Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.


Earth holds its breath

Some inert-gas isotopes in Earth's atmosphere can only have come from deep inside the planet. We thought we knew how much gas Earth gives up, and how it does it — but a challenge has emerged to the prevailing model.

Slowly, Earth is cooling. The culprit is solid convection, which transports hot, buoyant material to Earth's surface from as far down as the core–mantle boundary, 2,900 kilometres beneath us. When this material reaches the surface, some of it melts, and 'incompatible' elements that are of the wrong size, or the wrong ionic charge, to fit into the remaining solid mineral crystal diffuse into the molten rock. These elements become highly concentrated in the oceanic crust that forms as the melt cools. Volatile elements that are highly incompatible — the inert gases such as helium and argon are generally counted among these1,2,3 — also escape the melt, being 'degassed' into Earth's atmosphere. These elements take little further part in subsequent processes such as continental-crust formation, and are not recycled back into the mantle.

Because their history seems so simple, inert-gas isotopes are widely used to define our understanding of deep-Earth processes: whether there is a volatile-rich reservoir that has been trapped deep in the mantle since Earth's accretion; how efficient mantle convection has been during Earth's history; and the details of the processes controlling the physical and chemical evolution of our planet4. On page 299 of this issue5, Watson et al. present new data that — if their interpretation is correct — would blow apart the basic assumption of all previous inert-gas studies. They argue that argon is between 10,000 and 10 million times more compatible than was previously thought.

Argon is a particularly useful tracer of Earth's mantle convection. Its most significant isotope by far, 40Ar, is produced exclusively by the β-decay of the potassium isotope 40K. Whereas in Earth's early history the amount of 40Ar present was insignificant, today almost 1% of the atmosphere consists of 40Ar, representing about half of the 40Ar produced since Earth formed. The atmospheric level of 40Ar is thus a measure of the efficiency of mantle convection in bringing it from the depths to Earth's surface6,7. The amount of 40Ar in the mantle and the atmosphere must also add up, and so low 40Ar concentrations in the parts of the mantle we can see might mean that hidden parts have higher concentrations8 — an essential detail in reconstructing which parts of the mantle are convecting.

Most mantle melting, and so degassing, takes place at mid-ocean ridges, where oceanic crust is formed as two tectonic plates move apart. Only 10–15% of the available material is actually melted, and argon degasses from this material when it migrates to the surface in what amounts to a continual planetary exhalation. Numerical simulations assuming argon's almost total incompatibility and a vigour of mantle convection similar to that of today calculate a release of the gas over time in good agreement with the proportion of argon in today's atmosphere6.

But on the basis of laboratory tests on olivine and enstatite, two closely related silicate minerals that form the bulk of Earth's upper mantle, Watson and colleagues5 argue that at most only 10–15% of the argon contained in the same volume of melt would be released — rather as if the planet were holding its breath. The corollary is that, to supply the amount of argon that is currently in the atmosphere, convection in the mantle must have been much more vigorous in the past7.

The consequences go further. The helium–argon (4He/40Ar) ratios in basalt samples unaffected by degassing are similar to those predicted from the concentrations of their parent isotopes (potassium for argon, and the α-particle emitters uranium and thorium for helium) in the mantle4; so there can be little separation of helium from argon during melting. Thus, if Watson et al.5 are correct and argon is highly compatible, helium must be compatible too, and must also be only inefficiently extracted from solid minerals. The entire budget of inert gases in the mantle, including 40Ar, is currently calculated from the present-day flux of 3He into the oceans. This helium isotope was trapped in the Earth on its accretion and, in contrast to 40Ar, has no significant production sources in Earth's interior. If helium is compatible, the concentration of 3He in the mantle that maintains the observed outflow rate must also be significantly higher than was assumed.

If 3He levels are low in the convecting mantle, the background production of 4He by uranium and thorium decay in the convecting mantle would rapidly lead to much lower 3He/4He ratios than those observed in the mantle material that is exposed at mid-ocean ridges. The conventional picture thus requires a 3He flux from a volatile-rich geochemical reservoir to balance things out9 (Fig. 1a). Ocean-island basalt rocks from isolated volcanoes that usually form away from ocean ridges can have even higher 3He/4He ratios, and are thought to 'sample' this reservoir directly. Similarly, if the 40Ar concentration in the convecting mantle is low, the total amount of argon in the atmosphere and convecting mantle combined is less than that produced by potassium decay in the Earth. A hidden reservoir with high 40Ar is thus also needed for 'mass balance'8.

Figure 1: Two inside views.

Oceanic crust (grey) is subducted into Earth's interior at continental boundaries and, after residing in the deep mantle for a while, either mixes back into the convecting mantle or rises in plumes (large arrows) from the core–mantle boundary. a, If, as in the conventional view, the concentration of the helium isotope 3He is low in the convecting mantle, a supply of 3He (small arrows) would be needed from a volatile-rich reservoir in the deep mantle to maintain the observed 3He/4He ratios observed at mid-ocean ridges (4He is continuously produced in the mantle by radioactive decay of uranium and thorium)9. Low mantle 3He requires low mantle argon (40Ar) levels, so a deep 40Ar source is also required to balance the observed atmospheric argon concentration10. In this model, ocean-island basalts pick up some of this deep 3He-rich reservoir, explaining their high 3He/4He ratios. b, But if, as Watson et al. suggest5, the inert gases are highly compatible and so the concentration of 3He in the mantle is high, no 3He flux into the convecting mantle or hidden 40Ar reservoir is required to explain either high 3He/4He ratios at the mid-ocean ridges or the ratio of mantle argon to atmospheric argon. The high 3He/4He ratios of the ocean-island basalts might arise because they sample a deep reservoir that was depleted of uranium and thorium at an early stage in Earth's evolution, and is therefore 4He-poor11.

Although Watson et al.5 predict high concentrations of inert gases in the mantle, those levels cannot be more than about 3.5 times higher than current estimates10. This is the 'zero-paradox' reference concentration: the point at which, for argon, the amount of the gas in the whole convecting mantle and the atmosphere summed up would balance the total amount produced in the Earth. Zero-paradox 3He concentrations are the point at which the observed mid-ocean-ridge 3He/4He values balance with the estimated concentrations of uranium and thorium in the convecting mantle — and an external source of 3He (Fig. 1a) is no longer required. The high compatibility of the inert gases opens the way for helium-rich material to be recycled back into the mantle, without the uranium and thorium that would produce 4He (Fig. 1b). This has been suggested as an alternative explanation for the high 3He/4He values measured in ocean-island basalt rocks (see, for example, ref. 11).

So should we rush out to buy this new model? Not yet. With such high compatibility, getting the noble gases out of the mantle at all becomes a significant problem. If we can indeed directly scale 40Ar concentrations in the convecting mantle to the 3He flux into the oceans, 40Ar concentrations in the convecting mantle today would be two to three times higher than the zero-paradox reference value — we would have more 40Ar in the combined convecting mantle plus atmosphere than the Earth has produced since its formation. Watson and colleagues' results5 also point to ultra-slow argon diffusion rates even at mantle temperatures. This would mean that it is very hard for olivine, which is the dominant mineral of the upper mantle, to release its helium or argon at all unless melted. A simple, testable prediction of the model would be that freshly exhumed mantle olivine should contain 3He and 40Ar concentrations somewhere between currently accepted average mantle values and the zero-paradox values.

Like many such laboratory studies, the question arises as to how relevant the results truly are to the mantle. Watson et al.5 exposed the surfaces of their crystals of olivine and enstatite to hot, pressurized argon gas, and deduced high argon compatibility from the high argon concentration at the surface, and very slow diffusion rates from its low penetration into the rest of the crystal. Such surface effects have been observed, in less detail, in other minerals before12. But work on natural glass-olivine samples1, theoretical lattice-accommodation models2 and laboratory experiments2,3, involving crystals grown from a melt containing inert gases, have all resulted in the opposite conclusion — inert-gas incompatibility.

So does the Earth hold its breath? Someone has got it wrong. Let's hope we don't have to hold our own breath too long to find out who.


  1. 1

    Heber, V. S., Brooker, R. A., Kelley, S. P. & Wood, B. J. Geochim. Cosmochim. Acta 71, 1041–1061 (2007).

    CAS  Article  ADS  Google Scholar 

  2. 2

    Marty, B. & Lussiez, P. Chem. Geol. 106, 1–7 (1993).

    CAS  Article  ADS  Google Scholar 

  3. 3

    Brooker, R. A. et al. Nature 423, 738–741 (2003).

    CAS  Article  ADS  Google Scholar 

  4. 4

    Porcelli, D. & Ballentine, C. J. Rev. Mineral. Geochem. 47, 411–480 (2002).

    CAS  Article  Google Scholar 

  5. 5

    Watson, E. B., Thomas, J. B. & Cherniak, D. J. Nature 449, 299–304 (2007).

    CAS  Article  ADS  Google Scholar 

  6. 6

    van Keken, P. E. & Ballentine, C. J. J. Geophys. Res. 104, 7137–7151 (1999).

    Article  ADS  Google Scholar 

  7. 7

    Huang, J. & Davies, G. F. Geochem. Geophys. Geosyst. 8, Q03017 (2007).

    ADS  Google Scholar 

  8. 8

    Hart, R., Dymond, J. & Hogan, L. Nature 278, 156–159 (1979).

    CAS  Article  ADS  Google Scholar 

  9. 9

    O' Nions, R. K. & Oxburgh, E. R. Nature 306, 429–431 (1983).

    CAS  Article  ADS  Google Scholar 

  10. 10

    Ballentine, C. J., van Keken, P. E., Porcelli, D. & Hauri, E. H. Phil. Trans. R. Soc. Lond. A 360, 2611–2631 (2002).

    CAS  Article  ADS  Google Scholar 

  11. 11

    Parman, S. W. Nature 446, 900–903 (2007).

    CAS  Article  ADS  Google Scholar 

  12. 12

    Wartho, J.-A. et al. Earth Planet. Sci. Lett. 170, 141–153 (1999).

    CAS  Article  ADS  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Ballentine, C. Earth holds its breath. Nature 449, 294–296 (2007).

Download citation

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.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

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