This article was originally published in Treatise on Geochemistry, Second Edition published by Elsevier, and the attached copy is provided by Elsevier for the author's benefit and for the benefit of the author's institution, for non- commercial research and educational use including without limitation use in instruction at your institution, sending it to specific colleagues who you know, and providing a copy to your institution’s administrator.
All other uses, reproduction and distribution, including without limitation commercial reprints, selling or licensing copies or access, or posting on open internet sites, your personal or institution’s website or repository, are prohibited. For exceptions, permission may be sought for such use through Elsevier's permissions site at:
http://www.elsevier.com/locate/permissionusematerial
Warren P.H., and Taylor G.J. (2014) The Moon. In: Holland H.D. and Turekian K.K. (eds.) Treatise on Geochemistry, Second Edition, vol. 2, pp. 213-250. Oxford: Elsevier.
© 2014 Elsevier Ltd. All rights reserved. Author's personal copy
2.9 The Moon
PH Warren, University of California, Los Angeles, CA, USA GJ Taylor, University of Hawai‘i, Honolulu, HI, USA
ã 2014 Elsevier Ltd. All rights reserved.
This article is a revision of the previous edition article by P. H. Warren, volume 1, pp. 559–599, © 2003, Elsevier Ltd.
2.9.1 Introduction: The Lunar Context 213 2.9.2 The Lunar Geochemical Database 214
2.9.2.1 Artificially Acquired Samples 214 2.9.2.2 Lunar Meteorites 214 2.9.2.3 Remote-Sensing Data 215 2.9.3 Mare Volcanism 216 2.9.3.1 Classification of Mare Rocks 216
2.9.3.2 Chronology and Styles of Mare Volcanism 218 2.9.3.3 Mare Basalt Trace Element and Isotopic Trends 224 2.9.4 The Highland Crust: Impact Bombardment and Early Differentiation 227 2.9.4.1 Polymict Breccias and the KREEP Component 227 2.9.4.2 Bombardment History of the Moon 229
2.9.4.3 Impactor Residues: Siderophile and Fragmental 230 2.9.4.4 Pristine Highland Rocks: Distinctiveness of the Ferroan Anorthositic Suite 231 2.9.4.5 The Magma Ocean Hypothesis 235 2.9.4.6 Alternative Models 237 2.9.5 Water in the Moon 238
2.9.5.1 Traditional View of a Dry Moon 238
2.9.5.2 Water in Pyroclastic Glasses 239 2.9.5.3 Water in Apatite in Mare Basalts and KREEP-Related Samples 239 2.9.5.4 Water in the Lunar Mantle 240 2.9.5.5 Implications of Water in the Lunar Interior 240 2.9.6 The Bulk Composition and Origin of the Moon 241
Acknowledgments 242 References 242
2.9.1 Introduction: The Lunar Context Longhi, 1992, 2003; Warren and Wasson, 1979b), and in this sense, the Moon more resembles a planet than an asteroid. Stable isotopic data suggest a remarkably similar pedigree for Another direct consequence of the Moon’s comparatively the constituent matter of the Moon and Earth, given the great small size was early, rapid decay of its internal heat engine. But isotopic diversity among sampled components of the solar the Moon’s thermal disadvantage has resulted in one great system (Warren, 2011; Zhang et al, 2012). Yet, lunar materials advantage for planetology. Lunar surface terrains, and many are obviously different. The Moon has no hydrosphere and of the rock samples acquired from them, retain for the most virtually no atmosphere, and lunar materials show strong part characteristics acquired during the first few hundred Ma of depletions of even mildly volatile constituents, not just N2, solar system existence. The Moon can thus provide crucial O , and H O (e.g., Wolf and Anders, 1980). Oxygen fugacity insight into the early development of the Earth, whereas the 2 2 is uniformly very low (BVSP, 1981; Nicholis and Rutherford, terrestrial record of early evolution was largely destroyed by 2009). These idiosyncrasies have direct and far-reaching impli- billions of years of geological activity and the first 500 Ma of cations for mineralogy and geochemical processes. Basically, Earth history are missing altogether. Lunar samples show that they imply that mineralogical (and thus process) diversity is the vast majority of the craters that pervade the Moon’s surface subdued, a factor that to some extent offsets the comparative are at least 3.9 Ga old (Norman et al., 2006). Impact cratering dearth of available data for lunar geochemistry. has been a key influence on the geochemical evolution of the
The Moon’s gross physical characteristics play an important Moon, and especially the shallow Moon. The popular giant role. Although exceptionally large (radius¼1738 km) in rela- impact model holds that the Moon originated as a form of tion to its parent planet, the Moon is only 0.012 times as impact spall after a collision between the proto-Earth and a massive as the Earth. By terrestrial standards, pressures inside the doomed Mars-sized (or larger) body (e.g., Cameron and Ward, Moon are feeble: The upper mantle gradient is 0.005 GPa km 1 1976; Canup, 2004). (vs. 0.033 GPa km 1 in Earth), and the central pressure is The uppermost few meters of the lunar crust, from which slightly less than 5 GPa. However, lunar interior pressures are all lunar samples derive, are a layer of loose, highly porous, sufficient to significantly influence igneous processes (e.g., fine impact-generated debris – regolith or lunar ‘soil.’ Processes
Treatise on Geochemistry 2nd Edition http://dx.doi.org/10.1016/B978-0-08-095975-7.00124-8 213 Treatise on Geochemistry, Second Edition, (2014), vol. 2, pp. 213-250
Author's personal copy
214 The Moon
peculiar to the surface of an atmosphereless body, that is, effects such as a nearby impact melt (Cushing et al., 1999). But lunar of exposure to solar wind, cosmic rays, and micrometeorite granulitic breccias are almost invariably fine grained, and they bombardment, plus spheroidal glasses formed by in-flight tend to be ‘contaminated’ with meteoritic siderophile elements quenching of pyroclastic or impact-generated melt splashes, all (e.g., Cushing et al., 1999; Lindstrom and Lindstrom, 1986; are evident in any reasonably large sample of lunar soil (Eugster Warren et al., 1991b; however, cf. Treiman et al., 2010), so the et al., 2000; Keller and McKay, 1997). The lunar regolith is precursor rocks were probably mostly shallow impact breccias conventionally envisaged as having a well-defined lower bound- (brecciation and siderophile contamination being concen- ary, typically about 5–10 m below the surface (McKay et al., trated near the surface), and the heat source was probably 1991a); below the regolith is either (basically) intact rock or most often a proximal mass of impact melt. else a somewhat vaguely defined ‘megaregolith’ of loose but not Besides impactites, which are predominant near the bom- so finely ground-up material. Ancient highland terrains tend to barded surface, virtually all other lunar crustal rocks are igne- have roughly two to three times thicker developments of rego- ous or annealed-igneous. The superarid Moon has never lith than maria (Taylor, 1982). produced (by any conventional definition) sedimentary rock All lunar samples are derived through the regolith, so the and most assuredly has never hosted life. Even metamorphism detailed provenance of any individual lunar sample is rarely is of reduced scope, with scant potential for fluid-driven meta- obvious, and for ancient highland samples, never obvious. The somatism. Sampled lunar metamorphism is virtually confined closest approach to in-place sampling of bedrock came on the to impact-shock and thermal effects. Although regional burial
Apollo 15 mission, when many tens of clearly comagmatic metamorphism may occur (Stewart, 1975), deeply buried basalts were acquired within meters of their 3.3 Ga ‘young’ materials seldom find their way into the surface regolith, (and thus nearly intact) lava flow, so that their collective prov- whence all samples come. Annealing, among lunar samples, is enance is certain (Ryder and Cox, 1996). Lunar meteorites show more likely a product of simple postigneous slow cooling that impacts occasionally eject rocks clear off the Moon. How- (at significant original depth), or dry baking in proximity to ever, in a statistical way, most lunar rocks, even ancient highland an intrusion, or baking within a zone of impact heating. rocks, are found within a few hundred kilometers of their orig- The Moon’s repertoire of geochemical processes may seem inal locations. This conclusion stems from theoretical modeling limited and weird, but the Moon represents a key link between of cratered landscapes (Melosh, 1989; Shoemaker et al., 1970), the sampled asteroids and the terrestrial planets. Four billion plus observational evidence, such as the sharpness of geochem- years ago, at a time when all but microscopic bits of the ical boundaries between lava-flooded maria and adjacent Earth’s dynamic crust were fated for destruction, most of the highlands (e.g., Li and Mustard, 2000). Moon’s crust had already achieved its final configuration. Besides breaking up rock into loose debris, impacts create The Moon thus represents a unique window into the early important proportions of melt. A trace of melt along grain thermal and geochemical state of a moderately large object boundaries may suffice to produce new rock out of formerly in the inner solar system and into the cratering history of loose debris; the resultant rock would be classified as either near-Earth space. regolith breccia or fragmental breccia, depending upon whether surface fines were important in the precursor matter 2.9.2 The Lunar Geochemical Database (Sto¨ffler et al., 1980). Features diagnostic of a surface compo- nent include a smattering of glass spherules (of order 0.1 mm 2.9.2.1 Artificially Acquired Samples in diameter; typically a mix of endogenous mare-pyroclastic Six Apollo missions acquired a total of 382 kg of rocks and soil. glasses and impact-splash glasses) or abundant solar wind- Sampling was mostly either by simple scooping of bulk implanted noble gases (e.g., Eugster et al., 2000). regolith or by collection of large individual samples, mostly Elsewhere, impact melt may constitute a major fraction of much bigger than 0.1 kg (four of the six missions collected the volume of the material that becomes new rock, especially in individual rocks >8 kg). As a result, the particle size distribu- the largest events in which a planet’s gravitational strength limits tion of the overall sample is strongly bimodal. Of the total displacement and the kinetic energy of impact is mainly parti- Apollo collection, rocks big enough to not pass a 10 mm sieve tioned into heat (Melosh, 1989). Rocks formed in this manner comprise 70 wt%, yet the fraction between 1 and 10 mm adds are classified as impact-melt breccia and subclassified based on only 2–3 wt%, and the remaining 27–28 wt% of the material is whether they are clast-poor or clast-rich and whether their <1 mm fines (including core fines) (Vaniman et al., 1991). matrix is crystalline or glassy (Sto¨ffler et al., 1980). Obvious Three Russian unmanned Luna missions added a total of lithic and mineral clasts are very common in impact-melt brec- 0.20 kg of bulk regolith. The Luna samples are more valuable, cias, although the full initial proportion of clasts may not be gram-for-gram, than the Apollo samples because they represent evident in the final breccia. Some of the clasts may be so pul- three distinct sites. However, all nine of the lunar sample- verized that they become lost by digestion into commingled return sites are tightly clustered within the central-eastern superheated impact melt (Simonds et al., 1976). By some defi- region of the Moon’s nearside hemisphere. The nine sites can nitions, the term impact-melt breccia may be applied to prod- be encompassed within a polygon covering just 4.4% of the ucts of melt plus clast mixtures with initial melt proportion as Moon’s surface; if limited to rock-sampling (Apollo) sites, the low as 10 wt% (Papike et al., 1998; Simonds et al., 1976). polygon’s coverage is merely 2.7%. A few impactites feature a recrystallized texture, that is, they consist dominantly of a mosaic of grains meeting at 120 2.9.2.2 Lunar Meteorites triple junctions. These metamorphic rocks, termed granulitic breccias, may form from various precursor igneous or impac- R. L. Korotev (e.g., Korotev, 2005; 2012) maintains a fre- tite rocks, and the heat source may be regional (burial) or local, quently updated Internet list of lunar meteorites (or lunaites).
Treatise on Geochemistry, Second Edition, (2014), vol. 2, pp. 213-250
Author's personal copy
The Moon 215
As of mid-2012, the number of separate, unpaired finds was (Japan, 2007); Chang’e 1 (China, 2007); Chandrayaan 1 (India, rapidly approaching 80. Fortunately, in contrast to the Apollo 2008, with a spectrometry component, M3,fromNASA);Lunar and Luna sites, the distribution of source craters for lunar Reconnaissance Orbiter, LRO (NASA, 2009); Chang’e 2 (China, meteorites is almost random. This conclusion is based on the 2010); and Gravity Recovery and Interior Laboratory, GRAIL randomness of overall cratering on the present-day Moon (NASA, 2011). Among the pre-1994 remote-sensing geochemi- (Gallant et al., 2007; Ito and Malhotra, 2010; Le Feuvre and cal databases, the most notable (only recently superseded) is the Wieczorek, 2011; Morota et al., 2005) (cratering still shows a x-ray spectrometry data obtained for about 10% of the lunar slight bias toward the equator and 90 west longitude and surface on the Apollo 15 and 16 missions. But the key results, against both poles and 90 east) plus the finding (Gault, data for Mg/Si and Mg/Al, suffer from poor precision. For exam- 1983; Gladman et al., 1996) that the vast majority of Moon– ple, the average 1s error in Mg/Si reported for 22 large regions Earth journeys are not direct but involve a phase of geocentric by Adler et al. (1973) is 26%. or even heliocentric orbit. All of the lunar meteorites are finds, Clementine used four different cameras to map the global in some cases significantly weathered (Korotev, 2012). Lunar surface reflectance of the Moon at 11 different wavelengths, provenance is proven for these meteorites by a variety of evi- from the near ultraviolet (415 nm) to the near infrared dence, but the single most useful constraint comes from oxygen (2800 nm), in roughly one million images. Pixel resolution isotopes (Clayton and Mayeda, 1996; Hallis et al., 2010). generally varied from 100 to 300 m, but for selected areas, In addition to the usual potential for pairing among mete- Clementine’s highest resolution camera took 600000 images, oritic finds, lunar meteorites have an important potential for for four different wavelengths (415–750 nm), at typical pixel launch (source-crater) pairing (e.g., Zeigler et al., 2005). resolution of 7–20 m. Clementine multispectral images were Launch pairing can often be ruled out based on cosmic ray used to map surface mineralogy and, for a few elements, chem- exposure (CRE) constraints (e.g., Nishiizumi et al., 2002; Sokol ical composition, and even regolith maturity (i.e., average et al., 2008; Thalmann et al., 1996). The launch age is the sum extent of exposure to surface-regolithic processing) (e.g., of the terrestrial age (usually brief, of order 1–10 ka) plus the Lucey et al., 1998, 2000a,b; Pieters et al., 2002; Staid and p ‘4 ’ (all-directional exposure, Moon-to-Earth) CRE age. Unfor- Pieters, 2001; Wo¨hler et al., 2011). Translation from spectral tunately, the most unambiguous 4p CRE constraints, based on data into concentration is most straightforward for iron, using radioisotopic methods, become increasingly imprecise for CRE primarily data from 900 to 1000 nm, the region of a major þ ages much lower than 1 Ma and hopelessly imprecise for CRE Fe2 absorption band for pyroxene (Lucey et al., 2000a). The ages much less than 0.1 Ma, while physical modeling shows technique for titanium is more indirect, relying on the spectrum that, over time, most lunar meteorites have Moon-to-Earth slope as determined by the ratio between the 415 nm and journeys that take less than 0.1 Ma (Gault, 1983; Gladman longer wavelength (especially 750 nm) reflectances. Clementine et al., 1996). Another way of constraining launch pairing is data provided revelations about cryptomaria (Antonenko et al., based on the expectation that materials from a single source 1995), the petrology of moderately large impact craters as crater will generally show a degree of overall geochemical constraints on lateral and vertical heterogeneities within the similarity. This approach is most useful if one or both of the lunar crust (Cahill et al., 2009; Tompkins and Pieters, 1999), samples are a type of material (e.g., regolith breccia) that tends and compositional variations within the unfortunately remote to show region-specific composition. In this connection, it is (in relation to all Apollo and Luna sampling sites) South Pole– important to realize that the scale of the launch zone is far Aitken basin (Pieters et al., 2001). smaller than the full diameter of a crater, because ejection Lunar Prospector’s two most important geochemical map- velocity is a strong function of proximity to ground zero ping sensors were designed for gamma-ray spectrometry
(Warren, 1994). (GRS) and neutron spectrometry. These techniques measure In principle, a thorough sampling of the lunar surface the upper few decimeters of the regolith, whereas reflectance might yield pieces of Earth impact transported to the Moon and x-ray fluorescence (XRF) spectrometry measure no dee- (Armstrong et al., 2002), a large proportion of which would per than a few micrometers. The difference is not very im- presumably date from the era of heavy bombardment, 3.9 Ga portant, however, because the lunar regolith is extensively and before (see succeeding text). However, Earth’s atmosphere impact gardened. Prospector’s neutron spectrometer was is a major impediment to high-velocity rock ejection mainly designed to map the global distribution of regolith (Gladman and Chan, 2012), and rocks of terrestrial prove- hydrogen. Locally, very high H concentrations found near the nance have not been discovered among the meteorites hitting poles (Feldman et al., 2001) were interpreted as confirming Earth in modern times; only glasses (tektites) are known to Arnold’s (1979) suggestion that water liberated in impacts have reentered from space, and those are believed to have between comets and the Moon might have accumulated in reentered in the immediate aftermath of a cratering event. In cold traps within permanently shadowed regions near the any case, to date, no ‘terrene meteorite’ has been found among lunar poles. the lunar samples. Prospector’s GRS detector was of bismuth germanate type, with similar resolution to the NaI(Tl) detectors flown on the
Apollo 15 and 16 missions. The Prospector GRS database is superior because of vastly longer detector acquisition times 2.9.2.3 Remote-Sensing Data (i.e., better counting statistics) and because Prospector’s cover- The past two decades have been a golden age for lunar remote age was global, whereas the Apollo data covered no more than sensing, thanks to eight major missions: Clementine (BMDO/ 20% of the surface in two near-equatorial bands. The two ele- NASA, 1994); Lunar Prospector (NASA, 1997); SMART-1 ments most amenable to orbital GRS, iron and thorium, were (ESA, 2003); Selenological and Engineering Explorer, SELENE mapped to a spatial resolution of 45 km (Lawrence et al., 2007).
Treatise on Geochemistry, Second Edition, (2014), vol. 2, pp. 213-250
Author's personal copy
216 The Moon
Locally, application of sophisticated deconvolution methods to heterogeneities within the lunar crust (Cahill et al., 2009; conspicuous terrain features allows an effective spatial resolu- Klima et al., 2011). Combined GRS and reflectance data have tion improvement by a factor of 1.5–2. Prospector’s GRS been applied to study the great South Pole–Aitken basin achieved spatial resolution of 60 km for potassium and tita- (Hagerty et al., 2011; cf. Yamamoto et al., 2012) and also the nium (Prettyman et al., 2006), and also for the rare earth ele- Moon’s volumetrically minor but nonetheless important silicic ment (REE) samarium, based on a complex technique involving hills (Besse et al., 2011; Glotch et al., 2010, 2011; Hagerty comparison with data from Prospector’s neutron spectrometer et al., 2006a; Lawrence et al., 2005); of which the most intrigu-
(Elphic et al., 2000). The GRS also yielded maps at spatial ing, because of its unusual (farside) location, is Compton– resolution of 150 km for oxygen, magnesium, aluminum, Belkovich (Jolliff et al., 2011). LRO’s Diviner instrument was silicon, calcium, and thorium (Prettyman et al., 2006). These primarily flown to make detailed diurnal temperature maps of data, particularly for thorium (an exemplary incompatible trace the lunar surface, but it has also been used to map various element; on the Moon, such elements are strongly concentrated aspects of surface composition (e.g., Allen et al., 2012; into KREEP), confirmed early hints from the Apollo GRS that Greenhagen et al., 2010; Kusuma et al., 2012). the Moon’s crust shows a remarkable degree of global geochem- Early results from the XRF spectrometers on SMART-1 and ical asymmetry. For example, average surface concentration of Chandrayaan 1 have been reported by Swinyard et al. (2009), Th is 3.5 times higher on the hemisphere centered over Oceanus Narendranath et al. (2011), and Weider et al. (2012). Both the Procellarum compared to the hemisphere antipodal to Procel- XRF of Chang’e 1 (Wu, 2012) and SELENE’s reflectance spec- larum (although it should be noted that at the low end of trometer (Ohtake et al., 2012) have detected small regions of