hydrogen is as important as its total abundance. The deuterium/hydrogen ratio19, δD (Box 2), Water in glassy volcanic deposits is higher in three types of lunar volcanic glass (green, yellow and The first report of water inside lunar materials came at the Lunar orange) than in the trapped melt inside olivine crystals in orange and Planetary Science Conference in 2007 (ref. 11), soon fol- glass (Fig. 3a). The high δD in the glasses reflects H–D fractiona- lowed by detailed analyses of H, C, F, S and Cl in lunar volcanic tion from the droplets of lava when erupted. The measured δD of glasses1. The abundances were small, but detectable, in sensitive the melt inclusions ranges from +187 to +327‰, higher than the Hawaii Institute of Geophysics and Planetology, University of Hawaii, Honolulu, Hawaii 96822, USA. *e-mail: [email protected] NATURE GEOSCIENCE | VOL 7 | JUNE 2014 | www.nature.com/naturegeoscience 401 © 2014 Macmillan Publishers Limited. All rights reserved REVIEW ARTICLE NATURE GEOSCIENCE DOI: 10.1038/NGEO2173 Box 1 | Why lunar scientists thought the Moon was dry The long-held belief of a nearly anhydrous Moon was based on and after repeated heating to outgas adsorbed water, Q rose to sound reasoning. Weathering products were conspicuously absent 4,800 (ref. 78). Thus, the seismic properties of the lunar crust in lunar basalts, suggesting that water did not flow across the sur- fortified the idea that the Moon has a much lower water content face or leak out to alter the original igneous rocks. The basalts did than Earth. not contain any water-bearing minerals, but more importantly, nei- A few measurements did suggest that water was present in ther did intrusive rocks. Intrusive rocks crystallize at depth, where lunar samples, but all were rejected as terrestrial contamination. higher pressure increases the solubility of water in magmas71. If One study79 crushed samples of lunar basalt from both exterior water is present it would be expected to lead to the formation of surfaces and the rock’s interior to <250 μm, and measured the hydrous minerals such as amphibole at least some of the time, yet gases released from each sample while they were heated from 25 to no such hydrous minerals were found. Bulk chemical analyses 1,400 °C. A sharp peak in H2O for the surface sample and release 72 of lunar basalts indicated H2O concentrations of <100 ppmw , over a broad range of temperatures for H2O for the interior sample 73 compared with 1,000–3,000 ppmw H2O in Hawaiian basalts . was observed, similar to the release from measurements on lunar (Analytical techniques with low detection limits to measure soils. The investigators concluded that that the water in the surface water contents in individual mineral grains were not available.) sample was probably adsorbed from the terrestrial atmosphere79. Furthermore, the dry Moon hypothesis was consistent with low D/H and oxygen isotopes were also measured in regolith sam- contents of elements such as Cd and Bi74, which are volatile, but ples80. The solar wind, like the Sun, has low δD (approx. −1,000‰), 81 less so than H2O and H2. compared with the bulk Earth (−62.5‰). The water released by Another driver of the dry-Moon hypothesis was the surpris- heating the regolith has δD of −100‰, in the range of air sam- ingly weak attenuation of seismic waves as measured by seis- ples in Pasadena, California, where the samples were measured80, mometers left by the Apollo missions 75. Seismic attenuation is leading lunar scientists to conclude that the water was terrestrial expressed by the inverse of the ‘quality factor’, Q; the higher the contamination. Of course, at the time we did not know the δD of value of Q, the less attenuation. The upper crust of the Moon lunar interior water. (The δD nomenclature is described in Box 2.) has a Q of 3,000–5,000, compared with terrestrial values of at Evidence for lunar water was also reported from ‘rusty rock’, an least ten times smaller75,76.