News & Views | Published:

Earth science

Small differences in sameness

Nature volume 497, pages 4345 (02 May 2013) | Download Citation

Fresh evidence shows that the iron isotopic composition of Earth's silicate component does not, as was previously thought, reflect the formation of the planet's core at high pressure nor losses of material to space.

Writing in Earth and Planetary Science Letters, Craddock et al.1 provide strong evidence that iron-isotope differences between planetary samples reflect the origins of the samples themselves rather than isotopic fractionation during planet formation. Although this is a negative result, it says a lot about planet and core formation.

Meteorites provide an invaluable archive of the circumstellar disk from which the terrestrial planets and asteroids formed. With the advent of accurate mass spectrometry and its application to meteorite samples, it was soon shown that this disk had relatively uniform isotopic compositions. For example, the uranium found on Earth has the same atomic mass as the uranium found in meteorites from the asteroid belt that lies between Mars and Jupiter, showing that the mix of isotopes from diverse primordial circumstellar-disk material was about the same. With the more recent development of a technique called multiple-collector inductively coupled plasma mass spectrometry, it has become possible to explore this 'sameness' to much higher precision and in many more elements. This has led to a search for small, mass-dependent isotopic differences that may have been imposed by effects that acted to separate (fractionate) the isotopes during planet formation. The resolution of such effects could help to confirm or refute theories about the dynamic processes that formed Earth and its metallic core2,3,4.

This new area of isotope geochemistry has generated strong debate, because most isotopic differences reported so far have been small — less than about 100 parts per million per atomic mass unit (p.p.m. per AMU) — and have been obtained at the technical limits of what can be reliably resolved. Equally heated have been debates about whether systematic isotopic differences between samples can be scaled up to define planetary compositions at all. This is the focus of Craddock and colleagues' study.

Mass-dependent isotopic fractionation could in principle result from loss of planetary material to space through vaporization2,3,4,5, or loss to a planet's core during core formation6,7 (Fig. 1). In both cases, there could be a slight difference in terms of the ease of incorporation of a lighter isotope in one phase relative to another. These two phases could be vapour and liquid in the case of material lost to space, or silicate and metallic liquids in the case of core formation. Both of these processes are relevant to Earth's formation by mass accretion, which probably occurred at the same time as core formation by means of a series of large, stochastic, gravity-driven collisions over tens of millions of years.

Figure 1: Earth's formation and iron isotopic composition.
Figure 1

Earth formed by the cumulative accretion of smaller planets and impactors. Melting from the accretion energy of these impactors would have led to segregation of dense metal (yellow) from the residual silicate of the planet (red to black, with black denoting a lower degree of melting), resulting in concomitant growth of its metallic core. The figure illustrates schematically how the iron isotopic composition of the silicate part could have been modified or left unchanged during this process, depending on the conditions of accretion and core formation. a, Formation of the core at low pressure is thought to leave the composition unchanged6. b, Conversely, this composition should become heavy if the core formed at high pressures6,10. c, If volatilized iron is lost to space during a collision between the proto-Earth and a small planet, then the iron isotopic composition of the residual silicate Earth should be heavy5. d, If the proto-Earth grows by repeatedly colliding with planets with low-pressure cores and the metal mixes directly with metal and silicate with silicate13,14,15, the core will grow without a change in iron isotopic composition. Craddock and colleagues' results1, coupled with earlier findings6,8,10,11, provide evidence that b and c either resulted in no change in the iron isotopic composition, contrary to expectation, or were unimportant in the later history of Earth accretion and core formation.

Chemistry-based arguments have accumulated that Earth, and/or the various proto-planets that it incorporated during accretion, may have lost material to space from their outer silicate portions through erosion during the impacts3. As Earth became bigger, the gravitational energy released by accretion would have generated temperatures at which silicates and metals should have been vaporized2. Major growth phases through collisions are termed giant impacts, and the last such collision between Earth and a smaller planet named Theia, often referred to as 'the giant impact', led to the formation of the Moon from condensation and accretion in the resultant disk of vapour and debris2,4. If some of the material was lost to space rather than re-accreted to Earth and the Moon, then elements that should have only partially entered the vapour phase at these temperatures and pressures, such as lithium, silicon and iron, might show resolvable isotopic differences.

The iron isotopic composition of lunar basalt rocks has been found5 to be on average slightly enriched in the heavier iron isotopes (about 30 p.p.m. per AMU) compared with most terrestrial mantle-derived samples, mainly basalts. The average for data from Earth is in turn slightly heavier (about 30 p.p.m. per AMU) than that for data on basalts from Mars and the asteroid Vesta. Further work confirmed that lunar basalts can indeed have a heavy iron isotopic composition, although this depends on the types of basalt analysed8. Furthermore, lunar basalts do not have a heavy isotopic composition for the light element lithium9, which seems inconsistent with the idea that there were losses of lighter isotopes of iron during vaporization in the Moon-forming giant impact.

It has been argued instead that the Moon's apparent heavy iron isotopic composition might simply reflect that of the outer silicate part of Earth, which in turn was heavy because of the high pressure involved in core formation6. Recently, it was found that iron isotopes can become fractionated as a result of 'disproportionation' of ferrous iron into core-forming metal and oxidized ferric iron in the presence of perovskite minerals in the mantle10. Separation of this metal to the core is one mechanism that might explain why the silicate Earth is oxidized and why iron in terrestrial and lunar basalts is isotopically heavy, as it is for silicon7.

More detailed studies of Earth1,8,11, for which plentiful samples of the solid mantle are available, have raised the question of whether basalts are representative of planetary composition at all. In their favour, the mantle is compositionally heterogeneous, so individual fragments are not always representative. By contrast, basalts are derived by partial melting of large volumes of mantle and therefore provide a more effective method of averaging planetary heterogeneity. However, Craddock and colleagues' study clearly demonstrates that solid-mantle samples that have undergone melting have a lighter iron isotopic composition than basalts because of fractionation during melting. Furthermore, measurements11,12 of chondrites, a group of primitive meteorites with a similar chemical composition to that of the Sun (if volatile elements are subtracted), show that the iron isotopic composition of the silicate Earth is like that of chondrites and so is no different from that of the Sun and the average Solar System.

These results imply that high-pressure iron-isotope fractionation, which has been demonstrated both theoretically6 and experimentally10, did not in fact substantially affect the silicate Earth's residual iron. The disproportionation of ferrous iron in the presence of perovskite may not have been the mechanism by which the iron in the silicate Earth became oxidized. Also, alternative models for core formation exist that do not involve segregation of metal at high pressures. For example, Earth's core and residual silicate may have grown in part in a more direct fashion by accreting smaller planetary objects that had their own low-pressure cores and by separate admixing of these objects' metal and silicate reservoirs through density differences13. There is supporting evidence for such core–core mixing, both theoretically14 and in the silicate Earth's isotopic13 and chemical15 composition.

For some elements, such as silicon, the isotopic composition of the mantle is not fractionated greatly by melting and is heavy relative to both chondrites and samples from Mars and Vesta, with the most likely explanation being partitioning into the core7. The search is now on to determine which other elements have been isotopically fractionated by core formation and what this tells us about processes and conditions in the early Earth.

Craddock and colleagues' results also raise questions about the assumptions that have been made in simply comparing the iron isotopic compositions of basalts from Earth, the Moon, Mars and Vesta. It has been demonstrated over the past 10 years that the isotopic compositions of nearly all elements are the same in Earth and the Moon, leading to new models of lunar origins4. On this basis, the iron isotopic composition of the bulk Moon is probably also like that of chondrites, and the data for lunar basalts5,8 may also reflect fractionation, but fractionation associated with melting on the Moon.

Recently, it has been argued that zinc isotopes in lunar samples were fractionated during the giant impact16. The new data from Craddock et al. greatly strengthen the argument that mass-dependent isotopic fractionation of elements less volatile than zinc, such as iron, magnesium and lithium, did not occur as a result of losses to space. If material was lost to space during accretion3, it happened without isotopic fractionation of these elements, possibly because accretion was not as energetic as has been thought. Intriguingly, some of the latest simulations of the Moon-forming giant impact provide some support for this latter view4.


  1. 1.

    , & Earth Planet. Sci. Lett. 365, 63–76 (2013).

  2. 2.

    & Earth Planet. Sci. Lett. 262, 438–449 (2007).

  3. 3.

    & Phil. Trans. R. Soc. Lond. A 366, 4205–4238 (2008).

  4. 4.

    & Science 338, 1047–1052 (2012).

  5. 5.

    , , , & Earth Planet. Sci. Lett. 223, 253–266 (2004).

  6. 6.

    Science 323, 912–914 (2009).

  7. 7.

    , , , & Geochim. Cosmochim. Acta 75, 3662–3676 (2011).

  8. 8.

    et al. Earth Planet. Sci. Lett. 240, 251–264 (2005).

  9. 9.

    , & Earth Planet. Sci. Lett. 243, 336–353 (2006).

  10. 10.

    , , , & Earth Planet. Sci. Lett. 321–322, 54–63 (2012).

  11. 11.

    & Earth Planet. Sci. Lett. 252, 342–359 (2006).

  12. 12.

    & Geostand. Geoanal. Res. 35, 101–123 (2011).

  13. 13.

    Nature 427, 505–509 (2004).

  14. 14.

    & Earth Planet. Sci. Lett. 295, 177–186 (2010).

  15. 15.

    et al. Earth Planet. Sci. Lett. 301, 31–42 (2011).

  16. 16.

    , & Nature 490, 376–379 (2012).

Download references

Author information


  1. Alex N. Halliday is in the Department of Earth Sciences, University of Oxford, Oxford OX1 3AN, UK.

    • Alex N. Halliday


  1. Search for Alex N. Halliday in:

Corresponding author

Correspondence to Alex N. Halliday.

About this article

Publication history



Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing