RESEARCH NEWS & VIEWS
incompatible with prevailing GRB models. The
energy spectrum of the afterglow displayed a
shape (the Planckian form) that is characteris-
tic of a plasma whose electrons arrive at such an
equilibrium that we can say that it has a certain
temperature. Conversely, typical energy spec-
tra of GRBs and their afterglows follow power
laws that well describe the collision of materials
at high velocities, such as that resulting from
the impact of an explosion on its environment.
If the Christmas GRB originates from a
remote hypernova, as do long-lasting GRBs,
why is it so unusual? Could it be a completely
different phenomenon, only apparently similar
to a GRB? So far, observations have not unam-
biguously disclosed a host galaxy. The loca-
tion of the GRB in the sky is such that it could
belong to the Perseus arm of the Milky Way,
but also to a satellite of the Andromeda galaxy.
In their studies, Campana et al.2 and Thöne
et al.3 and set out to address these questions
and to explain the origin of the GRB.
Thöne et al.3 interpret a change in slope and
colour of the Christmas GRB afterglow’s opti-
cal emission as being due to the addition of a
supernova that emerges after 10 days, when the
afterglow has almost vanished. By comparing
the system with similar composite supernova–
GRB objects, they derive a distance for the sys-
tem of 1.6 gigaparsecs and an energy of about
1.4 × 1051 erg, compatible with a typical GRB.
Thöne and colleagues posit that the GRB
originates from a tightly bound binary system
composed of a neutron star and a supergiant
helium star. The helium star would transfer
mass to the neutron star, and the system would
evolve to a phase in which the stars would be
surrounded by a common gas envelope pro-
duced by the expansion of the external layers
of the helium star. The neutron star and the
core of the helium star would merge to form
a black hole or a highly magnetized neutron
star, known as a magnetar4, producing a GRB-
like jet that would eventually emerge through
the common envelope. The authors find that
this model agrees with the observed thermal
spectrum of the GRB. However, the case for
the additional supernova, which would be
unusually faint, is not compelling. There-
fore, the deduced GRB distance from Earth
is uncertain. Moreover, it isn’t clear whether a
host galaxy exists. The only optical object seen
near the position at which the GRB took place is
point-like and extremely faint: were it a galaxy it
would be the faintest ever to host a GRB.
Meanwhile, Campana et al.2 revive a model
proposed5 in 1973, soon after the discovery of
GRBs. In this model, a minor body, such as a
comet or an asteroid, flying by a neutron star
at a distance of less than 5,000 kilometres, gets
tidally disrupted, breaks into fragments and
produces a multi-peak GRB. The authors2 used
recent models for tidal-disruption events6 and
further assumed that the debris of the minor
body forms a transient disk around the star.
They compared these models with the available
4 8 | NAT U R E | VO L 4 8 0 | 1 D E C E M B E R 2 0 1 1
data for the Christmas GRB and found that a
model in which a minor body of 5 × 1020 grams
falls onto a neutron star in our Galaxy is
in excellent agreement, in both temporal
and spectral behaviour, with the data.
Both hypotheses2,3 are plausible and explain
numerous and complex data. But at least one
is wrong and the definitive proof — namely,
unambiguous determination of the GRB’s
distance from Earth — is missing. Thöne and
colleagues’ study raises several questions.
How many binaries pass through a phase
of a common envelope and produce GRB-
like explosions? Do all jets emerge from the
envelope, or is that the case for only a few of
them? How highly collimated is the jet? But
Campana and colleagues’ study also opens
up queries. How many minor bodies pass in
close proximity to neutron stars? And is the
deduced mass for the minor body conceiv-
able? The proposed mass is relatively large
(and consequently rare) for an object in the
Solar System, but larger bodies have actually
been found to orbit neutron stars. We could
EA RTH SC I EN CE
Redox state of
early magmas
A study of cerium in zircon minerals has allowed an assessment of the redox
conditions that prevailed when Earth’s earliest magmas formed. The results suggest
that the mantle became oxidized sooner than had been thought. See Letter p.79
B R U N O S C A I L L E T & FA B R I C E G A I L L A R D
A
prime goal of petrologists has been to
assess the redox conditions of magmas
produced throughout Earth’s evolu-
tion, because magmas are known to affect
various major phenomena — such as the
composition of volcanic emanations, which
are widely believed to affect the composition
of the atmosphere1. But research efforts have
been hampered by the lack of rocks from the
Hadean eon, which encompasses nearly the
first half-billion years of Earth’s existence.
On page 79 of this issue, Trail et al.2 report
their analysis of the sole mineral survivors of
the Hadean, zircon samples more than 4 bil-
lion years old. Their findings allowed them to
determine the ‘fugacity’ of oxygen in Hadean
magmatic melts, a quantity that acts as a meas-
ure of magmatic redox conditions. Unexpect-
edly, the zircons record oxygen fugacities
identical to those in the present-day mantle,
leading the authors to conclude that Hadean
volcanic gases were as highly oxidized as those
emitted today.
The continuous reshaping of the surface
© 2011 Macmillan Publishers Limited. All rights reserved
compute the likelihood of each hypothesis,
and perhaps discard one on the basis of statis-
tical considerations. Unfortunately, however,
such a computation would inevitably involve
considerable conjecture.
In short, not much more can be said about
the nature of the Christmas GRB — except that
the odds are that the event is a rare phenom-
enon that looks like a GRB but falls outside this
category. Whatever the case, it’s hard to escape
the fascination of a possible comet death on
Christmas Day. ■
Enrico Costa is at IASF-Roma, INAF, Via del
Fosso del Cavaliere 100, 00133 Roma, Italy.
e-mail: enrico.costa@iasf-roma.inaf.it
1. Costa, E. et al. Nature 387, 783–785 (1997).
2. Campana, S. et al. Nature 480, 69–71 (2011).
3. Thöne, C. C. et al. Nature 480, 72–74 (2011).
4. Barkov, M. V. & Komissarov, S. S. Mon. Not. R.
Astron. Soc. 415, 944–958 (2011).
5. Harwit, M. & Salpeter, E. E. Astrophys. J. 186,
L37–L39 (1973).
6. Lodato, G. & Rossi, E. M. Mon. Not. R. Astron. Soc.
410, 359–367 (2011).
and deep interior of our dynamic planet has
inevitably resulted in the loss of evidence of
Earth’s ancient composition. In this respect,
Earth contrasts sharply with other planetary
objects — particularly the Moon and Mars,
which are largely unaffected by large-scale geo-
dynamics. Thanks to the Apollo programme
and to meteor­ites inferred to be of Martian
origin, we have rock samples that provide direct
testi­monies of the Moon and Mars’s very early
past. But the same cannot be said of Earth.
Nevertheless, it is generally assumed that all of
these bodies were chemically reduced when
they first formed and then underwent either a
gradual or stepwise oxidation3.
One of the main arguments for a reduced
start is that, during or soon after accretion,
Earth must have entered a magma-ocean stage,
when a sizeable part of the planet was molten
because of the energy released by accretion
processes3. A massive segregation process must
then have occurred that separated metals from
silicates, thus allowing the formation of the
metallic core and the silicate-containing man-
tle. The formation of Earth’s core is thought to
have occurred 30 million to 60 million years

after the beginning of the Solar System. The
coexistence of silicate and metal liquids in the
magma ocean would have made conditions
severely reduced, with oxygen fugacity (f O2)
5 to 8 log units below the current value for the
upper mantle (Fig. 1).
Assuming that such a reduced starting
point is inescapable, how did Earth reach its
modern oxidized state, and how fast did this
happen? Persistent oxidizing conditions at
Earth’s surface started when the planet was
around 2.3 billion to 2.4 billion years old, a
phenomenon known as the Great Oxidation
Event1 (GOE). Rock analyses4 suggest that the
mantle’s redox state has been similar to its cur-
rent state from as early as 3.8 billion years ago.
Trail et al.2 wanted to look even further back in
time, so they developed a technique for study-
ing Hadean zircons, which are the only solids
dating from the first 500 million years of
Earth’s history that are known to have survived
unchanged to the present day. By analysing the
cerium content of the zircons and developing
a method to calibrate their results, the authors
were able to determine f O2 for Hadean magmas.
Their findings extend the mantle’s oxidized
realm to almost 4.4 billion years ago. Although
somewhat tenuous, this is the first direct evi-
dence of the redox state of the earliest Earth. If
the zircons analysed by the authors are repre-
sentative of the Hadean eon, this result shrinks
the duration of the reduced era of Earth’s
mantle to less than 150 million years. It also
increases the lag time between the oxidation
of the mantle and the subsequent oxidation
of the atmosphere (the GOE; Fig. 1), which
might seem to make it difficult to establish
a direct connection between the two events.
However, redox changes upon the release of
gas from magmas5 and changes in the pressure
of volcanic degassing6 could explain why the
atmospheric GOE happened so long after the
mantle’s oxidation. But what was the cause of
the mantle’s great oxidation event?
Planetary-scale geodynamics is often cited
as a mechanism for bringing oxidized mater­
ial into the deep, reduced mantle, and water
is thought to be responsible for oxidation7.
This concept is deeply entrenched among
geologists, although the details of the redox
mechanisms involved remain unknown.
Although a matter of strong debate8,9, Earth’s
current budget of volatile compounds (includ-
ing water) seems to have been reached no later
than 100 million years after core formation,
and so volatiles may have fuelled the mantle’s
great oxidation at about that time (Fig. 1).
However, it is known that terrestrial mag-
mas do not necessarily have a high oxidation
state, even if volatiles are abundant10,11, and
that the Moon’s reduced lavas contain a sub-
stantial amount of volatiles12. Taken together,
these facts argue against the existence of an
un­equivocal relationship between high f O2 and
the abundance of volatiles.
But the accretion of volatiles into Earth’s
NEWS & VIEWS RESEARCH
Arc magmas
Oxidized 4
Zircon Lithospheric mantle and basalts
3
2
1
Mantle great
oxidation event
0
log fO2– IW
–1
–2
Meteorite
bombardment
–3
–4
Core–mantle Oldest Oldest Atmospheric Present
–5 separation zircon rock GOE day
Reduced –6
0.001 0.01 0.1 0.6 2 4.5
Age (billion years)
Figure 1 | Evolution of the redox state of Earth’s mantle.  The oxygen fugacity ( f O2, measured in bar) of
the mantle — a measure of the mantle’s redox state — has varied over time. Here, log f O2 is plotted relative
to a standard value (known as the iron-wüstite buffer; IW); the scale on the x-axis is logarithmic. At first,
the mantle was highly chemically reduced, but it became more oxidized as Earth’s accretion proceeded
and as core–mantle redox equilibration occurred (blue arrow). Meteorite bombardment may have
contributed to oxidation of the mantle8 (orange dotted arrows) once it had segregated from the core. A
‘great oxidation event’ in the mantle (black dotted arrow) then occurred, in which f O2 rapidly increased.
On the basis of their analysis of 4.4-billion-year-old zircon samples, Trail et al.2 report that the mantle’s
redox state about 100 million years after core–mantle separation (green box, left) was similar to that of
the present-day lithospheric mantle (green box, right). The most highly oxidized mantle melts are the
most recent — the arc magmas (red box) that formed as a result of crust–mantle exchanges triggered by
subduction. The timing of some key events in Earth’s history are indicated, along with the times when the
oldest rock and zircon samples were formed; GOE is the Great Oxidation Event.
bulk is not the only possible mechanism for earlier periods of Earth’s history, thereby
oxidation of the mantle. Experiments have refining our knowledge about what f O2 rep-
shown13 that, at the pressure of the lower man- resents. Also much needed are experimental
tle, iron(ii) oxide is converted into iron metal or modelling studies aimed at understanding
and iron(iii) oxide — which means that large the mechanisms of redox processes in magma,
bodies such as Earth can self-oxidize their and the extent to which redox state reflects
mantle, whereas smaller ones cannot (or do so either the origins of magmas or later processes
to a lesser extent). In other words, a nominally that affect them. ■
dry mantle can be oxidized. This mechanism,
however, would require that wholesale mantle Bruno Scaillet and Fabrice Gaillard are at
convection and mixing occurred within a short the CNRS/INSU-Université d’Orléans,
period of time (less than 100 million years). ISTO, UMR 6113, 1a rue de la Férollerie,
Alternatively, changes in the mantle’s redox 45071 Orléans, France.
state could have occurred if the early magma e-mails: bscaille@cnrs-orleans.fr;
ocean crystallized and degassed, because of fabrice.gaillard@cnrs-orleans.fr
the redox effect of degassing5 and the greater
1. Holland, H. D. Geochim. Cosmochim. Acta 66,
affinity of iron(iii) for liquid relative to 3811–3826 (2002).
minerals14 (the iron(iii) would have become 2. Trail, D., Watson, E. B. & Tailby, N. D. Nature 480,
concentrated in the residual liquid, thus 79–82 (2011).
3. Wood, B. J., Walter, M. J. & Wade, J. Nature 441,
increasing the oxidation state of the liquid). 825–833 (2006).
Such processes would probably have affected 4. Li, Z.-X. A. & Lee, C.-T. A. Earth Planet. Sci. Lett. 228,
the entire mantle and would not have required 483–493 (2004).
5. Burgisser, A. & Scaillet, B. Nature 445, 194–197
large-scale convection. (2007).
Clearly, many parts of Earth’s early evolution 6. Gaillard, F., Scaillet, B. & Arndt, N. T. Nature 478,
229–232 (2011).
are still obscured by poor preservation of the 7. Kelley, K. A. & Cottrell, E. Science 325, 605–607
geological record and by our limited know­ (2009).
ledge of the mechanisms that drive magmatic 8. Albarède, F. Nature 461, 1227–1233 (2009).
9. Wood, B. J., Halliday, A. N. & Rehkämper, M. Nature
redox patterns. Even so, the zircon record 467, E6–E7 (2010).
revealed by Trail et al.2 firmly anchors one of 10. Mallmann, G. & O’Neill, H. St C. J. Petrol. 50,
the first redox steps of the infant Earth. This 1765–1794 (2009).
11. Scaillet, B. & Macdonald, R. J. Petrol. 42, 825–845
extraordinary tale is certainly not finished, (2001).
and future work will augment the existing 12. Saal, A. E. et al. Nature 454, 192–195 (2008).
data — perhaps the discovery of other zircon 13. Frost, D. J., Mann, U., Asahara, Y. & Rubie, D. C.
Phil. Trans. R. Soc. A 366, 4315–4337 (2008).
samples will allow f O2 to be determined for 14. Carmichael, I. S. E. Contrib. Mineral. Petrol. 106,
even more highly reducing conditions at 129–141 (1991).
1 D E C E M B E R 2 0 1 1 | VO L 4 8 0 | NAT U R E | 4 9
© 2011 Macmillan Publishers Limited. All rights reserved