Giant impact: accretion and evolution of the Moon - Implications for Earth, Mars and the Solar System as a whole
IMPLICATIONS FOR EARTH
THE SOLAR SYSTEM AS A WHOLE 2
0 Jacobs University Bremen
1 Jungfraujoch Commission - Swiss Academy of Sciences
2 International Space Science Institute , Bern
Our planetary system has not always been as serene as it appears to us today. Exploration of the Moon has shown that disastrous collisions and violent epochs have occurred in the early part of its history. Indeed, a collision of the Earth with another planet - the Giant Impact - is the most widely accepted theory for the origin of the Moon. Several hundred million years later, Moon and Earth received a Late Heavy Bombardment that created the large basins on the Moon and must have devastated the atmosphere and hydrosphere of the Earth.
ACCRETION AND EVOLUTION OF THE MOON
Tthey are unlikely. It has been proposed, for example, that the
he exact place and time of such disastrous events are
impossible to accurately retrace in time, but that does not mean
high density of Mercury is due to a giant impact as well, or
that Uranus and Neptune migrated outwards, ravaging small-object
populations until reaching their final positions several hundred million
years after their formation. Also the existence of extra-solar “Hot Jupiters”
– large planets circling close to a star – is explained by planet migration.
Obviously, the chance for life to form and survive on Earth, on Mars
or on exo-planets is affected by the intensity of such violent epochs in
The Giant Impact
The Giant Impact (depicted in Fig. 1) is consistent with major lunar
observations, namely (i) the Moon’s exceptionally large size relative to
its host planet; (ii) the high angular momentum of the Earth-Moon
system; (iii) the Moon’s extreme depletion of volatile elements; and (iv)
a differentiated global crust and mantle of the Moon that had quickly
followed its delayed formation.
For the most part, the latter two findings are based on chemical
analyses and age determinations of lunar rocks that had been brought back
to Earth by the Apollo astronauts between 1969 and 1972.
The Giant Impact set the initial conditions for the formation and
evolution of the Moon. The impactor – sometimes called Theia – is
assumed to have had at least the size of Mars and to have hit the Earth at
an oblique angle. The collision produced a dense protolunar cloud. Fast
accretion of the Moon from this cloud assured an effective storage of
gravitational energy as heat, producing early melting. A Magma Ocean
of global dimensions formed, and upon cooling, solidified into a crust2
and a mantle3 (cf. left-hand part of Fig. 2).
The Giant Impact occurred between 70 Ma and 110 Ma after the
solar-system matter had become isolated from galactic
]. At that time, heat and accretion were the only energy sources
capable of providing the energy for melting and differentiation of a
lunar-sized body4. Heating occurs near its surface and thus much heat
1 Editor’s Note: This special Feature is an updated summary of a paper in which
J. Geiss and A.P. Rossi [The Astronomy and Astrophysics Review (2013) On the
chronology of lunar origin and evolution – Implications for Earth, Mars and
the Solar System as a whole, 21:68 (54 pp)] covered the development of lunar
science over the past half-century and underlined the importance of
understanding the Moon’s history for the history of the Solar System as a whole.
Detailed physical arguments, models, numerical estimates and references can
be found in the review paper itself. Jo Hermans, EPN Science Editor
2 The lunar crust is feldspar-rich, composed mainly of anorthite CaAl2Si2O8,
with a density of ρ = 2.67 g cm-3.
3 The mantle is composed mainly of olivines and pyroxenes (Mg,Fe)SiO3 and
(Mg,Fe)2SiO etc, with densities of ρ ≈ 3.5 g cm-3 and ρ ≈ 3.6 g cm-3, respectively.
4 Decay of 26Al with a half-life of 0.7 Ma was the principal energy source for
melting and differentiating asteroids very early in Solar-System history. At the
time of the formation of the Moon, however, 26Al had disappeared and the
long-lived isotopes of K, U and Th with half-lives between 0.7 Ga and 14 Ga
are much too slow to produce the fast melting that is inferred from the ages
of lunar crust material.
m FIG. 1:
Origin of the moon
in a Giant Impact.
Paintings by William
the Earth– moon
system at 30 min,
5 h and 1000 a,
is lost by radiation with increasing temperature unless
a very fast accretion “buries” the heat as it is produced.
Accretion times as short as ≈ 103 a are needed, and model
calculations show that this is achieved at the high density
existing in the protolunar cloud that had been created as
a consequence of the Giant Impact.
During the formation of the Moon a poorly defined
intermediate layer rich in incompatible elements called
KREEP developed between crust and mantle5.
The maria: basins being filled by lava
Several 100 Ma after the Moon’s accretion, the decay of
the long-lived isotopes (40K, 232Th,235U and 238U) that were
concentrated in the KREEP-rich layer below the crust
had generated enough internal heat to induce partial
melting in the mantle. Lava extruded into large basins –
i.e., into the depressions that had been excavated on the
Moon as part of the Late Heavy Bombardment. Upon
cooling, the lava solidified into titanium-rich mare basalt
(cf. right-hand part of Fig. 2). This era of rock formation
may actually have lasted for nearly 3 Ga, from about 3.9
Ga before the present until ca. 1 Ga ago.
Mare basalt surfaces and other geologic units have
been identified on the lunar near side by use of chemistry
data from lunar orbit (i.e., from remote sensing in spectral
domains ranging from γ- and X-rays to the infrared).
How a relative timescale was established through crater
counting, and how this timescale was then calibrated by
radiometric dating of rocks returned from six Apollo
5 KREEP stands for potassium (K), rare earth elements (REE) and phosphorus
(P), which together with uranium and thorium had accumulated between
crust and mantle.
6 Three of the USSR Luna spacecraft returned samples that they had collected at
their landing spots.
7 ∆ 17O is defined such as to be unaffected by mass dependent chemical or physical
landing regions and three Luna6 landing spots is
presented in Box B. Detailed results, including geographical
information are shown in Fig. 3.
The general, coherent explanation of the lunar origin
and evolution sketched above is by and large complete.
There remain some questions that still need to be
resolved. In the following we briefly mention two such
issues. The first one shows the precision of some lunar
measurements, the second one also sheds light on early
Constraints from isotopes: the “Oxygen Isotope Puzzle”
The abundances of the chemical elements and their
isotopes – as measured in the returned lunar samples – are
generally compatible with the fractionation processes
that go along with, and follow the Giant Impact
origin of the Moon. But there remains an odd agreement
of isotope ratios, the so-called Oxygen Isotope Puzzle,
which might not a priori be expected as a consequence
of that scenario.
The “intrinsic” isotopic composition of an object is
commonly represented as ∆ 17O, the excess – or deficit – of
17O, relative to its terrestrial abundance7. For
solar-system objects, |∆ 17O| values are very small, typically 1‰
or less. For Mars, for example, ∆ 17O = 0.3 ‰ has been
determined8. Surprisingly, however, there is no significant
difference in the oxygen isotopic composition between
Moon and Earth.
Models have been presented indicating that this puzzle
is probably not severe enough for falsifying the Giant
Impact scenario. In any case, the need for explanation
would be less severe if |∆ 17O| of Theia was smaller than
the Martian value of 0.3 ‰. This is quite possible, since
Theia, by definition, crossed the Earth orbit during the
later phase of planet accretion, while Mars accreted its
matter farther away from Earth. So it is quite possible
that Theia’s |∆ 17O| was very small.
The lunar dichotomy – scientific and political surprise
In October 1959, only two years after the USSR had
launched the first artificial satellite, Sputnik 1, into orbit,
another Russian spacecraft, Luna-3, took images of the
far side of the Moon – and that side turned out to be very
different from the near side!9 Most notably, the typical
maria seen on the near side (cf. Fig. 3) are almost absent.
Photographing the far side of the Moon was a
technological feat: it meant on-board developing and scanning
of the exposed film, so that the images could be sent back
to Earth by telemetry10.
Some of the suggested explanations for the dichotomy
can be further tested by altimetry, and by gravimetric and
chemical analysis from lunar orbit. However, to definitely
decide what caused the dichotomy and when it happened
might require the return and analysis of documented
samples from the far side of the Moon11.
We now turn to the use of the lunar timescale for other
planets, in this case Mars, and finally discuss the lunar
evidence for the Late Heavy Bombardment.
Towards a Chronology of Mars
The spatial density of craters (the ‘crater frequency’) can
be used to infer information about the age of surfaces.
Crater frequency ages and ages of Martian meteorites
provide growing evidence for young lava flows, fluvial
activity and aeolian deposition on Mars. In Fig. 4 plots
of cumulative crater frequency12 vs. diameter are shown
for the ejecta blanket of the lunar North Ray Crater and
for a young volcanic area on Mars.
The crater frequency age of 46.5 Ma for the lunar
North-Ray Crater is in good agreement with its exposure
age of 50 Ma. Due to “resurfacing”, the cumulative crater
frequency levels off at small diameters in both plots, but
as expected, “resurfacing” for Martian areas is much faster
than it is for lunar areas.
Establishing a relative timescale for Mars is now
progressing fast: in the absence of documented samples from
Mars, crater frequencies are being provisionally converted
to actual ages (in Ma) by use of a theoretical Mars
So far, calibration of crater frequency ages with
radiometric ages or exposure ages of meteorites has not
been achieved, but this is beginning to change. In a
recent paper [
] it is concluded that the 55 km wide
Martian Mojave Crater is the source of many Shergottites8,
8 So far no samples have been returned from Mars, but
the rare SNC meteorites are known to come from Mars.
(SNC stands for the meteorite classes named after the
place where the first examples had been found: Shergotti,
9 The Moon’s rotation is tidally locked, i.e., its orbital and
rotational periods are the same; so there is a ‘far side’ of the Moon,
which we never see from Earth.
10 This happened long before today’s ubiquitous CCDs
11 Detailed exploration of the lunar far side is not
straightforward. It will most likely require telecommunication
via a lunar relay satellite, since electromagnetic waves
from the far side of the Moon do not reach the Earth.
12 Note that crater frequencies are spatial frequencies
(normally they have the dimension km-2).
b FIG. 2 (left) The fast
the Giant Impact led
to a magma Ocean
covering the lunar
surface, partly or
fully. Before the
towards the surface
and the heavier
towards the centre.
in an intermediate
layer. (right) Several
100 ma after lunar
accretion, lava that
was melted in the
mantle by the decay
of long-lived isotopes
of K, U and Th
extruded into large
basins and solidified
mare basalt. The
ages indicated refer
to the Mare Imbrium
and to mare basalt
samples collected at
the Apollo 15 landing
site (cf. Fig. 3).
because the meteorite ejection ages coincide with the
crater frequency age of that crater as calibrated with
a theoretical Mars Production Function. Both ages
cluster around 3 Ma. The expected production rate
on Mars for Craters with a diameter D > 55 km is less
than one in 10 Ma. Thus having a competitor for the
Mojave Crater as source for the 3-Ma-Shergottites is
rather unlikely. In fact, the ample ejection of meteorites
from the young Mojave Crater might be the reason
why in our epoch there are more Martian meteorites
than lunar ones!
Clustering of exposure ages is observed not only for
Shergottites, but also for other meteorite classes,
suggesting that crater production rates are likely to vary on
timescales of tens of Ma. For older surfaces such
variations will be evened out when observing Cumulative
The Late Heavy Bombardment
Radiometric dating revealed that four of the prominent large
lunar basins were created during a relatively short time
interval, about 4.1 Ga to 3.8 Ga before present (cf. Fig. 3 and
the illustration in Box A). When stratigraphic observations
showed that many of the other large basins on the near side
of the Moon were also excavated during or close to this
period, it was realised that the Moon had received a Late Heavy
Bombardment (LHB), about 0.5 Ga to 1.6 Ga after the Moon
had been accreted. Such a late clustering of major impacts
must have been caused by a special phase in the history of
the solar system, separate from the main accretion period.
Was there a major perturbation in the planetary system so
late in its history, and what could have been its cause?
Three discoveries, the lunar LHB, Trans-Neptunian
objects, and Jupiter-sized Exoplanets unexpectedly close
to their central stars[
] have renewed interest in planet
m FIG. 3 Two images of the near side of the moon with chronological
information. (left) Radiometric ages of four large basins (purple) and basalt
samples from six mare areas (blue) derived from analyses of samples collected
in the Apollo landing areas and the Luna landing spots (both of them are
marked on the right-hand image). (right)Crater frequency ages of mare areas
from high-resolution crater counting and mapping.The youngest mare areas
are in the OceanusProcellarum to theWest of MareImbrium; they correspond
to mare areas with a high KREEP content.
b FIG. 4 Crater frequency ages for the lunar North Ray Crater [
] and a young
volcanic area on mars.[
] Images used were from the NASA Lunar and mars
Reconnaissance Orbiters, respectively. The Crater Frequency ages given in
these plots were calculated by using the theoretical Production-Function
and Chronology-Function given by Ivanov [
] (see Box B).
migration, and led to the development of Nice Models13
for explaining late heavy bombardment epochs by planet
rearrangement well after the birth of the Solar System.
The classical Nice model assumes planets to have been
accreted from the proto-planetary disc more closely
together than we find them today. After dissipation of the
gaseous portion of the solar nebula, the planets slowly
migrated, due to their interacting with the remaining
planetesimals. Saturn moved away from Jupiter, increasing the
ratio between the periods of the two planets (PS and PJ),
until they got into the 1: 2 resonance at Pj/PS = 1/2 (today,
we have Pj/PS =1/2.48). The orbit of Saturn became very
eccentric, causing close encounters among large planets
and “a massive delivery of planetesimals to the inner solar
system” that could have caused the lunar LHB.[
For comparisons with specific Nice-model
predictions, the lunar LHB would have to be better defined than
it is today. The Orientale basin is the youngest among the
major lunar basins and may mark the end of the lunar
LHB. Its beginning is difficult to measure and even to
define. Determining instead the declining phase of the
bombardment by measuring the number of
basin-forming impacts in a given time interval near 4 Ga would be
easier and could significantly constrain theories of LHB
origin. Geiss & Rossi1 suggest that two
rover/samplereturn missions would do the job, one sampling the
blanket of the Orientale basin and the other exploring a region
to the Southeast of Mare Nectaris14.
The heavy bombardment of the Moon about 4.1 Ga to
3.8 Ga ago must have hit the Earth as well, causing lasting
devastation of our atmosphere and hydrosphere15.
Quantitative comparison of the effects of bombardment of Moon
and Earth depends on the origin of the impactor. For
objects arriving with high energy from, e.g., the region beyond
Uranus, impact energies on the Earth are only modestly
increased, and the Moon and Earth ratio of impact rates
are nearly proportional to their geometrical cross sections
(1: 13.4). On the other hand, for objects coming from the
inner edge of the asteroid belt, focussing and acceleration
by the terrestrial gravitational field strongly increases the
relative flux and also the impacting energy hitting the Earth.
Remnants of basins produced on Earth during the
Archean and Proterozoic epochs have been wiped out
by plate tectonics. However, the distribution of spherule
beds, interpreted as ejecta from large impacts, indicate
13 Nice models were first developed in Nice (France).
14 Following the delivery to the lunar surface of the Chinese
rover Chang’e 3 in December 2013, a sample return mission,
Chang’e 5 is scheduled for 2017.
15 The energy of the Imbrium impact on the Moon was
roughly a thousand times higher than the energy of the
Chixculub impact on Earth that caused the mass extinction at the
BOX A: TIMESCALES AND CHRONOLOGIES[
On Earth, a global relation was found in the 19th century between fossil
species and sedimentary rock strata, establishing a relative chronology
from the Cambrian to the present. In the 20th century this chronology
was calibrated by radiometric dating and extended deep into the
Precambrian. On the Moon, where there are no fossils, relative timescales
are based on the observed impact crater frequencies. Crater count ages
have now been determined for a large portion of the lunar surface and
calibrated with radiometric and some exposure ages of lunar samples
returned from six Apollo landing areas and three Luna landing spots.
Even from the rather limited geographical coverage available after the
Apollo and Luna missions, it was concluded that the crater production
rate decreased strongly for ages larger than 3.2 Ga and remained
approximately constant afterwards. By approximating the crater
production rate by a constant plus an exponential, Neukum, Ivanov and
] introduced the Cumulative Crater FrequencyTimescale
shown in the illustration below; this has since been the standard. The
timescale is well established only in the time interval 4 Ga to 3 Ga.
It cannot be much extended beyond 4 Ga, because there the crater
frequency is approaching saturation. Samples from the young areas
to the West of Mare Imbrium (cf. Fig. 3) could close the gap in the
calibrated timescale from 3 Ga to 1 Ga. Completing the lunar crater
count timescale is of foremost importance, because it will remain
indispensable for quite some time, as it can help for transferring
absolute timescales to Mars and other solar system bodies.
that basin forming impacts on Earth lasted to 3 Ga
before the present and beyond.[
] The lunar LHB ended
earlier, as inferred from the crater frequency curve in
the illustration of Box A or from the estimated age of
the Orientale basin.
The earliest, well-established traces of life appeared
about 3.5 Ga ago. The analysis of spherule beds implies,
however, that the epoch of heavy bombardment on Earth
had not yet ended at that time. Further data on the LHB
and post-LHB epochs on Moon, Earth and Mars as well
as on early traces of life on Earth might tell us whether
a heavy bombardment had delayed the creation of life
on our planet, whether all life was extinguished that had
existed before, or whether a primitive life form managed
somehow to survive the epoch of heavy bombardment. n
BOX B: AGE DETERMINATION
Radiometric Ages are determined by radioactive dating. This
technique is based on a comparison between the observed abundance
of naturally occurring radioactive isotopes and their decay products,
by use of the known decay rates.
Cosmic Ray Exposure Ages are based on the following model: a
collision excavates a rock that was buried and screened from cosmic
rays and ejects it into an orbit around the Sun, where it is fully
exposed to cosmic rays. Eventually, the orbiting rock will collide with
the Earth, where cosmic-ray exposure will virtually cease. The time
of exposure to cosmic rays is determined from nuclides produced by
cosmic-ray-induced nuclear reactions.
An exposure age (TEXP) is calculated from the abundances of a pair of a
stable (NST) and a radioactive (NRAD) cosmic-ray-produced nuclide, the
corresponding production rates PST and PRAD, and the decay constant
λ of the radioactive nuclide. If PST and PRAD were constant over the time
interval TEXP to T0 , then for the simple case of λT0 = 0 and λTEXP >>1,
the exposure age is given by
1 PRAD NST
TEXP = –——
λ PST NRAD
This relation is valid only if the irradiation conditions (cosmic-ray
intensity and energy spectrum, heliospheric modulation, shielding by
atmosphere or overlying solid material) were constant from TEXP to
T0. These conditions are usually met when calculating the time of
ejection of a meteorite from an asteroid, Mars or Moon.
Crater Count Ages are defined as the Cumulative-Crater-Frequency
[km-2] = Production-Function (D) x Chronology-Function (T) [km-2].
Because it is possible to separate the variables, the data can be
presented in two types of plots:
• Cumulative-Crater-Frequency versus Age (see graph in BOX A) and
• Cumulative-Crater-Frequency versus D, i.e., the lower integration
limit for the crater diameter.
The slope of the plot (cf. Fig. 4) determines the age of the cratered
area. This must, however, be calibrated by radiometric or exposure
ages of rocks, determined in the laboratory.
About the Authors
Johannes Geiss is Honorary Director
and was Executive Director (1995-2002)
of the International Space Science
Institute in Bern. He was Professor of Physics
at the University of Bern (1960-1991),
Associate Professor at the Marine
Laboratory of the University of Miami (1958-1959) and
Research Associate at the Enrico Fermi Institute, University
of Chicago (1955-1956). He is a Foreign Associate of the
US National Academy of Sciences and a Member of the
Academia Europaea. He received the NASA Medal for
Exceptional Scientific Achievement, the Albert Einstein
Medal of the Einstein Association in Bern, and the Bowie
Medal of the American Geophysical Union.
Martin C.E. Huber was Adjunct
Professor at ETH Zurich, Head of ESA’s Space
Science Department and President of
EPS (2004-2005). He was closely
associated with Europhysics Letters (now
EPL) and is a member of the Editorial
Board of The Astronomy and Astrophysics Review. He is
an Associate of Harvard College Observatory, a Member
of the Academia Europaea and immediate past President
of the Kommission für die Hochalpine Forschungsstation
Jungfraujoch of the Swiss Academy of Sciences (SCNAT).
Angelo Pio Rossi is a planetary scientist
based at Jacobs University Bremen. He
is a geologist by background and works
on Earth and Planetary Remote
Sensing and Comparative Planetology. He
is a Co-I of the High Resolution Stereo
Camera on board of ESA’s Mars Express and a member of
the Editorial Boards of Planetary and Space Science and
EGU Solid Earth. Previously he was a Research Fellow
in ESA’s Science Directorate and Staff Scientist in the
International Space Science Institute.
 see A.N. Halliday , Phil Trans R Soc A 366 , 4163 ( 2008 ).
 H. Hiesinger , C.H. Van der Bogert , J.H. Pasckert et al., J Geophys Res 117 , E00H10 ( 2010 ).
 F. Jagert , E. Hauber, Geoinf 2012 ( 2 ), 177 ( 2012 ).
 B.A. Ivanov , Space Sci Rev 96 , 87 ( 2001 ).
 S.C. Werner , A. Ody , F. Poulet , Science 343 , 1343 ( 2014 ).
 M. Mayor & D. Queloz , Nature 378 , 355 ( 1995 ).
 R. Gomes , H.F. Levison , K. Tsiganis , A. Morbidelli , Nature 435 , 466 ( 2005 ).
 B.C. Johnson and H.J. Melosh , Nature 485 , 75 ( 2012 ); W.F. Bottke , D. Vokrouhlicky , D. Minton et al., Nature 485 , 78 ( 2012 ).
 R. Kallenbach , J. Geiss and W.K. Hartmann , Space Sci Ser ISSI (SSSI) 12 ( 2001 ).
 G. Neukum , B.A. Ivanov , W.K. Hartmann , Space Sci Rev 96 , 55 ( 2001 ), and SSSI 12 , 55 ( 2001 ).