Download Compositions of Mercury`s earliest crust from magma ocean models

Earth and Planetary Science Letters 286 (2009) 446–455
Contents lists available at ScienceDirect
Earth and Planetary Science Letters
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e p s l
Compositions of Mercury's earliest crust from magma ocean models
Stephanie M. Brown a,⁎, Linda T. Elkins-Tanton b,1
a
b
Massachusetts Institute of Technology, Dept. Earth, Atmospheric, and Planetary Sciences, Building 54-823, 77 Massachusetts Avenue, Cambridge, MA 02139, United States
Massachusetts Institute of Technology, Dept. Earth, Atmospheric, and Planetary Sciences, Building 54-824, 77 Massachusetts Avenue, Cambridge, MA 02139, United States
a r t i c l e
i n f o
Article history:
Received 21 January 2009
Received in revised form 22 June 2009
Accepted 4 July 2009
Available online 4 August 2009
Editor: R.W.Carlson
Keywords:
Mercury
magma ocean
crustal composition
giant impact
core
a b s t r a c t
The size of the Mercurian core and the low ferrous iron bearing silicate content of its crust offer constraints
on formation models for the planet. Here we consider a bulk composition that allows endogenous formation
of the planet's large core, and by processing the mantle through a magma ocean, would produce a low-iron
oxide crust consistent with observations. More Earth-like bulk compositions require silicate removal, perhaps
by a giant impact, to create the planet's large core fraction. We find that the endogenous model can produce a
large core with either a plagioclase flotation crust or a low-iron oxide magmatic crust. Because a magma
ocean creates a gradient in iron oxide content in the resulting planetary mantle, the parts of the mantle
removed by a putative giant impact could result in either a high-iron oxide mantle in contradiction to current
crustal measurements, or a low-iron oxide mantle consistent with the current understanding of Mercury. If a
giant impact cannot preferentially remove shallow mantle material then the proto-Mercury must have had a
bulk low iron-oxide composition. Thus a specific bulk composition is required to make Mercury
endogenously, and either a specific process or a specific composition is required to make it exogenously
through giant impact. Measurements taken by the MESSENGER mission, when compared to predictions given
here, may help resolve Mercury's formation process.
© 2009 Elsevier B.V. All rights reserved.
1. Introduction
We model magma ocean solidification of several hypothetical bulk
mantle compositions of Mercury and predict the compositions of their
corresponding first crust, part of which may remain today. It may be
possible to compare our predicted crusts with measured compositions
from the MESSENGER mission and thus further constrain Mercury
formation mechanisms.
1.1. Surface and interior composition constrains
Data collected from Earth-based observation and by the three
flybys of Mariner 10 show that Mercury's surface is low in iron-bearing
silicates, containing between 0 and 3–4 wt.% FeO in the silicates
(Solomon, 2003; Denevi and Robinson, 2008). Mid-infrared spectra
have detected the presence of calcic plagioclase feldspar and low-FeO
pyroxene (Robinson and Lucey, 1997; Robinson and Taylor, 2001;
Solomon, 2003; Sprague et al., 2009). Recent measurements by the
flybys of Mercury MESSENGER are consistent with these ground-based
observations and limit total iron content to be less than 6 wt.%
(Solomon et al., 2008), with iron oxide in silicates at less than 2–3 wt.%
⁎ Corresponding author. Tel.: +1 617 253 1902.
E-mail addresses: [email protected] (S.M. Brown), [email protected]
(L.T. Elkins-Tanton).
1
Tel.: + 1 617 253 1902.
0012-821X/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsl.2009.07.010
(Robinson et al., 2008; McClintock et al., 2008). Recent observations
indicate the presence of a possible global darkening agent, thought to
be an opaque oxide, a non-silicate mineral (Denevi and Robinson,
2008; Robinson et al., 2008; McClintock et al., 2008).
While Mercury's surface silicates are iron oxide-poor, the bulk
planet contains a high metallic iron fraction. Mercury's high density
(Newcomb, 1895; Rabe, 1950; Anderson 1975; Anderson et al., 1987)
indicates a massive core (Urey, 1952; Anderson and Kovach, 1967;
Reynolds and Summers, 1969; Siegfried and Solomon, 1974). Uncertainties in the core radius and mass have been explored in response to
the recent finding that the outer core is molten (Solomon, 2007;
Margot et al., 2007), indicating the presence of a possible light alloying
element (Harder and Schubert, 2001; Hauck et al., 2007; Riner et al.,
2008). Core radius values range from 1675 km to 2075 km with a
corresponding 53–78% of the mass of the planet (Riner et al., 2008).
1.2. Mercury formation mechanisms
During the late 1970s and the 1980s, after Mariner 10, several
hypotheses for the origin of Mercury were motivated by Mercury's
high core fraction when compared to the other terrestrial planets.
These different hypotheses are based upon either planetary accretion
from compositions inherently high in metallic iron and low in silicates
(endogenous) or silicate removal from an initially larger Mercury
(exogenous). We consider three hypotheses as the most plausible and
distinct in recent literature (Solomon, 2003) and discuss them here.
(See Taylor and Scott, 2003 for details on the hypotheses.)
S.M. Brown, L.T. Elkins-Tanton / Earth and Planetary Science Letters 286 (2009) 446–455
In 1978, Weidenschilling proposed that aerodynamic drag in the
early solar system had a different effect on silicates than on the iron
that endogenously formed Mercury. Weidenschilling's aerodynamic
sorting hypothesis, based upon a nebular condensation model,
culminates with a crust produced by a crystal–liquid fractionation of
a silicate magma ocean (Weidenschilling, 1978; Solomon, 2003).
The remaining two hypotheses modify Mercury after planetary
accretion by removing silicates from the differentiated, cooled mantle.
The first relies on a giant impact to remove the silicates from a
Mercury 2.25 times the mass it is today (Benz et al., 1988, 2007). The
giant impact is the favored hypothesis among many researchers (Boss,
1998; Taylor and Scott, 2003) for the formation of Mercury in part due
to its similarities with the current theory of the formation of the
Moon.
The giant impact hypothesis relies upon accretion across a wide
radius of the protoplanetary disk in order to account for the high
relative velocity of the impactor. However, the other terrestrial planets
all have higher iron oxide contents in their crusts than does Mercury;
if the impactor came from a wide feeding range, crustal composition
would be expected to be more similar (Taylor and Scott, 2003).
In the final model discussed here, 70 to 80% of the silicate mantle is
vaporized by radiation from the hot young Sun and is then removed by
a strong solar wind (Cameron, 1985a; Fegley and Cameron, 1987).
Temperatures at the distance of Mercury would have to reach between
2500 K and 3500 K to vaporize the mantle. Cameron proposes a model
that forms the cores of the terrestrial planets before the Sun begins to
form, placing Mercury's core between the necessary high temperatures (1982; Decampli and Cameron, 1979, Cameron, 1985a,b, 1988;
Cameron et al., 1987). This Solar System formation model is not
consistent with more modern models reviewed and discussed by Boss
(1998, 2003) and Chambers (2003, 2004). Cameron himself realized
the unlikely conditions necessary to form a Mercury depleted in
silicates with this method and posits that the giant impact hypothesis
as more probable (Cameron, 1988).
1.3. Structure of models
With a view toward the likely formation processes of the other
terrestrial planets, we propose that Mercury's large core and iron
oxide-poor surface may be consistent with planetary formation from a
bulk composition richer in metallic iron than that inferred for the
Earth or Mars, rather than by solar ablation (Cameron, 1985a; Fegley
and Cameron, 1987) or giant impact (Benz et al., 1988, 2007). The
hypothesis presented here is parallel to Weidenschilling's (1978)
sorting theory in that we propose a primary bulk composition of
Mercury produced its large core, but rather than sorting alone, we
suggest that high temperatures and highly reducing conditions in the
early Solar System helped produce a Mercurian bulk composition that
would eventually accrete to form a large iron core with low-iron oxide
silicate mantle.
Reducing conditions are expected at high temperatures, when
metallic iron and sulfur will condense. At distances close to the young
Sun, material had the opportunity to become reduced because
temperatures were around 2500 K to 3500 K as Cameron (1985a,b)
notes. This material would have an unusual metallic iron fraction by
reducing iron otherwise bound into silicates. However, the Solar
System has been cooling since it once reached these high temperatures. Safronov (1972) places disk temperatures between 115 K and
186 K in the zone of Mercury based on solar radiation calculations. We
consider the situation in which Mercury was later accreted, at lower
temperatures, from material reduced at high temperatures. The
creation of Mercury's high metal fraction by accretion of metal-rich
chondrites material has previously been proposed and briefly
investigated by Taylor and Scott (2003).
Mercury likely formed from materials condensed at a range of
major axes from the Sun, but we suggest that high temperatures
447
around Mercury's orbit may have reduced their iron, regardless of
their origin. Equilibrium condensation (discussed by Latimer, 1950;
Lewis, 1972; Urey, 1952) predicts volatility and iron oxide content to
increase with increasing helicocentric distances. We are effectively
reconsidering the equilibrium condensation hypothesis within the
framework of radial mixing, which is the major prediction from
dynamical models of planetary accretion. We are not, however,
presenting a specific new nebular model here.
We present the results of models for the formation of Mercury that
use known chondritic meteorite compositions to match the large core
fraction of the planet and follow self-consistent models of magma
ocean solidification to produce predictions for Mercury's crustal
composition. Our hypothesis accounts for the large metallic iron
fraction of Mercury and predicts a bulk mantle and crust composition
low in iron oxide and volatiles and with refractory elemental
abundances similar to chondrites (none enriched). The results of
these models may be compared to the existing observations, and await
further data from the MESSENGER mission.
2. Models
The observations of Mercury's surface composition and its core
size offer an initial constraint on early planet- and crust-forming
processes. We assume that a magma ocean existed after the accretion
of the planet. We model the solidification of the magma ocean and
investigate the consequences of differing initial bulk compositions.
Significant changes in bulk compositions result in alterations in final
mineralogy and structure of the planet as well as initial crustal
compositions.
2.1. Compositions
The Mercury formation hypothesis presented here is motivated in
part by the discovery of chondrites with an unusually high metallic
iron composition: Bencubbinite (CB) chondrites. This rare primitive
material contains at least 60 wt.% metallic iron (Weisberg et al., 2000).
Should Mercury have been built from a CB-like bulk composition, its
metallic iron fraction may be entirely explained without further iron
reduction. Taylor and Scott (2003) propose that the accretion of
metal-rich CB chondritic planetesimals is a plausible alternative to
previous hypotheses for the formation of Mercury. Other chondrite
classes could not create Mercury even if all their constituent iron was
metallic (Fig. 1). Because Mercury orbits near the Sun, however,
ejection of its bulk planet-building material (possibly the CB material
composition) out to Earth's orbit is far less likely than is our obtaining
samples of bulk material consistent with building the Earth and Mars,
and we would therefore have fewer examples of its material
(Wetherill, 1984; Love and Keil, 1995; Gladman et al., 1996).
Current models of planetary accretion indicate that planets'
feeding zones extend over a range of distances from the Sun
(Raymond et al., 2006; O'Brien et al., 2006). Formation from a single
meteorite composition is therefore highly unlikely. The CB chondrites
offer an example composition with sufficient iron to build Mercury's
large core and possibly a resulting mantle with interestingly high
silica, but we do not argue that any planet is formed from any one
primitive composition.
As the constraints on Mercury's composition are few, we consider a
range of initial bulk compositions. In this study we use four potential
mantle compositions for Mercury. The first is the silica-rich composition that results from removing all iron oxide (reduced into a putative
core) from the average CB chondrite composition (Weisberg et al.,
2000; Lauretta et al., 2007). As the amount of iron oxide in this
composition relies upon the extent of the reducing conditions, we
consider a varying amount of iron oxide in our starting bulk magma
ocean compositions, while holding the remaining oxides constant.
This first CB chondrite composition is created by averaging the
448
S.M. Brown, L.T. Elkins-Tanton / Earth and Planetary Science Letters 286 (2009) 446–455
Fig. 1. Natural bulk metallic iron content for selected meteorite classes, and their metallic iron content if all iron oxide was also reduced to metallic iron. The metallic iron is presented
in combination with NiS as we assume in a simplistic way that the bulk of the nickel and sulfur enter the core. In these compositions, NiS ranges from 0 to 2.28 mass percent of the
planet. Only the CB chondrites have sufficient metallic iron to create Mercury's core through endogenous processes. Data: Lodders and Fegley (1998): CI, CO, CK, CR, EH, and EL;
Jarosewich (1990): CM, CV, L, LL, and eucrites; Kenkyujo et al. (1995): H; CB compositions averaged from Lauretta et al. (2007) and Weisberg et al. (2001).
compositions given in Lauretta et al. (2007) and Weisberg et al.
(2000), and results in a composition with superchondritic Si/Mg.
Because this may reflect sampling bias, we also use a CB chondrite
composition created by averaging the Weisberg et al. (2000) and
Lauretta et al. (2007) compositions that have chondritic Si/Mg. These
two compositions are referred to as “chondritic CB” and “nonchondritic CB,” respectively.
The third is a modeled average Earth mantle from Hart and Zindler
(1986) (Table 1). The Hart and Zindler (1986) composition is derived
from measured compositions of Earth mantle lherzolites (undepleted
fertile mantle rocks). Morgan and Anders (1980) provides the fourth
composition, produced from an assumed fractionation in the nebula of
solar elemental abundances from an equilibrium condensation model;
certain values were filled in by using lunar abundances and inferences
for Mercury. From equilibrium condensation and this specific
fractionation, all metal percentages increase nearer to the Sun in the
Morgan and Anders (1980) model, and therefore Cr2O3 is notably
higher than in the other compositions. The resulting mantle mineral
assemblages differ from other bulk compositions and thus produce
different predictions for the mantle and early crustal compositions of
the planet. The Hart and Zindler (1986) composition and the Taylor
and Scott (2003) CB composition fall between the CB chondritic and
the CB non-chondritic composition. Therefore we did not run another
Table 1
Bulk mantle compositions used in models.
Oxide
in wt.%
Morgan
and Anders
(1980)
Chondritic CB
Weisberg et al.
(2000, 2001),
Lauretta et al.
(2007)
Hart and
Zindler (1986)
Earth model
mantle
Taylor
and Scott
(2003) CB
Nonchondritic CB
Weisberg et al.
(2000, 2001),
Lauretta et al.
(2007)
SiO2
Al2O3
MgO
CaO
TiO2
Cr2O3
FeO
Mg #
Si/Mg
47.1
6.4
33.7
5.2
0.3
3.3
3.7
94.20
1.1
47
5.5
42
4
0.2
1
0–7
91–100
1.1
46.3
3.96
38.4
3.2
0.2
0.5
7.7
89.89
0.9
50.8
5.2
37.4
3.6
0.2
0.7
2
97.09
1.1
63
3.7
29
3
0.15
1
0–7
87–100
2.2
model with the CB composition of Taylor and Scott (2003). These
initial bulk compositions are given in Table 1.
2.2. Magma ocean solidification processes
We assume that the energy of accretion and core formation is
converted into enough heat to melt the planet, producing a wholemantle magma ocean with an initial depth of 600 km, from a radius of
1840 km at the core–mantle boundary to 2440 km at the surface.
Although magma oceans are not a certainty of planetary formation,
they are a likelihood, and here we start with the simplest assumption,
a whole-mantle magma ocean.
Magma ocean solidification proceeds through two major phases.
First, because the slope of the adiabat in the well-mixed magma ocean
is steeper than the slope of the solidus of the bulk magma ocean, as
cooling proceeds the adiabat first intersects the solidus at depth. Thus
the solidification of the magma ocean proceeds from the core–mantle
boundary to the surface (Solomatov, 2000; Elkins-Tanton et al., 2003).
The majority of mantle cumulate minerals possess a crystal site
that incorporates either magnesium or iron. Because magnesium is
incorporated preferentially, evolving magma ocean liquids, and
therefore later-solidifying minerals, are enriched in iron. This process
can produce solid mantle cumulates with densities that increase with
radius through iron enrichment, and this iron enrichment process is
critical for flotation of buoyant minerals in the magma ocean. Density
gradients in the solid cumulate can also be produced by mineral phase
changes.
In the second stage of magma ocean solidification, the gravitationally unstable solidified silicate mantle cumulates overturn to a stable
configuration as discussed in Elkins-Tanton et al. (2003, 2005a,b) and
Elkins-Tanton (2008). During this process, low-density cumulates that
originated at depth rise buoyantly and regions will melt adiabatically
as they near the surface, potentially producing an earliest igneous
crust. Simultaneously, higher iron oxide cumulates sink into the
planetary interior. Overturn creates a mantle that is gravitationally
stable and therefore resistant to the onset of thermal convection.
If overturn does not allow rising cumulates to melt, and no
flotation crust has formed, then the young planet will be left with a
magmatic crust produced by solidification of the magma ocean. This
crust would consist of the solidified last fractions of magma ocean
S.M. Brown, L.T. Elkins-Tanton / Earth and Planetary Science Letters 286 (2009) 446–455
evolved liquids, and it would not resemble a basaltic magma produced
by melting conventional mantle materials.
Mantle overturn is driven by the density instabilities that are
produced during mantle solidification. The density instabilities are
commonly produced by the incorporation of magnesium over iron
into the structures of the minerals. The final, near-surface layers in
these models contain significant fractions of titanium-, chromium- or
iron-bearing oxide minerals that will also drive overturn, as has been
suggested for the Moon (Hess and Parmentier, 1995). In the case of
low iron oxide in the original bulk composition, overturn is driven by
this opaque-bearing layer. However, its viscosity may be sufficiently
high to prevent wholesale overturn and opaque-bearing cumulates
may partially remain (Elkins-Tanton et al., 2002; Riner et al., 2009), a
possible outcome not modeled here, but discussed in Section 4.3. Note
in addition that any unstable density gradient is sufficient to drive
overturn, as long as viscosities are low enough to allow flow.
2.3. Modeling solidification
To model solidification we use existing magma ocean evolution
code from Elkins-Tanton et al. (2003, 2005a,b) and Elkins-Tanton
(2008), modified for Mercury. This code requires an a priori set of
assumptions about the mineral assemblages being solidified at every
step, and these assemblages are specific for each of our bulk
compositions. To predict crystallizing assemblages for Mercury we
used John Longhi's magma ocean code FXMO, which is a Gibbs free
energy minimization program for magma ocean solidification, along
with existing experimental data. The magma ocean evolution code
then calculates the equilibrium compositions and densities of each
fractionating increment of mantle cumulates, and tracks the composition of the evolving residual magma ocean liquids. These mineral
assemblages are reasonably simplifications of what can only be
determined through series of experiments on bulk and then evolving
magma ocean compositions. Results of these models, however, are not
highly sensitive to small changes in mineralogy.
In the non-chondritic CB bulk composition, high silica produces
mantle cumulate phases including quartz (Fig. 2). The Earth mantle
and other compositions contain far less silica and thus quartz is not
stable in its magma ocean solidification evolution. All models end
with evolved liquids rich in titanium and chromium, and thus solidify
a layer near the planetary surface that is rich in opaque minerals.
The solidus that we assume for all models is similar to that of a
terrestrial peridotite over the pressure range appropriate for Mercury's magma ocean. For the soilidus equation and other details of
model calculations, see Elkins-Tanton (2008). In these models the
solidus determines the temperatures of the newly solidified mantle
cumulates. Their temperatures in turn contribute to calculations of
cumulate density through the thermal expansivities of the minerals.
Solidi for a range of compositions do not vary enough to significantly
change the predictions of these models.
Mineral densities are calculated based on their temperature,
pressure, and composition (thermochemical parameters are given in
Elkins-Tanton (2008)). The model tracks the evolving magma ocean
liquid and atmospheric mass and composition. One percent interstitial
liquid is retained throughout solidification in each model. All solids
are assumed to retain their solidus temperatures during the entirety of
magma ocean solidification because the speed of solidification is fast
compared to thermal diffusion in the solid.
3. Results
There are two major opportunities for the formation of an early
crust through magma ocean processes. In the first, as hypothesized for
the Moon, plagioclase or other buoyant minerals float in a denser
coexisting magma ocean liquid and form a conductive lid. Iron oxide
enrichment creating progressively denser evolving magma ocean
449
liquids is the major contributor to plagioclase flotation (Wood et al.,
1970; Smith et al., 1970; Warren, 1985). The second process that may
produce an initial crust is cumulate overturn in the solid state
following magma ocean solidification, in which buoyant cumulates
rise from depth and melt adiabatically, producing secondary magmas
that erupt onto the planet's surface.
3.1. Flotation crusts
The planetary body most associated with a magma ocean is the
Moon, thought to have been produced from a body the size of Mars
impacting the Earth (Hartmann and Davis, 1975; Cameron and Ward,
1976). During solidification of the lunar magma ocean, plagioclase
became buoyant and floated to the surface to form the initial crust.
This plagioclase flotation has been suggested as a definitive marker of
a magma ocean, but flotation will only occur if the plagioclase is less
dense than its coexisting magma and if plagioclase stability is reached
before a significant crystal fraction forms (the network of crystals
would then prevent flotation).
For the N7 wt.% FeO Earth-like (Hart and Zindler, 1986) mantle
composition, plagioclase is stable and sufficiently buoyant to float in
the magma ocean liquids from the point it becomes a crystallizing
phase (Fig. 3e), assuming there is no crystal network to impede
flotation. The moderate iron and high chromium content of the
Morgan and Anders model composition similarly always produces a
liquid dense enough to float plagioclase (Fig. 3f). In contrast,
depending upon their iron contents, the iron oxide-poor magma
ocean liquids in the chondritic and non-chondritic CB composition
models do not always reach densities high enough to float plagioclase
(Fig. 3c, d).
To test the sensitivity of plagioclase flotation to iron oxide content
in the initial magma ocean bulk composition for the four compositions, we started at 0 wt.% FeO and incrementally added iron oxide
into the bulk composition (Figs. 3 and 4), and tested for plagioclase
buoyancy. Noting that our model results are dependent on bulk
composition and temperature, and limited by current understandings
of melt densities, we posit that buoyant minerals will not float in the
magma ocean unless its initial bulk FeO content is at least 2 wt.%–7 wt.%
(Figs. 3 and 4), depending upon the rest of the bulk composition. With
these higher initial iron oxide contents, in the case of the non-chondritic
CB composition, both plagioclase and quartz float in the Mercurian
magma ocean, creating a novel iron oxide-poor, silica-rich earliest crust.
The chondritic CB composition produces a simple plagioclase flotation
crust.
3.2. Volcanic crusts
In a second mechanism for creating an early crust, adiabatic
melting in upwelling solids during compositionally-driven overturn
can produce a melt that erupts onto the planetary surface. Even in an
initial magma ocean with insufficient iron oxide to float plagioclase,
sufficient cumulate density gradients are produced by titanium or
chromium enrichment to cause gravitationally driven solid-state
overturn following solidification. The solidification model produces
predictions for the original depth and composition of cumulates that,
during gravitationally driven overturn, rise sufficiently to melt (Fig. 5
shows results for the chondritic CB composition). Their exact melt
compositions cannot be calculated precisely because specific experimental studies would be required, but approximations can be made
based on the source bulk composition using the free energy
minimization program pMELTS (Ghiorso et al., 2002).
When the bulk magma ocean FeO content is below the critical 2 to
7 wt.% value, the earliest planetary crust is likely formed by erupting
lava that originates from melt during overturn. In all models one or
two regions that form at depth during solidification are sufficiently
buoyant and hot that they may melt adiabatically during overturn and
450
S.M. Brown, L.T. Elkins-Tanton / Earth and Planetary Science Letters 286 (2009) 446–455
contribute to this posited earliest igneous crust. Calculating the exact
melt composition is unrealistic first because of the simplicity of the
models and second because the melting behavior of these compositions is unknown, but rough estimates may be informative when
compared to surface spectral data.
In the chondritic CB composition model, low-FeO olivine +
orthopyroxene cumulates that originated below 300 km depth rise
sufficiently that they would partially melt during gravitationallydriven overturn (Fig. 6). Melts from this region would produce a very
low iron basaltic composition (Table 2).
The non-chondritic CB-like bulk composition with sufficiently low
iron oxide to avoid producing a flotation crust may form a second
interesting high-silica volcanic crustal composition (Table 2). In this
case, the mineralogy of the surface can be predicted to some degree by
calculating the mineral modes of the melt, solidified and equilibrated
at the surface (Table 2). The deep melting region produces an albite–
olivine–clinopyroxene rich crustal rock, perhaps a troctolite. The
shallow melting region produces a lava composed mostly of quartz
and some smaller amounts of clinopyroxene, anorthite, and spinel.
Each region contributes to about half of the melt volume. Both of these
Fig. 2. Cross-sections of the model Mercuries after crystallization and prior to overturn. The magma ocean solidifies from the core to the surface with the mineral assemblages shown.
S.M. Brown, L.T. Elkins-Tanton / Earth and Planetary Science Letters 286 (2009) 446–455
volcanic crusts, produced during magma ocean cumulate overturn,
would have very low iron oxide content in their silicate minerals.
In summary, by modeling solidification of a Mercury magma ocean,
it is possible to predict the composition of the planet's initial crust,
which may remain today. Using a CB chondrite composition successfully reproduces Mercury's core fraction and creates a mantle very low
in iron oxide. Unless FeO exceeds 2 to 7 wt.% in the bulk mantle,
plagioclase will not float and form an earliest crust. The initial crust on
such a planet will consist of high-magnesia lava erupted from melting
produced during solid-state mantle overturn. With higher initial iron
451
oxide fractions, the non-chondritic CB chondrite composition will
create an initial flotation crust of quartz and plagioclase, but dominated
by quartz, in contrast to the Moon. Mantle compositions more similar
to the Earth will likely produce a plagioclase flotation crust. These
scenarios create distinct and measurable predictions.
4. Discussion
The strongest existing constraints on Mercurian formation models
are its very low iron bearing crustal silicates (Robinson and Lucey,
Fig. 3. Density of minerals crystallizing from the magma ocean as a function of depth. The dotted line shows the density of the evolving coexisting magma ocean liquid. In some cases
plagioclase or plagioclase and quartz become less dense than the liquid and can float. Flotation is only suppressed in compositions with less than ~ 2–7 wt.% iron.
452
S.M. Brown, L.T. Elkins-Tanton / Earth and Planetary Science Letters 286 (2009) 446–455
Fig. 4. The sensitivity of the non-chondritic CB bulk composition to FeO. The lines show the density of evolving magma ocean liquids as a function of radius and iron content. Initial
iron content is marked on each curve; liquid density changes primarily because iron is enriched in liquids as low-pressure minerals are fractionated from it (see Fig. 3). The
plagioclase and the quartz lines are the densities of these minerals as a function of pressure in the planet. For this specific composition, ~ 1.5 wt.% FeO is necessary to float the shown
minerals, the lowest of all the compositions. Across all bulk compositions, 2–7 wt.% FeO is necessary to float buoyant minerals. Near the end of solidification high crystal fraction may
prevent flotation, though the physics of this process are poorly understood, so the depth of buoyancy required to allow flotation is unconstrained.
1997; Robinson and Taylor, 2001; Solomon, 2003; Denevi and
Robinson, 2008; Sprague et al., 2009) and the large metallic iron
core (Urey, 1952; Anderson and Kovach, 1967; Reynolds and Summers,
1969; Siegfried and Solomon, 1974). In our magma ocean models,
there are two opportunities to produce an earliest crust that may
remain on the planet today and can be compared to observed
constraints. The first is through flotation of buoyant minerals during
magma ocean solidification, and the second is through eruptions of
magma produced by adiabatic melting during solid-state cumulate
mantle overturn after magma ocean solidification.
4.1. Endogenous models for producing a low-iron oxide silicate crust
There are similarly two processes that may create a low-iron oxide
crust: plagioclase may float in a magma ocean, or mantle source
regions themselves low in iron oxide may melt and erupt a low-iron
oxide crust. Previous researchers have estimated that the amount of
iron oxide in the lavas on the surface of Mercury directly reflects the
amount of iron oxide within the melting mantle source regions of the
Fig. 5. Mercurian cumulate mantle density profiles of chondritic CB bulk composition
with 0.2 wt.% FeO after magma ocean solidification (dashed line), and following
gravitational overturn to stability (solid line). The mineral labels correspond to the
dashed, pre-overturn density profile.
planet, because iron oxide has a partial melt partition coefficient ~ 1
(Robinson and Taylor, 2001).
On a planet the size of Mercury any of the bulk mantle
compositions considered create a plagioclase or plagioclase + quartz
flotation crust, provided the bulk composition has at least 2–7 wt.%
FeO. Thus a chondritic CB Mercury with N2 to 7 wt.% FeO in its mantle,
or a Mercury with a Hart and Zindler (1986) Earth-like mantle, or a
Morgan and Anders (1980) model mantle might all present a low iron
oxide ancient crust through flotation of buoyant phases (Table 2).
The lavas produced by adiabatic melting during solid-state overturn
would not likely erupt through a plagioclase flotation crust in the absence
of crustal thinning, for example, from giant impacts. Should a flotation
crust not form, as in the case of a low-FeO silicate mantle, then the lavas
may erupt to form the earliest crust. A low-iron CB chondritic Mercurian
mantle would likely produce a low-iron oxide earliest igneous crust and
would be entirely consistent with the existing data on Mercury's surface.
Taylor and Scott (2003) produce crustal composition predictions
from a series of possible Mercury bulk compositions, including those
of Morgan and Anders (1980), the vaporization model by Fegley and
Cameron (1987), Goettel (1988), and a CB-chondrite like composition.
Fig. 6. Bulk mantle composition for chondritic CB starting composition after overturn to
gravitational stability. The shaded regions are hot enough to experience adiabatic
decompression melting as they rise during overturn and would produce a low-iron
oxide silicate crust.
S.M. Brown, L.T. Elkins-Tanton / Earth and Planetary Science Letters 286 (2009) 446–455
453
Table 2
Earliest crustal compositions resulting from magma ocean models.
Bulk Mercurian
mantle
composition
Non-chondritic CB avg. b ~ 2 wt% FeO
Source of crustal
material
CB
chondritic
average
Chondritic CB
avg b ~ 2 wt.%
FeO
Chondritic
CB avg. N
~ 2 wt.% FeO
Hart and
Zindler (1986)
Earth
Morgan and
Anders (1980)
Morgan and
Anders
(1980)
Adiabatic melting Adiabatic melting Flotation crust
of overturning
of overturning
deep cumulates
shallow cumulates
Adiabatic
melting of
overturning
cumulates
Flotation
crust
Adiabatic
melting of
overturning
cumulates
Adiabatic
melting of
overturning
cumulates
10% melt of 10% melt of
bulk
bulk
composition composition
Reference
This study
This study
This study
This study
This study
This study
This study
Taylor and
Scott
(2003)a
Taylor and
Scott
(2003)a
Flotation crust
mineralogy
None
None
Quartz N plagioclase None
Plagioclase
Plagioclase
Plagioclase
None
None
56
43
Trace
1
Trace
0.1
Not modeled
0.1
5 – 50
71
15
7
6
Trace
0.5
Not modeled
0.1
5 – 50
49.2
10.9
14.4
18.5
1.7
1.6
0.3
3.7
44.7
19
13.1
18.8
0.9
0.3
Volcanic crust:
SiO2
MgO
Al2O3
CaO
TiO2
Cr2O3
Na2O
FeO
Resulting crustal
Thickness (km)
a
Non-chondritic CB
avg. N2 wt.% FeO
~ 40 – 60
43–48
52–57
Trace–0.2
Trace–0.5
Trace
Trace
Not modeled
0.1 – 0.2
10 – 100
~ 40 – 60
~ 30
3.2
~ 30
No flotation was considered in these models.
They calculate a 10% partial melt at 1 GPa of the bulk compositions
using John Longhi's phase-equilibria program MAGPOX. The igneous
crust produced in our models contain much lower aluminum and
calcium content than the Taylor and Scott (2003) lava predictions, and
in the non-chondritic CB bulk composition case, much higher silica.
These differences arise from the material that melts: Our models
incorporate a differentiated planet, while the Taylor and Scott (2003)
model assumes a homogeneous bulk mantle composition.
If the mantle is heterogeneous, as our models indicate, Mercury
may carry a relatively high iron oxide fraction at depth following
magma ocean overturn, while crustal lavas contain little iron,
reflecting their low-iron source regions (Fig. 7). This may be evident
on the Moon, as various bulk compositions (Dreibus and Wänke, 1985;
Longhi, 2003) range between 7.5% and 14% FeO, while measured lava
basalts contain 15.5%–22% FeO (Papike et al., 1998).
Thus, the bulk mantle density of Mercury would still be low and
consistent with estimates, but regions within the interior may contain
more iron oxide and have a higher density. Harder and Schubert
(2001) have shown that the modeled size of Mercury's core depends
directly upon assumptions of its composition. Models of the internal
structure of Mercury are constrained only by the radius and mean
density, but improvements on characterization of the moment of
inertia and the gravitational field of the planet will help determine the
size of the core if we assume densities for the crust, mantle, and core
(Solomon, 2007; Margot et al., 2007). In Mercury interior models a
mantle containing higher iron oxide fraction at depth might predict a
smaller core radius. It may be difficult to distinguish the effect of the
mantle density, however, as the uncertainties in the core composition
may dominate the size of the core (Hauck et al., 2007; Riner et al.,
2008). Note, though, that the low-iron oxide bulk mantle compositions favored in this modeling predict mantle densities between 3000
and 3100 kg m− 3, within a range of some of the models of Hauck et al.
(2007).
4.2. Exogenous models for producing a low-iron oxide crust
To make Mercury with the observed core and crust characteristics
from a more Earthlike bulk composition, one of the exogenous models
Fig. 7. Iron oxide content of mantle cumulates following solidification but before overturn, for chondritic CB bulk composition with 1 wt.% FeO in bulk composition. Fractionation
produces higher iron oxide in cumulates solidifying last, nearer the surface (though these will likely sink during gravitationally-driven overturn). This figure demonstrates the ability
of magma ocean fractionation to create heterogeneous mantle source regions for later melting.
454
S.M. Brown, L.T. Elkins-Tanton / Earth and Planetary Science Letters 286 (2009) 446–455
must be considered. Here we examine the Benz et al. (1988, 2007)
giant impact model.
If the proto-Mercury had a core fraction similar to the Earth's, then
the proto-Mercury was a planet about the size of Mars, with a radius of
~ 3400 km. The position of high-iron oxide mantle cumulates at the
time of impact may determine whether the resulting Mercury has a
high-iron oxide or a low-iron oxide crust. During such a giant impact it
is possible that the shallower mantle will be lost preferentially to the
deep mantle (Kokubo, personal communication, and Andrews-Hanna
et al., 2009). If the giant impact occurs after cumulate overturn when
high-iron oxide cumulates had sunk into the planetary interior, the
deep mantle may be the part of the proto-planet retained for the new,
less massive mantle, and it will be rich in iron oxide. The energy of
impact will likely create a second magma ocean, and flotation of
buoyant low-iron phases such as plagioclase may be possible. An
igneous crust created in this secondary magma ocean, however, will
be too iron oxide-rich to match observations of Mercury.
Alternatively, the giant impact may occur at or toward the end of
magma ocean solidification. In this second scenario iron oxide has
been enriched in evolving magma ocean liquids and resides largely
near the surface of the planet and overturn has not begun to carry the
iron oxide-rich mineral assemblages to depth. If the impact occurs at
that moment a majority of the iron oxide may be lost from the planet,
and any secondary igneous crust will therefore contain silicate
minerals low in iron oxide. This may describe how a Mercury formed
by giant impact produced a crust low in iron-bearing silicates,
previously unexplained (Taylor and Scott, 2003).
The timeframe required for this scenario is relatively well defined;
impact must occur after a majority of the magma ocean is solidified,
perhaps within 50,000 years of the last giant accretionary impact, and
before cumulate overturn begins, in less than one million years
(Elkins-Tanton, 2008). This is a generous interval during which protoMercury may have experienced another giant impact following an
initial magma ocean in this accretion scenario.
If a giant impact can preferentially remove the shallow mantle,
therefore, then a giant impact on a newly-solidified and not yet
overturned cumulate mantle can result in a low-iron oxide mantle
consistent with surface observations. If a giant impact necessarily
mixes the mantle, then Mercury must have an inherently low-iron
oxide bulk silicate composition; otherwise its surface would consist of
either plagioclase flotation or higher-iron oxide lavas.
4.3. Opaques on the surface of Mercury
Our models indicate that an oldest volcanic crust on Mercury may
form with few or no opaque minerals. Instead, these opaque phases
solidify in a layer toward the end of the magma ocean solidification,
analogously to models of the lunar magma ocean, and reside in the
uppermost mantle at the end of solidification. In these models the
source regions for ilmenite and chromite solidify between the surface
and 120 km depth (Figs. 2, 3, and 5).
These late, high-titanium and high-chromium cumulates will not
crystallize until temperatures have fallen to about 1150 °C (Hess and
Parmentier, 1995; Van Orman and Grove, 2000). By the time the dense
material has finished solidifying and any gravitational instabilities
begun to form, the temperature would have fallen farther. These low
temperatures near the Mercurian surface produce high viscosity in the
opaque-bearing layer. The viscosity may be sufficient, in fact, to
prevent their overturn (Elkins-Tanton et al., 2002) and leave them
near the planetary surface. Riner et al. (2009) discuss the possibility
that the spectra of the low reflectance material seen on Mercury's
surface reflect this magma ocean oxide layer, excavated by later
impacts (see also Robinson and Lucey, 1997; Robinson and Taylor,
2001). The material could also be delivered to the surface by later
magmatic activity, either from standard planetary convection or from
impacts. This is a topic of ongoing work.
5. Conclusions
Here we have considered a range of initial Mercurian mantle
compositions, from an Earth-like composition to a non-chondritic CB
high silica composition. Models of conditions in the early solar system
may inform the likelihood of these compositions. During planetary
formation the Sun's emissions in the visible were far weaker than they
are today; however, high-energy emissions were much stronger
(Ribas et al., 2005). This strong solar wind and X-ray flux may have
produced a more metallic iron-rich composition nearer the Sun.
Conversely, dynamical models of planetary formation indicate that
radial mixing of planetary embryos during oligarchic accretion is
ubiquitous, though less extreme nearest the Sun (O'Brien et al., 2006).
Taken together, a more reduced, CB-like composition at Mercury's
radius is a reasonable hypothesis, though more Earth-like bulk
compositions are also reasonable given the likelihood of radial mixing.
Highly reducing conditions near the Sun in the early solar system
might have important implications for the composition of Mercury's
core: under highly reducing conditions iron is likely to form bonds
with alternative anions, such as sulfur. FeS is likely to participate in
core formation. Such conditions might encourage the formation of a
planetary core with an unusual fraction of light alloys, and thus allow
the preservation of today's molten core (Harder and Schubert, 2001;
Hauck et al., 2007; Riner et al., 2008).
Formation of Mercury from a low-iron oxide CB bulk composition
would create a planet with a sufficiently large metallic iron core and
low iron oxide crust to match observations of Mercury. A low-iron
oxide ancient crust can be made by plagioclase flotation, and in the
case of a non-chondritic CB composition, a plagioclase plus quartz
flotation crust. These crusts are only possible, however, if the bulk
mantle composition has more than 2 to 7 wt.% FeO and can thus
produce a dense magma ocean liquid.
Formation from compositions with these iron oxide contents
would also create at least portions of the planetary mantle with high
iron oxide through fractionation in the magma ocean. Later melting of
these regions would produce a higher-iron oxide volcanic crust, which
would be a critical clue to mantle compositional variations and thus
formation processes if later found by remote sensing. If all the
magmatic crust is low in iron oxides in its silicate fraction, then all the
mantle source regions they represent are similarly low in iron oxide.
A giant impact model that preferentially removes shallow mantle
cumulates may also be consistent with the low iron-bearing crustal
silicates found on Mercury, but this requires one of two conditions. In
the first, Mercury can be formed from a higher-iron oxide bulk
composition, but the giant impact preferentially removes the higher
iron cumulates from the upper half of the mantle before gravitationallydriven cumulate overturn carries them to depth. Alternatively, the
proto-Mercury already had an unusually low-iron oxide bulk silicate
composition.
Evidence for a plagioclase flotation crust on Mercury found by
MESSENGER would be consistent both with an endogenous model
with N2 to 7 wt.% FeO in the bulk mantle, or with an exogenous
giant impact model. If, conversely, the low-iron oxide silicate crust
is not primarily plagioclase and is ubiquitous, then Mercury's
bulk mantle composition must be low in iron oxide and would be
consistent with formation from a low-iron oxide CB-like reduced
bulk composition.
Acknowledgements
A suggestion from Paul Hess was the initial inspiration for this
research. The National Science Foundation Astronomy program and
the MIT Undergraduate Research Opportunities Program funded
this work. This paper has been greatly improved by discussions
with and reviews by Nancy Chabot, Larry Nittler, and an anonymous
reviewer.
S.M. Brown, L.T. Elkins-Tanton / Earth and Planetary Science Letters 286 (2009) 446–455
References
Anderson, J.D., 1975. Planetary geodesy. Reviews of Geophysics and Space Physics 13,
274–275.
Anderson, D.L., Kovach, R.L., 1967. The composition of the terrestrial planets. Earth and
Planetary Science Letters 3, 19–24.
Anderson, J.D., Colombo, G., Espsitio, P.B., Lau, E.L., Trager, G.B., 1987. The mass, gravity
field, and ephemeris of Mercury. Icarus 71, 337–349.
Andrews-Hanna, J.C., Zuber, M.T., Banerdt, W.B., 2009. The Borealis Basin and the origin
of the martian crustal dichotomy. Nature 453, 1212–1216 The newest version of the
giant impact origin for the Martian northern lowlands: a single glancing blow.
Benz, W., Slattery, W.L., Cameron, A.G.W., 1988. Collisional stripping of Mercury's
mantle. Icarus 74, 516–528.
Benz, W., Anic, A., Horner, J., Whitby, J.A., 2007. The origin of Mercury. Icarus 132, 189–202.
Boss, A.P., 1998. Temperatures in protoplanetary disks. Annual Review of Earth and
Planetary Sciences 26, 53–80.
Boss, A.P., 2003. The solar nebula. In: Turekian, K.K., Holland, H.D. (Eds.), Treatise on
Geochemistry. Elsevier Ltd, pp. 1–23.
Cameron, A.G.W., 1985a. The partial volatilization of Mercury. Icarus 64, 285–294.
Cameron, A.G.W., 1985b. Formation and evolution of the primitive solar nebula. In:
Black, D.C., Matthews, M.S. (Eds.), Protostars and Planets II. University of Arizona
Press, Tucson, pp. 1073–1099.
Cameron, A.G.W., 1988. Origin of the solar system. Annual Review of Astronomy and
Astrophysics 26, 441–472.
Cameron, A.G.W., Ward, W.R., 1976. The origin of the Moon. Lunar and Planetary
Institute Conference Abstracts 7, 120–123.
Cameron, A.G.W., Decampli, W.M., Bodenheimer, P., 1982. Evolution of giant gaseous
protoplanets embedded in the primitive solar nebula. Icarus 49, 298–312.
Cameron, A.G.W., Benz, W., Slattery, W.L., 1987. Planetary collision calculations: origin of
Mercury. Lunar and Planetary Institute Conference Abstracts 18, 151–152.
Chambers, J.E., 2003. Planet formation. In: Turekian, K.K., Holland, H.D. (Eds.), Treatise
on Geochemistry. Elsevier Ltd, pp. 1–17.
Chambers, J.E., 2004. Planetary accretion in the inner Solar System. Earth and Planetary
Science Letters 223, 241–252.
Decampli, W.M., Cameron, A.G.W., 1979. Structure and evolution of isolated giant
gaseous protoplanets. Icarus 38, 367–391.
Denevi, B.W., Robinson, M.S., 2008. Mercury's albedo from Mariner 10: implications for
the presence of ferrous iron. Icarus 197, 239–246.
Dreibus, G., Wänke, H., 1985. Mars, a volatile-rich planet. Meteoritics 20, 367–382.
Elkins-Tanton, L.T., 2008. Linked magma ocean solidification and atmospheric growth
for Earth and Mars. Earth and Planetary Science Letters 271, 181–191.
Elkins-Tanton, L.T., Van Orman, J.A., Hager, B.H., Grove, T.L., 2002. Reexamination of the
lunar magma ocean cumulate overturn hypothesis: melting or mixing is required.
Earth and Planetary Science Letters 196, 249–259.
Elkins-Tanton, L.T., Parmentier, E.M., Hess, P.C., 2003. Magma ocean fractional crystallization and cumulate overturn in terrestrial planets: implications for Mars. Meteoritics and Planetary Science 38, 1753–1771.
Elkins-Tanton, L.T., Parmentier, E.M., Hess, P.C., 2005a. The formation of ancient crust on
Mars through magma ocean processes. Journal of Geophysical Research 110, E12S01.
Elkins-Tanton, L.T., Zaranek, S., Parmentier, E.M., 2005b. Early magnetic field and
magmatic activity on Mars from magma ocean overturn. Earth and Planetary
Science Letters 236, 1–12.
Fegley, B., Cameron, A.G.W., 1987. A vaporization model for iron/silicate fractionation in
the Mercury protoplanet. Earth and Planetary Science Letters 82, 207–222.
Ghiorso, M.S., Hirschmann, M.M., Reiners, P.W., Kress III, V.C., 2002. The pMELTS: a
revision of MELTS aimed at improving calculation of phase relations and major
element partitioning involved in partial melting of the mantle at pressures up to
3 GPa. Geochemistry, Geophysics, Geosystems 3 (5), 1–36.
Gladman, B.J., Burns, J.A., Duncan, M., Lee, P., Levison, H.F., 1996. The exchange of impact
ejecta between terrestrial planets. Science 271, 1387–1392.
Goettel, K.A., 1988. Present bounds on the bulk composition of Mercury — implications
for planetary formation processes. In: Vilas, F., Chapman, C.R., Matthews, M.S.
(Eds.), Mercury. University of Arizona Press, City, pp. 613–621.
Harder, H., Schubert, G., 2001. Sulfur in Mercury's core? Icarus 151, 118–122.
Hart, S.R., Zindler, A., 1986. In search of a bulk Earth composition. Chemical Geology 57,
247–267.
Hartmann, W.K., Davis, D.R., 1975. Satellite-sized planetesimals and lunar origin. Icarus
24, 504–514.
Hauck, S.A., Solomon, S.C., Smith, D.A., 2007. Predicted recovery of Mercury's internal
structure by MESSENGER. Geophysical Research Letters 34, L18201.
Hess, P.C., Parmentier, E.M., 1995. A model for the thermal and chemical evolution of the
Moon's interior: implications for the onset of mare volcanism. Earth and Planetary
Science Letters 134, 501–514.
Jarosewich, E., 1990. Chemical analyses of meteorites: a compilation of stony and iron
meteorite analyses. Meteoritics 25, 323–337.
Kenkyujo, K.K., Keizo, Y., Hideyasu, K., 1995. Catalog of the Antarctic meteorites:
collected from December 1969 to December 1994, with special reference to those
represented in the collections of the National Institute of Polar Research. The
Institute, Tokyo.
Latimer, W.M., 1950. Astrochemical problems in the formation of the Earth. Science 112,
101–104.
Lauretta, D.S., Goreva, J.S., Hill, D.H., Killgore, M., 2007. Bulk compositions of the CB
Chondrites Bencubbin, Fountain Hills, MAC 02675, and MIL 05082. Lunar and
Planetary Institute Conference Abstracts 38, 2236.
455
Lewis, J.S., 1972. Metal/silicate fractionation in the Solar System. Earth and Planetary
Science Letters 15, 286–290.
Lodders, K., Fegley, B., 1998. The Planetary Scientist's Companion. Oxford University
Press, New York.
Longhi, J., 2003. A new view of lunar ferroan anorthosites: postmagma ocean
petrogenesis. Journal of Geophysical Research 108, 5083.
Love, S.G., Keil, K., 1995. Recognizing Mercurian meteorites. Meteoritics 30, 269–278.
Margot, J.L., Peale, S.J., Jurgens, R.F., Slade, M.A., Holin, I.V., 2007. Large longitude
libration of Mercury reveals a molten core. Science 316, 710–714.
McClintock, W.E., Izenberg, N.R., Holsclaw, G.M., Blewett, D.T., Domingue, D.L., Head, J.W.,
Helbert, J., Mccoy, T.J., Murchie, S.L., Robinson, M.S., Solomon, S.C., Sprague, A.L.,
Vilas, F., 2008. Spectroscopic observations of Mercury's surface reflectance during
MESSENGER's first Mercury flyby. Science 321, 62–65.
Morgan, J.W., Anders, E., 1980. Chemical composition of Earth, Venus, and Mercury.
National Academy of Sciences, Proceedings 77, 6973–6977.
Newcomb, S., 1895. The Elements of the Four Inner Planets and the Fundamental Constants
of Astronomy. Government Printing Office, Washington.
O'Brien, D.P., Morbidelli, A., Levison, H.F., 2006. Terrestrial planet formation with strong
dynamical friction. Icarus 184, 39–58.
Papike, J.J., Ryder, G., Shearer, C.K., 1998. Lunar samples. In: Papike, J.J. (Ed.), Reviews in
Mineralogy: Planetary Materials. Mineralogical Society of America, pp. 5.1–5.234. DC.
Rabe, E., 1950. Derivation of fundamental astronomical constants from the observations
of Eros during 1926–1945. Astronomical Journal 55, 112.
Raymond, S.N., Quinn, T., Lunine, J.I., 2006. High-resolution simulations of the final assembly
of Earth-like planets I. Terrestrial accretion and dynamics. Icarus 183, 265–282.
Reynolds, R.T., Summers, A.L., 1969. Calculations on the composition of the terrestrial
planets. Journal of Geophysical Research 74, 2494–2511.
Ribas, I., Guinan, E.F., Güdel, M., Audard, M., 2005. Evolution of the solar activity over
time and effects on planetary atmospheres. I. High-energy irradiances (1–1700 Å).
Astrophysical Journal 622, 680–694.
Riner, M.A., Bina, C.R., Robinson, M.S., 2008. Internal structure of Mercury: implications of a
molten core. Journal of Geophysical Research 113, E08013.
Riner, M.A., Lucey, P.G., Desch, S.J., McCubbin, F.M., 2009. Nature of opaque components
on Mercury: insights into a Mercurian magma ocean. Geophysical Research Letters
36, L02201.
Robinson, M.S., Lucey, P.G., 1997. Recalibrated Mariner 10 color mosaics: implications for
mercurian volcanism. Science 275, 197–200.
Robinson, M.S., Taylor, G.J., 2001. Ferrous oxide in Mercury's crust and mantle. Meteoritics
and Planetary Science 36, 841–847.
Robinson, M.S., Murchie, S.L., Blewett, D.T., Domingue, D.L., Hawkins, S.E., Head, J.W.,
Holsclaw, G.M., Mcclintock, W.E., Mccoy, T.J., Mcnutt, R.L., Prockter, L.M., Solomon, S.C.,
Watters, T.R., 2008. Reflectance and color variations on Mercury: regolith processes
and compositional heterogeneity. Science 321, 66–69.
Safronov, V.S., 1972. Evolution of the Protoplanetary Cloud and the Formation of the
Earth and the Planets. Israel Program for Scientific Translations, Jerusalem.
Siegfried II, R.W., Solomon, S.C., 1974. Mercury — internal structure and thermal
evolution. Icarus 23, 192–205.
Smith, J.V., Anderson, A.T., Newton, R.C., Olsen, E.J., Crewe, A.V., Isaacson, M.S., Johnson, D.,
Wyllie, P.J.,1970. Petrologic history of the moon inferred from petrography, mineralogy
and petrogenesis of Apollo 11 rocks. Geochimica et Cosmochimica Acta Supplement 1,
897–925.
Solomatov, V.S., 2000. Fluid dynamics of a terrestrial magma ocean. In: Canup, R.M.,
Righter, K. (Eds.), Origin of the Earth and Moon. University of Arizona Press, Tuscon,
pp. 323–338.
Solomon, S.C., 2003. Mercury: the enigmatic innermost planet. Earth and Planetary
Science Letters 216, 441–455.
Solomon, S.C., 2007. Hot news on Mercury's core. Science 316 (5825), 702–703.
Solomon, S.C., McNutt, R.L., Watters, T.R., Lawrence, D.J., Feldman, W.C., Head, J.W.,
Krimigis, S.M., Murchie, S.L., Phillips, R.J., Slavin, J.A., Zuber, M.T., 2008. Return to
Mercury: a global perspective on MESSENGER's first Mercury flyby. Science 321, 59–62.
Sprague, A.L., Donaldson Hanna, K.L., Kozlowski, R.W.H., Helbert, J., Maturilli, A., Warell, J.B.,
Hora, J.L., 2009. Spectral emissivity measurements of Mercury's surface indicate Mg- and
Ca-rich mineralogy, K-spar, Na-rich plagioclase, rutile, with possible perovskite, and
garnet. Planetary and Space Science 57, 364–383.
Taylor, G.J., Scott, E.R.D., 2003. Mercury. In: Turekian, K.K., Holland, H.D. (Eds.), Treatise
on Geochemistry. Elsevier Ltd, pp. 477–485.
Urey, H.C., 1952. The Planets: Their Origin and Development. Yale University Press, New
Haven.
Van Orman, J.A., Grove, T.L., 2000. Origin of lunar high-Ti ultramafic glasses: constraints
from phase relations and dissolution kinetics of clinopyroxene–ilmenite cumulates.
Meteoritics and Planetary Sciences 35, 783–794.
Warren, P.H., 1985. The magma ocean concept and lunar evolution. Annual Review of
Earth and Planetary Sciences 13, 201–240.
Weidenschilling, S.J., 1978. Iron/silicate fractionation and the origin of Mercury. Icarus
35, 99–111.
Weisberg, M.K., Prinz, M., Humayun, M., Campbell, A.J., 2000. Origin of metal in the CB
(Bencubbinite) chondrites. Lunar and Planetary Institute Conference Abstracts 31, 1466.
Weisberg, M.K., Prinz, M., Clayton, R.N., Mayeda, T.K., Sugiura, N., Zashu, S., Ebihara, M.,
2001. A new metal-rich chondrite grouplet. Meteoritics and Planetary Science 36,
401–418.
Wetherill, G., 1984. Orbital evolution of impact ejecta from Mars. Meteoritics 19, 1–12.
Wood, J.A., Dickey Jr., J.S., Marvin, U.B., Powell, B.N., 1970. Lunar anorthosites and a
geophysical model of the moon. Geochimica et Cosmochimica Acta Supplement 1,
965–988.