Two studies show that evaporation of molten rock was intrinsic to the formation of Earth and other rocky bodies in the Solar System, suggesting that violent collisions played a key part in the formation process. See Letters p.507 & p.511
The concentrations of elements in meteorites called chondrites are thought to reflect the chemistry of the early Solar System — an idea reinforced by the fact that chondrites have a similar composition to that of the Sun. However, Earth and comparable bodies in the Solar System are depleted in some of the more volatile rock-forming elements relative to chondrites1. In general, the more volatile the element, the greater its affinity for gas rather than rock, and the more depleted it is in these bodies. The origin of the volatility trend is hotly debated2,3,4, but two studies now suggest an answer to this abiding geochemical question.
On page 511, Hin et al.5 demonstrate that isotope ratios of magnesium in Earth, Mars and some asteroids provide evidence for extensive evaporation of molten rock. On page 507, Norris and Wood6 report experiments showing that Earth's concentrations of several chalcophile (sulfur-loving) elements are also best explained by the partial melting and evaporation of rock. Both studies propose energetic collisions between planetesimals — the asteroid-sized rocky precursors to planets — as the likely cause of rock melting and evaporation.
During their formation, Earth and many other rocky bodies melted to form a mantle and a crust, with a denser, metallic core. Although most people think of molten rock as being unlikely to evaporate, laboratory experiments7 have shown that such material exposed to the vacuum of space does indeed evaporate, with more-volatile elements evaporating faster than less-volatile ones. In addition, the rapidly vibrating lighter isotopes of an element evaporate faster than the more languid heavy isotopes, resulting in an excess of heavy isotopes in the melted rock that remains.
Hin et al. report an impressive set of high-precision data on the relative abundances of two stable isotopes of magnesium, 24Mg and 25Mg, showing that Earth and similar rocky bodies are slightly enriched in the heavier isotope 25Mg, relative to the chondrites from which they formed. The enrichment is precisely what would be expected if molten rock in planetary precursors had evaporated, conjuring up images of lava floating in the vacuum of space. The authors' data breathe new life into earlier suggestions8 that Earth and chondrites differ in their magnesium isotopic compositions.
Magnesium isotopes are especially useful for such analyses because magnesium is a lithophile (rock-loving) element. Consequently, this element is not lost to planetary atmospheres, nor does it dissolve in metallic cores at temperatures relevant to planetary formation. It is therefore an unambiguous tracer of the history of rock. Hin et al. show that if the evaporation of molten rock was slow enough to allow for thermodynamic equilibrium to occur between the rock and its vapour, previously documented enrichments of the heavy isotopes of silicon9 and iron10 in some melted bodies could also be explained by vapour loss, rather than by sequestration in the core, as previously suggested11,12,13.
In a separate but related study, Norris and Wood melted basaltic rock in a furnace under controlled conditions. They discovered that the evaporation of moderately volatile chalcophile elements accounts for several vexing observations regarding the relative concentrations of these elements in the rocky portion of Earth. In particular, the authors found that the experimentally determined volatilities explained the pattern of depletion of these elements in the rocky Earth14 if the partial pressures of oxygen at the time of evaporation were relatively high, similar to those intrinsic to planet formation (partial pressure is the pressure generated by a component of a mixture of gases). Such conditions could have prevailed only after the hydrogen gas left over from the formation of the Sun had been dispersed by strong stellar winds. Because this dispersal took several million years15, Norris and Wood's experiments not only point to evaporation as a key process, but also constrain the timing of the evaporation events.
But how did the molten rock form? According to both studies, cataclysmic collisions of planetesimals caused the melting and vapour loss (Fig. 1). Both teams correctly point out that the velocities of hot gas can overcome the force of gravity only for small planetesimals — those with a mass about half that of Pluto. Therefore, Earth and other large bodies might have inherited the chemical imprints of vapour loss from these smaller building blocks. Alternatively, computer simulations of giant impacts such as the one that formed the Moon16 allow for vapour loss through more-complicated scenarios.
The conclusions of the two studies differ in important, if nuanced, ways. Hin et al. propose a series of liquid–vapour equilibrium events triggered by planetesimal collisions, in which the rates of condensation and evaporation become equal before the vapour escapes. The question arises as to whether the collisions would really lead to such episodic equilibration and vapour loss. Conversely, Norris and Wood invoke the kinetics of evaporation, rather than equilibrium in the strictest sense. It remains to be seen whether these conclusions are in serious conflict.
The results of the two studies are not yet universally applicable. Mars and Earth, for example, have different silicon isotope ratios that are not easily explained by the authors' models. Moreover, the isotopic effects of elements such as silicon and iron dissolving in the cores of planetary bodies must be accounted for. Distinguishing the effects of core formation from those of vapour loss on isotopic compositions will take further study.
Unresolved problems notwithstanding, the physical chemistry of melting and evaporation could ultimately prove to be a key arbiter in competing models of planet formation. The current studies are not the first to suggest that volatile-element depletion and isotope separation resulted from collisions17, but their relative success should encourage further exploration of the potential role of collisions in determining the chemical and isotopic compositions of planets.Footnote 1
McDonough, W. F. & Sun, S. Chem. Geol. 120, 223–253 (1995).
Allègre, C., Manhès, G. & Lewin, E. Earth Planet. Sci. Lett. 185, 49–69 (2001).
Young, E. D. Earth Planet. Sci. Lett. 183, 321–333 (2000).
Day, J. M. D. & Moynier, F. Phil. Trans. R. Soc. A 372, 20130259 (2014).
Hin, R. C. et al. Nature 549, 511–515 (2017).
Norris, C. A. & Wood, B. J. Nature 549, 507–510 (2017).
Davis, A. M., Hashimoto, A., Clayton, R. N. & Mayeda, T. K. Nature 347, 655–658 (1990).
Young, E. D., Tonui, E., Manning, C. E., Schauble, E. & Macris, C. A. Earth Planet. Sci. Lett. 288, 524–533 (2009).
Pringle, E. A., Moynier, F., Savage, P. S., Badro, J. & Barrat, J.-A. Proc. Natl Acad. Sci. USA 111, 17029–17032 (2014).
Sossi, P. A., Nebel, O., Anand, M. & Poitrasson, F. Earth Planet. Sci. Lett. 449, 360–371 (2016).
Polyakov, V. B. Science 323, 912–914 (2009).
Georg, R. B., Halliday, A. N., Schauble, E. A. & Reynolds, B. C. Nature 447, 1102–1106 (2007).
Shahar, A. et al. Science 352, 580–582 (2016).
Palme, H. & O'Neil, H. St. C. in Treatise on Geochemistry 2nd edn, Vol. 3 (eds Holland, H. D. & Turekian, K. K.) 1–39 (Elsevier, 2014).
Pecaut, M. J. & Mamajek, E. E. Mon. Not. R. Astron. Soc. 461, 794–815 (2016).
Lock, S. J. & Stewart, S. T. J. Geophys. Res. Planets 122, 950–982 (2017).
Poitrasson, F., Halliday, A. N., Lee, D.-C., Levasseur, S. & Teutsch, N. Earth Planet. Sci. Lett. 223, 253–266 (2004).
About this article
Space Science Reviews (2018)