The Moon 245
8.5 Kreep Basalt
KREEP basalt, an enigmatic rock type with only a few
undisputed examples, is highly enriched in incompatible
elements (KREEP) but has a more primitive Mg/ (Mg+
Fe^2 +) ratio. This combination of primitive and evolved com-
ponents suggests that they are derived, like the members of
the Mg suite, from different sources and may be impact
melts. Probably the Apennine Bench formation is com-
posed of KREEP basalt. This formation appears to have
formed close in time to the excavation of the Imbrium Basin.
8.6 Breccias
A consequence of the massive bombardment that pulver-
ized the lunar highlands is that the rocks returned from the
lunar highlands are breccias, usually consisting of rock frag-
ments or clasts set in a fine-grained matrix. Lunar breccias
are usually divided into monomict, dimict, and polymict
breccias, consisting, respectively, of a single rock type, two
distinct components, and a variety of rock types and impact
melts. Polymict breccias, usually involving several genera-
tions of breccias, are the most common rock type returned
from the lunar highlands. They are further subdivided into
fragmental breccias, glassy melt breccias, crystalline melt
breccias (or impact melt breccias), clast-poor impact melts,
granulitic breccias, and regolith breccias.
8.7 The Magma Ocean
The geochemical evidence is clear that at least half and
possibly the whole Moon was molten at accretion. This stu-
pendous mass of molten rock is referred to as the “magma
ocean,” and a very energetic mode of origin of the Moon,
such as provided by the giant impact hypothesis, is required
to account for it. The crystallization of such a body is difficult
to constrain, or even imagine, from our limited terrestrial
experience. A possible scenario is that initial crystallization
of olivine and orthopyroxene formed deep cumulates. As
the Al and Ca content of the magma increased, plagio-
clase crystallized and floated in the bone-dry melt, form-
ing rockbergs that eventually coalesced to form the lunar
highland crust, around 4450±20 million years ago. The
first-order variation in thickness from nearside to farside
is probably a relic of primordial convection currents in the
magma ocean. Excavation by large basin impacts has subse-
quently imposed additional substantial variations in crustal
thickness.
Plagioclase was a very early phase to crystallize, as all
lavas derived from the interior bear the signature of prior re-
moval of Eu (and Sr) (see Fig. 19). Accordingly, the magma
ocean was probably enriched in Ca and Al over typical
terrestrial values, a conclusion reinforced by the more re-
centGalileo,Clementine, andLunar Prospectordata. The
implication is that the Moon was enriched in these and
other refractory elements compared to our estimates of the
terrestrial mantle. Continued crystallization of the magma
ocean eventually produced KREEP, which appears to have
pervaded and has been intimately mixed into the highland
crust on the nearside. The crystallization of the magma
ocean was probably asymmetric, as shown by the variations
in crustal thickness and the apparent concentration of the
residual KREEP melt under the nearside. Crystallization
of the main phases was complete by 4400 million years ago,
and the final KREEP residue was solid by about 4360 mil-
lion years ago.
The crystallization sequence portrayed here was far from
peaceful. During all this time, the outer portions of the
Moon were subjected to a continuing bombardment, which
broke up and mixed the various components of the highland
crust. Perhaps coeval with these events was the intrusion
into the crust of the Mg suite. Probably some local over-
turning of the deeper cumulate pile may have occurred,
but such events did not homogenize the interior that later
produced a wide variety of mare basalt compositions.8.8 Lunar Crustal Terranes
Geochemical mapping carried out by theClementineand
Lunar Prospectormissions has resulted in a significant ad-
vance in our understanding of the detailed structure of the
lunar highland crust. Based on the FeO and Th abundances
measured by theClementineandLunar Prospectormis-
sions, the crust can be divided into three major terranes:
(1) the Feldspathic Highland Terrane (FHT), (2) the Pro-
cellarum KREEP Terrane (PKT), and (3) the South Pole–
Aitken Terrane (SPAT) (Fig. 20). The Feldspathic Highland
Terrane constitutes the feldspathic lunar crust formed by
flotation from the magma ocean. The Procellarum KREEP
Terrane results from the intrusion (or mixing in) of the resid-
ual KREEP liquid from the last stages of crystallization of
the magma ocean. The South Pole–Aitken Terrane is the
result of the subsequent excavation of the 2500 km diame-
ter South Pole–Aitken Basin, that stripped off most of the
upper crust over that region and whose ejecta contributed
significantly to the thickness of the farside anorthositic crust,
north of the basin. The interior of the South Pole–Aitken
Basin, the deepest basin on the Moon, has a more mafic
(Fe- and Mg-rich) composition relative to the more felds-
pathic lunar highlands, but it is not clear that the impact
has uncovered the lunar mantle. It would be of great in-
terest to study this area in detail, as no excavated mantle
samples have ever been identified in the returnedApollo
samples.
The South Pole–Aitken Basin, where most of the up-
per crust is missing, has been preserved for over 4.1 billion
years without significant isostatic compensation occurring.
As this is the oldest and largest recognized lunar basin, the