Simulations of Earth's growth show a correlation between the timing of the Moon's formation and the amount of mass that Earth accreted afterwards. This relationship provides a way of measuring the age of our planet. See Letter p.84
The age of the oldest objects in the Solar System is known with remarkable precision — 4,567 million years1,2 — thanks to recent strides in the dating of meteorites. Unfortunately, applying the same dating methods to Earth yields an age that is frustratingly fuzzy. Our planet formed sometime during the first 150 million years of the Solar System's history3,4,5, but we do not know when. On page 84 of this issue, Jacobson et al.6 propose a way of measuring the time at which Earth finished forming using numerical simulations of the planet's growth and its chemical composition. Their result: Earth formed in 95 million years, with an uncertainty of about 32–39 million years, making the planet about 4,470 million years old.
A big part of the problem with dating Earth is that our planet did not appear overnight. Starting from humble beginnings, Earth gradually accumulated material over an extended period of time. Indeed, it is still gaining mass today in the form of meteorites and interplanetary dust particles. What is needed is a milestone in Earth's growth at which we can say that the planet was essentially complete. A widely adopted milestone is the major collision with a planet-sized body that is thought to have formed the Moon7 (Fig. 1). Current theory suggests that Earth experienced several of these 'giant impacts' during its formation, with the Moon-forming impact being the last. Each impact mixed together material that would end up in Earth's metallic core and rocky mantle, as well as adding bulk to both.
The prolonged nature of Earth's growth and the existence of multiple giant impacts complicate the task of dating Earth using radiometric clocks — those that combine known decay rates of radioactive materials with measurements of how these materials and their decay products are distributed within Earth today. For one thing, it is unclear just how much mixing of core and mantle material occurred with each giant impact, and to what extent the radiometric clocks were reset as a result.
Earth probably retains a memory of several of these events, rather than just the last one. Giant impacts may have ejected some of the planet's more volatile elements into space, distorting radiometric dating systems that assume that all of the radioactive decay products are still present. It is also possible that Earth's core and mantle continued to interact for some time after the last giant impact.
This is where modelling the growth of our planet could pay dividends. The formation of the Sun's rocky planets probably passed through several stages8, beginning with micrometre-sized dust grains in the nascent Solar System, proceeding to asteroid-sized bodies known as planetesimals, and then to a few dozen Moon-to-Mars-mass planetary embryos. The accumulation of these planetary embryos into the modern rocky planets through occasional giant impacts was by far the slowest stage, and it largely determined how long Earth took to form.
Jacobson et al. have modelled this final growth stage, beginning with various populations of planetary embryos and planetesimals. They also examined two widely different scenarios for what the giant planets of the Solar System were doing during this time. En route, the authors found an interesting and remarkably robust correlation: the timing of the last giant impact on Earth is inversely related to the amount of mass the planet accumulated afterwards from leftover planetesimals. If the last giant impact occurred early on, there would have been plenty of planetesimals left for Earth to sweep up afterwards. If the last giant impact was late, few planetesimals would have remained, and Earth's growth would have largely ceased.
This correlation provides an independent way of dating the Moon-forming impact, provided that we can measure the amount of material that arrived subsequently. Fortunately, there is a way to do this. Several elements, such as iridium and platinum, show a strong tendency to move into Earth's core. During the upheaval of each giant impact, these elements leached from the planet's mantle, bonding with heavy, iron-rich material destined to sink to the core. After the last giant impact, Earth's mantle should have been almost completely stripped of iridium,platinum and their cousins.
In practice, these elements are present in small amounts in the mantle, and in the same relative proportions seen in many meteorites9. To many researchers, this suggests that Earth acquired a fraction of its mass after the last giant impact, when the core and mantle had ceased separating. Jacobson et al. combine the measured mass of this extra material with the correlation from their simulations to date the Moon-forming impact and then deduce Earth's age. They find it very unlikely that Earth finished forming in the first 38 million years of the Solar System. Their favoured time — 95 million years — is compatible with some radiometric-dating estimates for when Earth's core finished forming4, and makes Earth comfortably older than the oldest minerals known to have formed in its crust10.
Naturally, Jacobson and colleagues' method is only as valid as our picture of how planets form. The standard model for planet formation in the Solar System is far from complete, and it has had a torrid time lately trying to explain some aspects of extrasolar planetary systems, such as the existence of Earth-sized planets orbiting very close to their star11. If conditions during the growth of the planets were different from those assumed by Jacobson et al., then the authors' estimated age for Earth could be incorrect.
Despite this caveat, it is encouraging to see studies that combine the fundamental physics inherent in numerical simulations of planet formation with the wealth of information available on Earth's composition. Understanding how and when the Sun's planets formed is immensely challenging, and researchers need every tool available. Studies such as the one by Jacobson and colleagues may be the best hope for understanding how and when our planet came to be.
Amelin, Y. et al. Earth Planet. Sci. Lett. 300, 343–350 (2010).
Connelly, J. N. et al. Science 338, 651–655 (2012).
Yin, Q. et al. Nature 418, 949–952 (2002).
Allègre, C. J., Manhès, G. & Göpel, C. Earth Planet. Sci. Lett. 267, 386–398 (2008).
Halliday, A. N. Phil. Trans. R. Soc. A 366, 4163–4181 (2008).
Jacobson, S. A. et al. Nature 508, 84–87 (2014).
Canup, R. M. & Asphaug, E. Nature 412, 708–712 (2001).
Morbidelli, A., Lunine, J. I., O'Brien, D. P., Raymond, S. N. & Walsh, K. J. Annu. Rev. Earth Planet. Sci. 40, 251–275 (2012).
Walker, R. J. Chem. Erde Geochem. 69, 101–125 (2009).
Valley, J. W. et al. Nature Geosci. 7, 219–223 (2014).
Dong, S. & Zhu, Z. Astrophys. J. 778, 53 (2013).