Emergence of a Habitable Planet – University of Texas at …

Loading...

Taking too long?

Reload document

| Open in new tab

Space Sci Rev (2007) 129: 35–78
DOI 10.1007/s11214-007-9225-z

Emergence of a Habitable Planet

Kevin Zahnle
Alex Halliday
Norman H. Sleep

Nick Arndt
Euan Nisbet

Charles Cockell
Franck Selsis

Received: 16 March 2006 / Accepted: 17 January 2007 / Published online: 25 July 2007
© Springer Science+Business Media B.V. 2007

Abstract We address the first several hundred million years of Earth’s history. The Moon-
forming impact left Earth enveloped in a hot silicate atmosphere that cooled and condensed
1,000 yrs. As it cooled the Earth degassed its volatiles into the atmosphere. It took
over
∼
another
2 Myrs for the magma ocean to freeze at the surface. The cooling rate was de-
termined by atmospheric thermal blanketing. Tidal heating by the new Moon was a major
energy source to the magma ocean. After the mantle solidified geothermal heat became
climatologically insignificant, which allowed the steam atmosphere to condense, and left
behind a
500 K CO2 atmosphere. Thereafter cooling was governed by how
quickly CO2 was removed from the atmosphere. If subduction were efficient this could have

100 bar,

∼

∼

∼

K. Zahnle (!)
NASA Ames Research Center, MS 245-3, Moffett Field, CA 94035, USA
e-mail: Kevin.J.Zahnle@nasa.gov

N. Arndt
LGCA, University Joseph Fourier, 1381 rue de la Piscine, 38401 Grenoble, France

C. Cockell
Planetary and Space Sciences Research Institute, Open University, Milton Keynes, MK7 6AA, UK

A. Halliday
Department of Earth Sciences, University of Oxford, Oxford, OX1 3PR, UK

E. Nisbet
Department of Geology, Royal Holloway, University of London, Egham, Surrey, TW20 0EX, UK

F. Selsis
Ecole Normale Supérieure de Lyon, Centre de Recherche Astronomique de Lyon, 46 Allée d’Italie,
69364 Lyon, Cedex 07, France

F. Selsis
CNRS UMR 5574, Université de Lyon 1, Lyon, France

N.H. Sleep
Department of Geophysics, Stanford University, Stanford, CA 94305, USA

(cid:1)
(cid:1)
(cid:1)
(cid:1)
(cid:1)
(cid:1)
36

K. Zahnle et al.

∼

taken as little as 10 million years. In this case the faint young Sun suggests that a lifeless
Earth should have been cold and its oceans white with ice. But if carbonate subduction were
inefficient the CO2 would have mostly stayed in the atmosphere, which would have kept the
surface near
500 K for many tens of millions of years. Hydrous minerals are harder to
subduct than carbonates and there is a good chance that the Hadean mantle was dry. Hadean
heat flow was locally high enough to ensure that any ice cover would have been thin ((cid:2) 5 m)
in places. Moreover hundreds or thousands of asteroid impacts would have been big enough
to melt the ice triggering brief impact summers. We suggest that plate tectonics as it works
now was inadequate to handle typical Hadean heat flows of 0.2–0.5 W/m2. In its place we
hypothesize a convecting mantle capped by a
100 km deep basaltic mush that was rela-
tively permeable to heat flow. Recycling and distillation of hydrous basalts produced granitic
rocks very early, which is consistent with preserved (cid:4) 4 Ga detrital zircons. If carbonates in
oceanic crust subducted as quickly as they formed, Earth could have been habitable as early
as 10–20 Myrs after the Moon-forming impact.

∼

Keywords Hadean Earth
Planetary atmospheres

Moon-forming impact
Late heavy bombardment

Origin of Earth

Magma oceans

1 Introduction

Percival Lowell, the most influential popularizer of planetary science in America before
Sagan, described in lively detail a planetology in which worlds formed hot and dried out as
they aged (Lowell 1895, 1906, 1909). Large worlds cooled slowly, and were still evolution-
arily young in 1895, “while in the moon we gaze upon the last sad age of decrepitude, a
world almost sans air, sans sea, sans life, sans everything.” One reason is that gases escape
to space. “The maximum speed [a molecule] may attain Clerk–Maxwell deduced from the
doctrine of chances to be seven-fold the average. What may happen to one, must eventually
happen to all.” Another reason presumes cooling. “As the [internal] heat dissipates, the body
begins to solidify, starting with the crust. For cosmic purposes it undoubtedly still remains
plastic, but cracks of relatively small size are both formed and persist. Into these the surface
water seeps. With continued refrigeration the crust thickens, more cracks are opened, and
more water given lodgement within, to the impoverishment of the seas.” In many respects
the modern story, if not the prose, broadly resembles Lowell’s.

Lowell’s speculations were rooted in Lord Kelvin’s concepts of time. Kelvin derived the
age of the Earth from the near surface thermal gradient (Kelvin 1895; Schubert et al. 2001;
Wood and Halliday 2005). He made the explicit assumption that the Earth cooled by thermal
conduction and the implicit assumption that the Earth harbored no unknown energy sources.
He obtained an age for the Earth of 25 million years. Kelvin also computed the age of the
Sun, in this case by presuming a convecting body for which gravitational contraction was
the only energy source, and he obtained a similar age. These are the sort of coincidences that
make for a robust theory, or at least a stubborn theorist, and Lowell was one among many
to accept these arguments. In the context of Kelvin’s history of brief time, monotonically
cooling planets made sense: fate was ruled by the surface-to-volume ratio.

The discovery of radioactive heating triggered a relatively brief (and in retrospect ill-
considered) counter-reaction in favor of a cold early Earth, in which the only primary
source of heating was radioactive decay. In this story the slow internal build up of ra-
diogenic heat eventually led to internal melting after hundreds of millions or even bil-
lions of years. A credible consequence of cold formation might be a hydrogen–methane–
ammonia primary atmosphere (Urey 1951). Such an atmosphere would be conducive to

(cid:6)
(cid:6)
(cid:6)
(cid:6)
(cid:6)
Emergence of a Habitable Planet

37

the abiotic synthesis of complex organic molecules (Miller 1953). Cold formation got a
foothold in textbooks, but the enormous gravitational energy released during accretion was
never plausibly made to go away. Hot formation eventually returned to favor when it be-
came more fully appreciated that accretion took the form of giant impacts (Safronov 1972;
Wetherill 1985).

Of more moment to us here is that Lowell placed the origin of life in a Hadean realm
of geothermal heat hidden from the Sun. Perhaps he saw no choice; 25 million years is not
necessarily a lot of time. It is now known that the mantle cools by solid state convection, and
that the Earth is more than 4.5 billion years old. This leaves plenty of time. Yet the suspicion
remains widespread that life arose on Earth in a Hadean realm that is hidden from the rock
record (Cloud 1988; Chyba 1993). The Hadean is important because it set the table for all
that came later (ibid).

1.1 The Hadean Today

Today the Hadean is widely and enduringly pictured as a world of exuberant volcanism,
exploding meteors, huge craters, infernal heat, and billowing sulfurous steams; that is, a
world of fire and brimstone punctuated with blows to the head. In the background a red
Moon looms gigantic in the sky. The popular image has given it a name that celebrates
our mythic roots. As Kelvin and Lowell understood, a hot early Earth is an almost inevitable
consequence of fast planetary growth. The apparent success of the Moon-forming impact hy-
pothesis (Benz et al. 1986; Hartmann et al. 1986; Stevenson 1987; Canup and Righter 2000;
Canup and Asphaug 2001; Canup 2004) has probably evaporated any lingering doubts. Earth
as we know it emerged from a fog of silicate vapor.

1.2 Defining the Hadean

Discord confuses what “Hadean” means or should mean (Nisbet 1985, 1991, 1996). One
choice has been to define the Hadean as the time before the first rock (currently the Acasta
Gneiss, dated to 4.00–4.03 Ga, Bowring and Williams 1999). This puts the Hadean into
the same category as the fastest mile or the tallest building. Another choice is to define it
as the time before the first evidence of life. This definition was in use at one time. Before
Cloud split it into the Hadean and the Archean Eons, there had been a lifeless “Azoic” Eon.
“Archean” means “beginning” in the context of life (Nisbet 1982). This definition is consis-
tent with geological convention but is open to endless debate over what constitutes evidence
of life. Later, Cloud (1983, 1988) set the origin of life in the Hadean. A potentially useful
definition is to synchronize the end of the Hadean with the end of the heavy bombardment
of the inner solar system. This would encourage comparisons between planets. On the other
hand, the end of the late bombardment is not (yet?) well defined as an instant in time, nor has
it shown itself clearly in the terrestrial record. This leaves picking an arbitrary date. Cloud
(1983) used 3.8 Ga, others have used 4.0 Ga. All of these definitions are in effect equal at
present.

The Hadean record is not data rich. Any tale of the Hadean truly told would be so ob-
scured with qualifications, caveats and prevarications that the reader would need a GPS
system just to follow the narrative thread. We have opted instead to present a web of spec-
ulations in flat declarative sentences, constrained by basic physics when possible. This is
the same point of view taken by Stevenson (1983) in an earlier essay on the topic. That our
authoritative-seeming sentences often differ from Stevenson’s authoritative-seeming sen-
tences can be taken as a sign of progress.

38

K. Zahnle et al.

2 Astrophysical Context

2.1 The Interstellar Environment

Stars can form in dense clusters in which massive stars live short, brilliant lives, or they
can form in quiet low-density suburbs where massive stars are rare. Massive stars dominate
their environment. In general massive stars are very hot and extremely luminous and most of
their light is emitted in the UV; such a star can emit 1010 times more UV than does our Sun.
A nearby massive star can be a bigger source of ionizing UV radiation to the solar system
than the Sun itself. Interstellar UV can drive photochemistry, and it can also photoevaporate
the nebular disk from which the Sun and planets formed (Adams et al. 2004). Stellar UV can
also drive off primary atmospheres of small planets. As massive stars hurry toward death
they unleash enormous stellar winds that pollute the nebula with fresh products of stellar
nucleosynthesis. The biggest stars end as supernovae. Supernovae provide the prime source
of short-lived radionuclides such as 26Al and 60Fe. Astronomical observations of γ -rays
from 26Al decay imply that the current average 26Al/27Al ratio in the interstellar medium is
6 (Diehl et al. 2006). This is notably lower than the primordial solar system ratio of
9
×
105 yrs. The implication is that
5(cid:1) 25
10−
the solar nebula was enriched with the products of a recent nearby supernova or supernovae.
Evidently the Sun did not form in a quiet low-density suburb (Adams and Laughlin 2001).
Nearby supernovae could have had other interesting effects on the Sun’s environment. But
massive stars destined for supernova last only a few million years (Arnett 1996). By the time
the Sun reached the main sequence it was well entrenched in its suburban tract home. Any
further speculation on these matters is beyond the scope of this essay.

5 (Bizzarro et al. 2004). The half-life of 26Al is 7

10−

×

×

2.2 The Faint Young Sun

According to the standard model, the Sun has steadily brightened since it arrived on the
Main Sequence 4.52 billion years ago (the Zero-Age Main Sequence, or “ZAMS”). In the
next billion years the Sun brightened from about 71% to 76% of its current luminosity.
Standard solar evolution is shown in Fig. 1.

The faint young Sun imposes a stringent constraint on the climate of the young Earth
(Ringwood 1961; Sagan and Mullen 1972). Without the addition of potent greenhouse gases
the early Earth should have been at most times and places frozen over. This is important and
will be discussed in more detail in the following.

The one way to make the young Sun brighter is to make it more massive than it is now.
The Sun loses mass through the solar wind. At current rates the mass loss is tiny, amounting
to only 0.01% of the Sun’s mass over 4.5 Gyrs. To be as bright as it is now, the ZAMS
Sun would have needed 6% more mass (Sackmann and Boothroyd 2003). This amount of
mass loss far exceeds what is plausible. By studying stellar winds of a half-dozen Sun-like
stars, Wood et al. (2002) found that a Sun-like star loses about 0.5% of its mass after it
reaches the Main Sequence. This is too small to be important. Wood et al. (2005) have since
characterized the winds of another half-dozen solar analogues. According to the newer study
the total mass loss from the main sequence Sun was only

0.1% of its initial mass.

There is little evidence bounding mass loss from very young stars1. In 2002 Wood et al.
argued that the empirical upper limit on X-ray flux implies a parallel upper limit on mass

∼

1When stars are still accreting they generally have extremely large stellar winds, but these typically do not
last more than a few million years at most, and given that the stars are accreting, the winds do not imply that
the star is on net losing mass.

Emergence of a Habitable Planet

39

Fig. 1 The first 3 billion years of solar evolution. The solid curves show luminosity evolution. Main se-
quence luminosity evolution follows Sackmann et al. (1993). Pre-Main Sequence evolution (shaded region)
is adapted from D’Antona and Mazzitelli (1994). The range of uncertainty is determined by mass loss. Pre-
ferred mass loss follows Wood et al. (2002, 2005). Sensitivity to mass loss is scaled from Sackmann and
Boothroyd (2003). The upper bound on luminosity arbitrarily multiplies Wood et al.’s best estimate by a
factor 4(cid:1) 56(cid:10) −
is the age of the Sun in Gyrs. Even with these relatively enormous solar winds
the Sun’s luminosity is barely affected. The solar wind, X-ray, and EUV evolutions (broken curves) follow
Wood et al.’s recommendations and references therein. These latter are aspects of solar activity rather than
solar luminosity—young stars are generally more active than the sedate modern Sun. The observed scatter in
X-ray luminosities of young Sun-like stars (not shown) implies an order of magnitude uncertainty during the
Hadean

1, where (cid:10)

loss rates; in 2005 they showed evidence that stellar winds may be smaller in stars younger
than 0.7 Ga than they become later. This is rather surprising. Still, the data offer no support
for a markedly more massive young Sun. The range of solar evolutions permitted by Wood
et al.’s mass loss rates is shown in Fig. 1.

Often overlooked is that, irrespective of mass loss, the Sun’s luminosity was far from con-
stant in the 50 Myrs it took to contract to the main sequence2. Figure 1 includes a model of
the Sun’s pre-main-sequence evolution beginning at 1 Ma (D’Antona and Mazzitelli 1994).
During the first few million years the Sun was brighter and redder than it is now. At 10
million years it was only half as bright as it is now, while at 30 million years it was almost
precisely as bright as it is now. Thereafter the Sun faded to its ZAMS luminosity as the
nuclear fires took over.

These time scales are comparable to the time scales currently seen as relevant to ter-
restrial planet accretion. Runaway growth of the first generation planets is thought to have
taken no more than 1 Myr (Lissauer 1993; Chambers 2004). Planetary embryos, at first
embedded in the nebula, would have emerged to see a bright red Sun. Earth and Venus
were built by collisions between planetary embryos over some
50 Myrs. As they grew
the planets would have experienced major changes in solar luminosity. These changes
were important because they determine the physical state of water in our part of the
Solar System. As the Sun changed brightness the water condensation front would have

∼

2If this time scale looks familiar, it is: it’s Kelvin’s time scale for gravitational contraction. This is the part of
the Sun’s evolution that predates the onset of significant nuclear fusion.

40

K. Zahnle et al.

swept back and forth through the solar system. For a planet at Earth’s distance from
the Sun, at 2 Myr any water present would have been vapor, while at 10 Myr the wa-
ter would have been ice, and ice would have stable at Venus. It is possible in Mars we
are looking at a planet that is old enough to remember these times (Lunine et al. 2003;
Halliday and Kleine 2006). In any event, the history of volatiles is sensitive to solar lumi-
nosity, and hence the eventual states of Venus and Earth would have been sensitive to the
growth spurts of the young Sun.

2.3 The Active Young Sun

In contrast to luminosity in general, the active young Sun was a much stronger source of
ultraviolet light, X-rays, and solar wind than it is today (see Fig. 1; see also the chapter
by Kulikov et al. 2007, this issue). This inference is based on empirical observations of
hundreds of young solar analogs. The theory is not fully developed, but in broad outline
stellar activity (sunspots, flares, UV, X-rays, etc.) is directly related to the strength of the
magnetic field, which in turn is generated from the star’s rotation. As the star ages it loses
angular momentum through the stellar wind. Solitary stars are like spinning tops. They all
slow down.

3 The Age of the Earth and Solar System

There are no rocks surviving from the first 500 Myrs of Earth’s history. The oldest zircon
grain found thus far yields an age (cid:4) 150 Myrs after the start of the Solar System (Wilde et al.
2001). Therefore, deducing Earth’s earliest history is strongly dependent on geochemistry,
theory and comparison with other solar system objects using meteorites and returned sam-
ples. The Earth was formed through successive accretion events involving objects as large
as other planets. As such the Earth has no simple “age” because it formed from combining
earlier formed planetary objects which already had established their own differentiated reser-
voirs, including cores and atmospheres. We can determine the rate at which the Earth grew
by making certain assumptions about the degree of mixing and equilibration between these
planets as they coalesced. We can also define the start of the Solar System and this growth
history very precisely. Chondrites are the most common form of meteorite landing on Earth.
They are thought to represent early dust and debris from the circumstellar disk from which
the planets grew. Most chondrites contain refractory Ca–Al-rich inclusions (CAIs) enriched
in elements expected to condense at very high temperatures from a hot nebular gas. These
are the oldest objects yet identified that formed in the Solar System. CAIs from the Efre-
movka chondrite have been dated by 235(cid:3) 238U–207(cid:3) 206Pb at 4(cid:1) 5672
0(cid:1) 0006 Ga (Amelin et
al. 2002). This is the current best estimate of the start to the Solar System and hence defines
a more precise slope to the meteorite isochron (called the “Geochron”) first established by
Patterson (1956) (Fig. 2). To sort out the growth history of planets it is necessary to use
short-lived nuclides, dynamic simulations of planet formation and petrological constraints
on likely core formation scenarios.

Short-lived nuclides provide a set of powerful tools for unraveling a precise chronology
of the early solar system. The advantage of these is that the changes in daughter isotope can
only take place over a restricted early time window; there is no correction for the effects of
decay over the past 4.5 billion years. A disadvantage is that the parent isotope can no longer
be measured. Hence its abundance at the start of the solar system must first be determined
by comparing the isotopic composition of the daughter element in rocks and minerals of
independently known age. Only then can it provide useful age constraints.

(cid:8)
Emergence of a Habitable Planet

41

Fig. 2 The current best estimates for the time-scales over which very early inner solar system objects and
the terrestrial planets formed. The approximated mean life of accretion (τ ) is the time taken to achieve 63%
growth at exponentially decreasing rates of growth. The dashed lines indicate the mean life for accretion
deduced for the Earth based on W and Pb isotopes (Halliday 2003, 2004; Kleine et al. 2002; Yin et al. 2002).
The earliest age of the Moon assumes separation from a reservoir with chondritic Hf/W (Kleine et al. 2002;
Yin et al. 2002). The best estimates are based on the radiogenic ingrowth deduced for the interior of the Moon
(Halliday 2003, 2004; Kleine et al. 2005b). See text for details of other sources. Based on a figure in Halliday
and Kleine (2006)

The short-lived nuclides provide most of the information on the first 50 Myrs of the
solar system. For example, as well as CAIs, most chondrites also contain chondrules, drop-
shaped ultramafic objects with strange textures thought to reflect rapid heating, melting and
quenching of pre-existing material in a dusty disk. Using 26Al–26Mg it has been shown
that some of these chondrules formed as much as 1 to 3 million years after the start of
the Solar System (Russell et al. 1996; Bizzarro et al. 2004) (Fig. 2). Therefore chondrites,
the meteorites that contain chondrules, though primitive in composition, must have formed
millions of years after the start of the solar system. This is interesting because simula-
tions of planetary accretion indicate that dust should have accumulated into 1000 km-sized
planetary embryos in just a few hundred thousand years—much less than the time indi-
cated from chondrule formation. In fact we now have excellent isotopic evidence that a
range of accretion styles were involved in the formation of the terrestrial planets. Before
discussing this it is worth first explaining the theories behind the formation of Earth-like
planets.

3.1 Planetary Accretion

A variety of theories have been advanced for how terrestrial planets form. For a recent
review see Chambers (2004). In broad terms the rates of accretion of Earth-like planets will
be affected by the amount of mass in the disk itself. If there is nebular gas present at the
time of accretion the rates are faster. In fact the absence of nebular gas is also calculated to
favor eccentric orbits, which gas would dampen (Agnor and Ward 2002). The presence of
solar noble gases in the Earth and Mars is consistent with these requirements. In the simplest
terms accretion of terrestrial planets is envisaged as taking place in four stages:

42

K. Zahnle et al.

(1) Settling of circumstellar dust to the mid-plane of the disk.
(2) Growth of planetesimals up to
1 km in size.
(3) Runaway growth of planetary embryos up to
(4) Oligarchic growth of larger objects through late-stage collisions.

103 km in size.

∼

∼

Stage 1 takes place over time scales of thousands of years and provides a relatively dense
plane of material from which the planets can grow. The second stage is the most poorly
understood at present but is necessary in order to build objects that are of sufficient mass
for gravity to play a major role. Planetesimals would need to be about a kilometer in size in
order for the gravitationally driven stage 3 to start.

We do not know how stage 2 happens, although clearly it must. Scientists have succeeded
in making fluffy aggregates from dust, but these are all less than a cm in size. How does one
make something that is the size of a house or a stadium? One obvious suggestion is that
some kind of glue was involved. Volatiles would not condense in the inner solar system. Not
only were the pressures too low, but the temperatures were probably high because of heating
as material was swept into the Sun (Boss 1990). An alternative is that, within a disk of dust
and gas, collective effects can sort or gather particles into pockets of locally high density that
might promote collisional coagulation or gravitational collapse (Weidenschilling and Cuzzi
1993; Cuzzi et al. 2005). Local separation and clumping of the material might also lead to
larger scale gravitational instabilities, whereby an entire section of the disk has relatively
high gravity and accumulates into a zone of concentrated mass (Ward 2000).

However they are formed, runaway growth builds these 1 km-sized objects into 1,000 km-
sized objects. The bigger the object the larger it becomes until all of the material available
within a given feeding zone or heliocentric distance is incorporated into planetary embryos.
This is thought to take place within a few hundred thousand years (Kortenkamp et al. 2000).
The ultimate size depends on the amount of material available. Using models for the density
of the solar nebula it is possible that Mars-sized objects could originate in this fashion.

Building objects that are the size of the Earth is thought to require a more protracted
history of collisions between such planetary embryos. Wetherill (1986) ran Monte Carlo
simulations of terrestrial planetary growth and some runs with planets of the right size and
distribution to be matches for Mercury, Venus, Earth and Mars. He monitored the time scales
involved in these “successful” runs and found that most of the mass was accreted in the first
10 Myrs, but that significant accretion continued for much longer. Wetherill also tracked
the provenance of material that built the terrestrial planets and showed that, in contrast to
runaway growth, the feeding zone concept was flawed. The planetesimals and planetary
embryos that built the Earth came from distances that extended over more than 2 AU. More
recent calculations of solar system formation have yielded similar results (Canup and Agnor
2000; Raymond et al. 2004).

Such planetary collisions would have been catastrophic. The energy released is sufficient
to raise the temperature of the Earth by thousands of degrees. The most widely held theory
for the formation of the Moon is that there was such a catastrophic collision between a
Mars-sized planet and the proto-Earth when it was approximately 90% of its current mass.
The putative impactor planet, sometimes named “Theia” (the mother of Selene who was the
goddess of the Moon), struck the proto-Earth with a glancing blow generating the angular
momentum of the Earth–Moon system.

3.2 Tungsten Isotopic Tests for Earth Formation Models

The above models of planet formation differ with respect to timing and can therefore be eval-
uated using isotope geochemistry. The 182Hf–182W chronometer has been particularly useful