NEEP602 Course Notes (Fall 1996)
Resources from Space

American Mineralogist, Volume 76, pages 773-784, 1991

Evolution of the Moon: Apollo model

P.O. Box 14338, Albuquerque, New Mexico 87191, U.S.A


The major stages of lunar evolution as derived from information obtained from the Apollo and Luna missions can be defined as follows:

1. Beginning-4.55 eons. The moon formed contemporaneously with the Earth. Geophysical data and parentless Pb and volatiles in pyroclastic glasses of Imbrium age derived from deep source regions indicate that initially undifferentiated material accreted below about 480 km.

2. Magma Ocean-4.5-4.4(?) eons. Accretionary melting, volatile depletion, and crystal settling and fioating differentiated the outer 400-500 km of the moon into an anorthositic crust 50-70 km thick and an ultramafic upper mantle containing urKREEP residual liquids.

3. Cratered Highlands-4.4(?)-4.2(?) eons. Impacts capable of forming craters at least 50 km in diameter saturated the Pre-Nectarian lunar crust, producing intense crustal brecciation and increased thermal insulation of the interior.

4. Old Large Basins and Crustal Strengthening-4.2(?)-3.9 eons. Pre-Nectarian large basins formed with rapid isostatic adjustment of the crust. Residual urKREEP liquids apparently moved upward into the pervasively fractured crust. The removal of underlying liquid combined with interlocking intrusions strengthened the crust.

5. Young Large Basins-3.9-3.8 eons. Large impact basins of Nectarian and Imbrium age formed in a crust strong enough to support mascons and mass deficiencies. Materials related to this stage of large basin formation, including ejecta blankets, debris fiows, and shock melted lava, covered most of the near side, if not most of the Moon. A lunar cataclysm of this age does not appear to be required to explain the rarity of older, datable breccia samples.

6. Basaltic Maria-3.8-3.0(?) eons. Mare basalt, produced by downwardly progressing partial melting of the mantle, erupted at the lunar surface and intruded the subsurface. The sequence of nearside surface eruptions proceeded roughly radially from northern Tranquillitatis, the region of minimum gravitational potential. Early volatiledriven eruptions of crustal debris contributed to light plains deposits. Late mafic pyroclastic materials appear to be a mixture of differentiated mantle and undifferentiated volatiles derived from below 400 km.

7. Mature Surface-3.0(?) eons to present. Formation of rayed craters occurred in all regions. Maturation of the lunar regolith continued, including implantation of solar wind gases. Bright swirls formed over large areas.


The Apollo Program gave us a first order understanding of the sequence of events and the processes by which the Moon evolved over 4.55 eons. This conclusion, so far, relates only to events and processes after the Moon's formation near the Earth. Speculation about what happened during formation has intensified in recent years with broad, but not unanimous, support for fission of the already differentiated Earth after collision of a "Mars-sized asteroid" (Hartmann. 1986).

Information about the Moon was accessible even to our distant ancestors as they planned monthly hunting and gathering around ancient African lakes. Had they looked closely at the full Moon, as they may well have done, they would have seen light regions interrupted in some places by sharply defined circular areas. These circular areas and other more irregular patches would have appeared relatively dark. If our ancestors had good eyes and squinted, they would have seen, even then, several very bright spots on both the light and the dark surfaces.

In the 16th century, the observations and logic of Copernicus, following in the footsteps of Aristarchus and probably others before who remain unrecorded, demonstrated that the Moon orbited the Earth while both bodies orbited the sun. Then, in the 17th century, Galileo invented the telescope. Through this remarkable example of human ingenuity and technology, Galileo soon discovered that the light areas were cratered terra, or highlands; the dark areas were less heavily cratered basins, or lowlands, at first mistaken for seas and called maria; and the bright spots were craters with vast systems of bright radial patterns, or rays.

Kepler, Newton, and others, also in the 17th century, provided the mathematical means of calculating the Moon's shape (roughly a sphere), size (radius about 1738 km), mean density (3.34 g/cm3), and the moments of inertia (about 0.400 for the principal moment). These facts, although slightly inaccurate, placed limits on composition, that is, the overall Moon resembles stony meteorites that fall on the Earth and had significantly less Fe relative to the Earth and the sun. The same facts, however, also limited the possible distribution of mass, that is, the Moon must be a roughly uniform solid sphere. With any logical bent, these same observers may have realized that the Moon had no clouds or other apparent protection from the ravages of space and the sun.

Gilbert's (1893) studies in the l9th century marked the beginning of modern considerations about the Moon. During the mid-20th century, indirect analyses by Baldwin (1949), Urey (1952), and Kuiper (1959) contributed new insights about composition, age, shape, motions, and surface properties. In the years just preceding human footsteps on the Moon's surface, and as it became apparent that humans might soon go there, knowledge of this small planet's past expanded rapidly through specialized investigations. Shoemaker (1962), Hackman (1962), and many others (see Wilhelms, 1987) used telescopic photographs and observations, photographs from space probes, and extensions of Steno's law of superposition to add refinements to the more ancient text about the Moon.

The highlands were seen to be saturated with large meteorite impact craters at least 50 km in diameter and clearly constituted the oldest visible crust of the Moon. Many highland basins appeared to contain relatively smooth, lightcolored plains. The analyses by the Surveyor VII spacecraft demonstrated that this light crust contains significantly more Ca and Al than the mare (Turkevich et al., 1969, p. 336). The large circular basins not only penetrated the upper crust but cut across large irregular basins as well. Tracking of Lunar Orbiter spacecraft disclosed that at least some of the youngest large basins overlay significant mass concentrations (Muller and Sjogren, 1968).

Early in this modern but still remote study, observations at low sun angles disclosed that the dark maria were vast plains of remarkably fluid volcanic flows, visually similar to flood basalts on Earth. Surveyor V and Vl analyses confirmed these conclusions (Gault et al., 1968). The flows partially filled many of the large circular and irregular basins on the Moon's near side and covered some of the light plains. Soviet photography of the lunar far side (Whitaker, 1963), as well as later Lunar Orbiter coverage (Kosofsky and El Baz, 1970), showed that the far side of the Moon consisted of an almost continuous expanse of cratered highlands with significantly fewer dark maria than the near side.

It soon became clear that little major internally generated activity had occurred at the Moon's surface after the last major episode of mare formation. As studies became more detailed, an overprint of subtle features became apparent. A lengthy ridge and volcanic island system existed in the Moon's largest western mare region, Procellarum (see, for example, McCauley, 1967); strange sinuous rilles and lightcolored swirls were found; and extensive, relatively young fault systems, in particular, grabens, cut large regions of the surface (see Wilhelms, 1987, chapter 6, p. 244, 256). Overall, a pulverized rock layer, defined as "regolith" (Shoemaker et al., 1967), covered essentially all of the surface, attesting to extensive and probably continuous meteoritic bombardment and radiation from space.

Table 1 outlines the relative sequence of events that formed a plausible model of lunar evolution before any pieces of rock or scoops of soil arrived from the lunar surface.


An early post-Apollo model for the evolution of the Moon proposed by Schmitt (1975a) attempted to provide a consistent framework for the interpretation of Apollo and Luna exploration data (see Fig. 1) then available. Revision and expansion of this model may help to focus the scientific planning of future exploration. Table 2 summarizes the broad characteristics of the updated model.

Beginning: 455 eons

The one largely undisputed conclusion about the origin of the Moon since the acquisition of Apollo samples places its formation at about 4.55 eons (Tatsumoto et al., 1977), apparently contemporaneously with the Earth. The process of formation remains unclear; however, that process probably could not have produced a fully or largely

molten moon. Subsequent differentiation of a molten moon would have eliminated relatively undifferentiated material rich in volatiles (summarized by Delano, 1986) and parentless Pb (Tera and Wasserburg, 1976) required as components in the source materials for orange, black, and green pyroclastic (picritic) volcanic deposits. This constraint suggests a lunar beginning through accretion of initially relatively cold material. Catastrophic separation of the Moon from a preexisting and differentiated Earth, as suggested recently by Hartmann (1986) and many others, may not be able to satisfy such a constraint.

Magma ocean: 4.55-4.4(?) eons

As lunar accretion accelerated, melting (Smith et al., 1970; Wood, 1970) and volatile depletion of the outer several hundred kilometers of the Moon occurred, creating a melted shell (Schmitt, 1975a) or magma ocean (Walker, 1975; Wood, 1975; Warren, 1985). The more commonly used term, magma ocean, designates this first major stage of lunar evolution. Differentiation of the magma ocean (see review by Taylor, 1982, chapter 8) resulted in a roughly 5070 km thick crust of ferroan (Dowty et al., 1974) anorthosite and olivine norite in roughly equal proportions (Korotev et al., 1980) and an ultramafic upper mantle several hundred kilometers thick, possibly with its base at 400480 km (Goins et al., 1979). The residual liquid in the differentiating magma gradually took on the chemical characteristics attributed to urKREEP (Warren and Wasson, 1979; Warren, 1988), the proposed parent of rocks rich in K, REE, and P described initially by Hubbard et al. (1971). Left undisturbed, this liquid would probably remain largely uncrystallized owing to concentration of radioisotopes and the insulating nature of the increasingly brecciated crust.

Prior to the development of a physically coherent crust, large accretionary impacts would have mixed the stilldifferentiating melt into the protocrust, adding both an overall gabbroic contaminant, the olivine norite of Korotev et al. (1980) and many impactforced intrusions and extrusions. Once the crust could maintain its overall coherency in the face of continued high but decreasing impact frequency, strong evidence exists that intrusions from the differentiating mantle magma were emplaced (Warren and Wasson, 1980; James, 1980). The local differentiation of these intrusions appears to be the source of the Mgrich suite of pristine clasts found in highland breccias such as those reviewed by Prinz and Keil (1977).

Crystallization ages greater than 4.4 eons (see Wilhelms, 1987, Table 8.4) measured for many of the examples of pristine mafic and ultramafic clasts strongly suggest that these may constitute remnants of differentiated intrusions from this period. As might be expected from this process of late accretionary mixing of differentiating magma, Pieters's (1987) spectroscopic studies of the central peaks and ejecta blankets of Copernican and Eratosthenian craters indicate the presence of large bodies of Mgrich rocks below the upper 10 or so km of the crust. The lower age limit of 4.4(?) eons assigned to the magma ocean stage corresponds to the minimum crystallization ages measured for clearcut samples of the Mgrich suite of highland breccia clasts.

Late in the magma ocean stage, uncrystallized and relatively low density urKREEP would have concentrated near the base of the still forming and continuously brecciated crust. (See Schmitt, 1975a; Ryder and Wood, 1977; Warren and Wasson, 1979; however, Ryder, 1990, now advocates cataclysmic brecciation at 3.9 eons, discussed subsequently.) In response to particularly large impacts, some urKREEP also may have been separated from the mantle and incorporated in the crust either as disseminated ejecta or as coherent intrusions. KREEP model ages of 4.4-4.3 b.y. (Lugmair and Carlson, 1978; summarized by Taylor, 1982) support this conclusion and support an overlap between the magma ocean stage and the following cratered highlands stage.

During accretion and melting, any immiscible ironsulfur liquid could be expected to settle quickly to the base of the magma ocean. Ultimately, as radioisotopic heating permitted solid state flow, this liquid would have moved toward the center of the Moon. Except for some purging of chalcophile and siderophile elements, the downwardly migrating ironsulfur liquid probably did not affect the composition of undifferentiated and still relatively cool lower mantle material. The formation of a lunar core has not been conclusively proved (see Wiskerchen and Sonett, 1977; Goins et al., 1979; summarized by Taylor, 1982, p. 358359); however, the weight of evidence suggests the presence of a fluid core with an upper limit of about 500 km in radius.

Table 3 summarizes the major lunar features present at the completion of the magma ocean stage of lunar evolution.

Cratered highlands: 4.4(?)-4.2(?) eons

The cratered highlands stage of lunar evolution (early Pre-Nectarian) represents a time between about 4.4(?) and 4.2(?) eons when impacts on the lunar surface produced a saturation of craters as large as 60-70 km, that is, curves of crater size vs. frequency approach a slope of I at these diameters (Wilhelms, 1987, p. 145). The lower age limit for this stage has been selected as that suggested by Taylor (1982, p. 238-240) for the resetting of most radioisotopic Ar clocks of possible Pre-Nectarian age; however, a final selection of a lower limit requires samples that clearly record Pre-Nectarian events. As discussed below, the possibility remains that no such samples have yet been recognized or collected.

Significant regional homogenization (Pieters, 1987) of the upper crust and intense brecciation of the lower crust to at least a 25km depth (Toksoz, 1974) occurred during this stage. Any extremely large impact events, such as those proposed for Procellarum and South PoleAitken (see Wilhelms, 1987, p. 145), would have disrupted and thinned the upper crust and potentially triggered surface eruptions of KREEP-related lavas (Wilhelms, 1987, p. 143144). The increasingly insulating character of the progressively more intensely brecciated upper crust allowed the gradual accumulation of radiogenic heat necessary to eventually partially remelt source regions in the upper mantle that produced Imbrium and younger basaltic lavas.

The cratered highlands stage merged with the next stage as the overall frequency of impacts declined. Table 4 summarizes this stage's modification of features present at the end of the magma ocean stage.

Old large basins and crustal strengthening: 4.2(?)-3.9 eons

All but the largest of the 29 clearly recognizable Pre-Nectarian large basins (Wilhelms,1987, chapter 8) formed after the period of intense overall cratering had waned. Otherwise most surface expression of basin structure would have been obscured. Qualitative comparison of the Pre-Nectarian large basins with younger large basins (see Wilhelms, 1987, Fig. 5.22, Plates 5 and 6) indicates that major strengthening of the lunar crust occurred during this stage. The younger basins (Nectarian and Lower Imbrium systems) are sharply defined and circular. Central mass concentrations or mascons (Muller and Sjogren, 1968) surrounded by mass deficiencies under mountain rims several thousand meters high underlie the young basins. The Pre-Nectarian older basins are only irregularly circular with relatively low rims and are largely compensated isostatically.

The absence of significant central mascons and rimmass deficiencies associated with Pre-Nectarian basins demonstrates that just prior to Nectarian time, the lunar crust had little strength to resist isostatic adjustment. Although the selection of the Nectaris impact event as the beginning of this time unit evolved from the lunar mapping program of the 1960s and 1970s (Stuart-Alexander and Wilhelms, 1975), the usefulness of its selection is reinforced by the Nectaris Basin being the oldest of the mascon basins. The old large basins and crustal strengthening stage of lunar evolution therefore encompasses major crustal changes as well as the formation of all the still recognizable nonmascon large basins. The age limit on the end of the stage coincides with the formation of Nectaris, concluded by Wilhelms (1987, p. 168-170) to be about 3.92 eons.

The formation of Pre-Nectarian basins in the now coherent crust had three major effects: (1) redistribution of upper crustal material as ejecta, ejecta blankets, debris flows, and impact melt flows, (2) regional thinning and thickening of the crust (see Bills and Ferrari, 1977, Fig. 3), and (3) fracturing of the lower crust beneath each basin. After each large basin formed, the relatively low density urKREEP liquid would move upward into the deeply and closely fractured crust (Schmitt, 1988). Following the excavation of the deepest of the older large basins formed about 4.2 Ga, KREEP-related lavas may have locally reached the floor of such basins (see Wilhelms, 1987, p. 143-144).

Becoming cooler and significantly contaminated with unconsolidated and largely anorthositic crustal debris, the KREEP-related liquids could be expected to crystallize into networks of compositionally varied intrusions. Removal of underlying urKREEP liquid, combined with an interlocking boxwork of solidified intrusions, would have strengthened the crust. Crystallization ages of 4.2 eons for a few KREEP-related materials (Nyquist et al., 1977) may support this hypothesis of Pre-Nectarian solidification of KREEP-related intrusions.

Finally, the downward migration of ironsulfur liquid, which began in the melted shell stage, may have been completed in this stage. This would be suggested if the activation of a dynamorelated dipole magnetic field created the measured remnant magnetic fields (see Taylor, 1982, p. 364-370).

Table 5 summarizes changes in major lunar features produced by the old large basin and crustal strengthening stage of lunar evolution.

Young large basins: 3.9-3.8 eons

Between 3.9 and 3.8 eons, 14 large impact basins of Nectarian and Lower Imbrium age (Wilhelms, 1987, chapters 9 and 10) formed in a crust strong enough to support significant mascons and mass deficiencies. The initial crustal penetration of the Imbrium event apparently excavated particularly large quantities of KREEP-related intrusions or flow material and distributed it to where Apollo 14 sampling (see Swann et al., 1977) and geochemical remote sensing (Metzger et al., 1977) found it concentrated. The work of the Consortium Indomitabile (1974) on a layered breccia boulder examined and sampled at the base of mountains bordering Serenitatis strongly suggests that other basinforming events also excavated significant quantities of now buried KREEP-related material (see Schmitt, 1975b).

Tera et al. (1974) suggested that a lunar cataclysm, including the events of the cratered highlands stage and the formation of all large basins, occurred between 3.9 and 3.8 eons. This suggestion recently has been revived in detail by Ryder (1990). A cataclysmic compression of all the events associated with the formation of old and young large basins as well as the crater saturation in the highlands does not appear to be supported by the analysis leading to the model of lunar evolution presented here.

First, it seems unlikely that there would be a hiatus of half a billion years in significant impact activity after such a remarkably violent accretionary beginning. Second, the cataclysm hypothesis requires that within 100 m.y., there must occur both a much compressed cratered highlands stage and the formation of two groups of large basins with markedly different relative ages and physical characteristics. Third, within the same 100 m.y. and between the formation of the two groups of large basins, the crust must strengthen significantly. Fourth, the ages of impact melts sampled by Apollo 16 in the highlands show ages that range from 3.7 to 4.2 eons (James, 1981), indicating that at least some pre-3.9 eon impact melts exist where one might expect to find them.

Finally, in the context of the cataclysm hypothesis, it should be noted that the sampling of impact breccias and melts during Apollo and Luna missions took place well within the portion of the Moon most affected by Nectarian-age cratering and basin formation, which may account for a major part of the 3.9-3.8 eon bias in impact melt ages. In addition, as suggested by Wilhelms (1987, p. 190 -191), a significant concentration of radiometric ages can be expected statistically from the natural consequences of a declining cratering rate. A final resolution of this controversy, however, may have to await a broader and more geologically selective collection of samples from other regions of the Moon.

The formation of large basins makes major modifications to the crust many crater diameters beyond the initial point of excavation. Lithostatic resistance at depth appears to divert the force of the impact radially both in the effects of crustal deformation and in the movement of ballistic and flowing ejecta. Lunar photogeologic mapping of Orientale (McCauley, 1987) and investigations of large breccia boulders by Apollo 17 (Schmitt and Cernan, 1972) at the Valley of TaurusLittrow traverse Stations 2, 6 (Fig. 2), and 7 (in the mountain ring surrounding the Serenitatis Basin) illustrate at different scales the complexity of processes that characterized the formation and interaction of large basins. A definitive study has been made (Consortium Indomitabile, 1974) of Boulder 1 at Station 2 at the base of the South Massif. The geological model for the formation of this complexly layered breccia presented by Schmitt (1975b) spans most of the first onehalf billion years of lunar evolution. At least three major dynamic events that modified and deposited impact ejecta contributed to Boulder 1's present structure, geochemistry, and petrology.

An additional view into the processes of large basin formation comes from the very large breccia boulder described and sampled at Station 6 near the base of the North Massif of the Valley of TaurusLittrow (Schmitt and Cernan, 1972; Schmitt, 1973; Apollo Field Geology Investigation Team, 1973). The Station 6 boulder had rolled down the side of the North Massif from a point about 1200 m below the top of the massif. The boulder (Fig. 2) included a contact between blue-gray and tangray breccias, both containing numerous clasts of the anorthositic and Mg-rich suites of rocks. The blue-gray breccia contains small vesicles within about 1 m of its contact with the coarsely vesicular tangray breccia. This evidence of high-temperature contact metamorphism indicates that the tan-gray breccia intruded the bluegray breccia as a partially molten mass with sufficient heat to melt portions of its host.

After the completion of the two stages of large basin formation defined here, regional deposits of crustal ejecta and shock-melted lava covered many regions of the Moon. Except for the shock-melted lava exposed inside the outer Rook Mountain ring of the Orientale Basin, the Maunder Formation (McCauley, 1987, p. 76-77), younger mare units cover most of such material inside the basins themselves. Impact events into shock-melted lavas, however, would have distributed samples of this type over most of the Moon, a fact that should be considered in the interpretation of fragments of Alrich rock types in the lunar regolith.

Lunarwide debris deposits in closed basins, now recognized as light plains, terra plains, and Cayley plains (Wilhelms, 1987, p. 216-220) probably have several origins. Many clearly are deposits of fine gas-charged debris ejected during large impact events and transported across the lunar surface as regional debris flows. Such flows would have settled preferentially in basins, leaving relatively smooth plains. Orbital observations of the lunar far side (Schmitt, unpublished data, 1972) included views of large, markedly smooth floored craters without visible central peaks, suggesting that the central peaks as well as any irregular floor and slump material has been covered by later deposits. The absence of younger maria except in craters excavated in the deepest of these smooth floored craters, has left these relationships exposed.

Although relatively shortlived, the young large basins stage resulted in major modifications to the model of lunar evolution presented here. Table 6 summarizes these changes.

Evolution of the Moon: Apollo Model (continued)

Back to Lecture 10 (fall'96)

Back to Lecture 10 (fall'97)

Back to Syllabus (fall'96)

Back to Syllabus (fall'97)

University of Wisconsin logo

University of Wisconsin Fusion Technology Institute  · 439 Engineering Research Building  · 1500 Engineering Drive  · Madison WI 53706-1609  · Telephone: (608) 263-2352  · Fax: (608) 263-4499  · Email:

Copyright © 2003 The Board of Regents of the University of Wisconsin System. For feedback or accessibility issues, contact
This page last updated August 21, 2003.