The discovery of an inner giant planet in the unusually massive solar system around the star HR 8799 creates an ensemble of planets that is difficult to explain with prevailing theories of planet formation. See Letter p.1080
The solar system around the star HR 8799 should not exist. This system is unlike any other known: it is a massive system that has multiple massive planets, with each giant planet containing many times the mass of all the planets in our Solar System combined. However, on page 1080 of this issue, Marois and collaborators1 present new images of HR 8799 in which yet another equally massive planet is visible.
Previous work2 had imaged three planets around HR 8799, and now we have the surprise discovery of a fourth, HR 8799e, an inner, massive planet (about 10 Jupiter masses) located some 14.5 astronomical units from the star (1 AU is the average distance from Earth to the Sun). One might question the importance of the discovery of another extrasolar planet when more than 500 are known. But the HR 8799 system is the only solar system known to have multiple outer planets (the other three planets, HR 8799b, HR 8799c and HR 8799d, orbit respectively at approximately 68, 38 and 24 AU from the host star, and have estimated masses of about 7, 10 and 10 Jupiters).
As HR 8799 is the only known example of a wide (greater than 25 AU) solar system with multiple giant planets, astronomers were curious to know whether the star's planets could have formed by gravitational collapse3 — one of the most popular theories of outer-planet formation. This theory posits that outer giant planets form from the fragmentation of the disk of gas and dust that develops around stars when they are young. In a process rather like the way binary stars form, a gravitational instability in the disk fragments it and quickly (on a timescale of 10,000 years) leads to the formation of gas-giant planets3. But the discovery of an inner planet such as HR 8799e at 14.5 AU poses a tricky puzzle. At this distance, the disk was neither cold enough nor rotating slowly enough to fragment and undergo gravitational collapse in situ to form HR 8799e3.
To explain the formation of this latest planet, Marois et al.1 appeal to the dominant theory of giant-planet formation: a slower process than gravitational collapse (about 3.5 million years at a distance of 10 AU) in which solid dust grains conglomerate into solid cores of tens of Earth masses and then gravitationally accrete disk gas to grow to Jupiter masses. Such a 'core-accretion' process itself is only marginally fast enough at 14.5 AU to build up HR 8799e's roughly 10 Jupiter masses before the disk gas accretes onto the star in less than 10 Myr. This formation timescale problem3 becomes even more vexing if one considers that, at about 2.6 times the distance HR 8799e is from the host star, HR 8799c would require about 20 times longer (more than about 200 Myr) to grow to the same mass at 38 AU — long after the disk has lost all its gas. What's more, at 68 AU, HR 8799b's formation is truly problematic, requiring an even longer timescale (many times the age of the star) to have formed in situ by core accretion. Hence, neither of the two favoured theories of giant-planet formation can explain how all the planets around HR 8799 formed: HR 8799e is too close to have formed by gravitational collapse, and HR 8799c and HR 8799b are too far out to have formed by core accretion (Fig. 1).
Perhaps all of these massive planets formed at much larger distances (more than at least 50 AU) by the gravitational collapse of an unusually massive disk and then migrated quickly inwards to their current positions, somehow sweeping into a dynamically stable set of 1:2:4 orbital resonances1 (where, for every one orbit of planet c, there are two of d and four of e). This does not really help the situation, however, because it is unlikely that such a massive planet as HR 8799e could have migrated from about 50 to 14.5 AU by means of tidal torques from the residual gas that had not been used to build up the planets. The converse theory, by which the planets all form through core accretion within about 10 AU and then slowly move outwards by scattering lesser objects (planetesimals) inwards, is also problematic because there is probably too limited a reservoir of planetesimals to move a 7-Jupiter-mass object such as HR 8799b outwards some 58 AU. So, despite having a clear view of the system — thanks to the power of adaptive-optics systems and large ground-based telescopes — we cannot currently explain how all four planets formed in a coherent, coeval fashion.
A key strength of direct imaging is that photons can be collected from these self-luminous young planets as they contract, allowing the planetary spectra to be observed (to calculate temperatures and luminosities). The observed brightness of HR 8799b in direct images is much lower than would be expected from its observed temperature, given that evolutionary models indicate that HR 8799b must have a radius larger than that of Jupiter1,2,4. This 'under-luminosity' problem is typical of around half of the extrasolar planets imaged to date. One possible explanation is that dusty, thick, planetary-scale high-latitude cloud 'bands' absorb/scatter light when viewing a young planet over its pole. For example, the 'under-luminous' planets in the HR 8799 system are probably being viewed close to 'pole-on'5, perhaps leading to less light emitted in the direction of Earth. By contrast, 'edge-on' giant planets, such as β-Pictoris b6, look brighter because light streams freely from the brighter equatorial regions between the dark cloud bands. Clearly, further theoretical (and direct imaging) work will be needed to identify the ultimate cause of this under-luminosity problem.
The future holds much promise for more surprises in the field of direct imaging of extrasolar planets. However, it seems unlikely that any other massive outer planets will be found around HR 87997. There is always a chance, though, that low-mass terrestrial planets lie within the star's 10-AU-radius 'asteroid' belt. The next chapter in this story will soon be written by even more powerful ground-based, adaptive-optics imagers8,9 and, let us hope, by more powerful pathfinding, space-based planet- and disk-imaging telescopes10. These pathfinders should eventually lead to a terrestrial-planet-finding telescope even capable of taking spectra of Earth-like planets. Such an achievement could address one of the most pivotal questions in science: how common are truly Earth-like planets and life in our Universe?
About this article
This article and the paper under discussion1 were published online on 8 December 2010.