Extrasolar Planets Dimitar D

Vol 451|3 January 2008 Q&A ESA/C. CARREAU ESA/C.

ASTRONOMY Extrasolar planets Dimitar D. Sasselov Hundreds of planets are known to orbit stars other than the Sun, and unprecedented observations of their atmospheres and structures are being made. It’s an invaluable opening for understanding the planets’ diverse natures, the formation of our Solar System, and the possibility of habitable planets beyond our home.

2 How many planets outside our Solar for at least one complete orbit of the planet. A will dim the star’s light by a fraction (Rp/Rs) , System have we found so far? wobble can also be detected directly by care- where Rp and Rs are the respective radii of the The number is going up week by week, but at fully observing the position of the star with planet and star. For a planet the size of Jupiter the moment there are about 270 confirmed respect to other stars (a technique known as and a star the size of the Sun, this dimming extrasolar planets — exoplanets, as they’re astrometry), but this is technically even more effect would be roughly 1%, which is easily known in the jargon. Not bad, considering the demanding. detectable even with amateur equipment. The first one was discovered just 13 years ago, and catch is that a transit requires the planet’s orbit they’re not that easy to spot. What other ways are there of finding to be almost exactly edge-on as we look at the exoplanets, besides the Doppler effect? star, which is extremely unlikely. Tens of thou- What makes exoplanets difficult to find? Gravitational lensing is another common sands of stars must be monitored with patience They are, by definition, far away and orbiting method, and also exploits the star–planet mass to detect periodic dimming in just a handful. close to a star that is far bigger and brighter ratio. For this, a source of light is needed, usu- than them. Light contrast ratios of 1010 (at ally another star, behind the star being inves- What can these crude remote-sensing visible wavelengths) to 107 (in the infrared) tigated. As we look at it, the light from the techniques tell us about exoplanets? between a star and planet make direct detec- background star will be bent by the gravity of A surprising amount. The Doppler technique tion by imaging extremely hard. Exploiting the intervening star. If the intervening star has gives the size, period and eccentricity (devia- mass ratios, which are generally in the region an attendant planet, this will alter the lensing tion from a perfect circle) of the planet’s orbit, 3 5 of 10 –10 , is slightly less daunting. Indeed, the effect in a noticeable way. as well as a quantity Mpsin(i) that contains most popular way of spotting an exoplanet is both the planet’s mass Mp and the inclination, through a star’s ‘wobble’, which is caused by the And what about the transiting method? i, of its orbit to our line of sight. The transit gravitational pull of an orbiting planet. This Transiting is perhaps the most fruitful method technique contributes the planet’s radius and a is measured through the Doppler effect — a of observation, because of the information we direct value for i. Combining these two param- shift in wavelength — in the spectrum of vis- can glean about the planet. It exploits the least eters can thus provide the actual value of Mp, ible light from the star. It’s a tiny effect, propor- daunting inequality between star and planet and also — a big prize, this, because it tells us tional to the mass ratio. Measuring it requires — the factor 10 to 100 difference in size. As it something about what the planet is made of patience, because the wobble must be followed passes across the disk of its parent star, a planet — the planet’s mean density, calculated from 29 NEWS & VIEWS Q&A NATURE|Vol 451|3 January 2008

is easy — it probably has a larger core, with log planetary mass (Jupiter masses) more heavy elements in general. But for exo- a Rocky planet 0.01 0.1 1 planets that are less dense, there comes a point Crust 12 TrES-4 when even a planet made just of hydrogen isn’t 1.6 enough to explain its low density. How can we get around this difficulty? 10 HD 209458b 1.4 A possible explanation is additional heating, either because of some persistent heat source Silicate Iron HD 189733b 1.2 or a seriously delayed cooling of the planet. mantle core /He 8 ure H P adii) One obvious source of heat is the parent star km)

4 b Ocean planet 1 itself: the exoplanets in Figure 1 are mainly Water ocean 10 Jupiter × ‘hot Jupiters’ that orbit extremely close to their ( Jupiter r 6 e stars. But then it’s unclear why some planets dius H Saturn % H/ 0.8 50 dius ( would absorb much of that heat and others would not. The same problem — accounting Gliese 436b 0.6 for the entire range of observed planetary sizes 4 H/He Planetary ra Neptune 10% — also bedevils other proposed solutions. Planetary ra s Uranus lanet n p ea 0.4 c d iron What more can remote sensing tell us Water O k an 2 Roc about a planet, besides basic features ice Earth such as its size and density? Rich in iron 0.2 This is where transiting exoplanets come Figure 2 | Journey to the centre of a super- Venus into their own. Their well-constrained basic 0 0 Earth. Super-Earth planets, with masses 1 10 100 1,000 parameters and natural ‘on–off switch’ allow between two and ten times that of Earth, come log planetary mass (Earth masses) their weak signal to be calibrated accurately. in at least two varieties: a, rocky; b, ocean. Both That means we can perform basic spectros- types originate in a well-mixed structure of Figure 1 | The gamut of transiting copy: atomic sodium and hydrogen have solids (silicates) and volatiles (such as water and exoplanets. The composition of a planet will already been detected in exoplanets’ upper ammonia), with trace amounts of hydrogen and determine its average density and where it will atmospheres. Unprecedented spectroscopy noble gases. The structure differentiates very lie on a plot of radius against mass. Here, various and thermal mapping of the atmosphere of the quickly and in a strictly predictable way: iron and planets within (red circles) and beyond (grey transiting hot Jupiter HD 189733 b has been siderophile elements (high-density transition circles) the Solar System are compared with metals that like to bond with iron) precipitate possible at the point where it is seen ‘side-on’, into a core, and excess water, ammonia and so models of different interior composition: from just as it disappears behind its star. Here, the an interior made of rock and iron alone (Venus on precipitate above a silicate mantle. If water is infrared light flux from the planet can be dis- present in significant quantities, most of it will and Earth approximate to this state) to a gaseous –4 composition of just hydrogen and helium (of cerned at a level just 10 that of the star’s flux. be under pressures in excess of 10 gigapascals, which Jupiter is the closest example in our Solar NASA’s infrared Spitzer space telescope has and thus in a solid state — ice — despite the high System). Owing to difficulties in spotting such proved perfect for this task. temperatures. An ocean planet will be larger small bodies, few exoplanets at the small, dense than a rocky one owing to the lower density of end of the scale have so far been found. Many What have we learnt about the water, and the size of a planet of a given mass will giant planets have been detected that have very atmosphere of HD 189733 b? depend on the proportion of core, mantle and low densities above the upper range of pure First, we have a measure of the temperature dif- water. Models show that an uncertainty of less than 5% in radius and less than 10% in mass are hydrogen–helium planets (dotted line) — the ference between the day and night side, which exoplanet TrES-4 is the most extreme example needed to distinguish ocean from rocky planets. — and these are a serious challenge to theory. is about 230 °C. In such an extremely close orbit — HD 189733 b circulates around its star at just 3% of the Earth–Sun distance — a and oxygen. Absence of water might indicate the radius and the mass. More than 20 transit- planet’s rotation and orbital revolution become thermal and radiative loss, as seems to have ing planets are already known, and the radius synchronized, with one side permanently facing been the case with our small and hot planetary and mass of several have been determined to the star. (In the same way, the Moon always neighbour Venus. Hot Jupiters should be mas- better than 4% and 8% accuracy, respectively. presents the same side to Earth.) During tran- sive enough to stave off complete loss, but then But there’s a sting in the tail of this simple, sit, we thus see the night side, and the heat flux again we know little about the extreme condi- surprisingly accurate method — most of the we see coming from this side indicates that the tions in their upper atmospheres. By measur- planets found so far are much less dense than planet’s atmosphere must transport heat very ing the ‘depth’ (the loss or gain in intensity) our theories predict (Fig. 1). efficiently from the day side. The hottest and of eclipse or transit in different infrared wave- coldest regions are not, as would be expected, length bands, Spitzer can crudely sample the Why are less-dense planets a problem seen at the longitudes facing exactly towards molecular lines not only of water, but also of for theoretical models? and away from the star, respectively, but are other gases of interest, such as carbon dioxide Most exoplanets known so far are Jupiter-like slightly offset from these longitudes, indi cating and methane. Water has already been seen on ‘gas giants’, with masses between 0.5 and 3 a very dynamic, windy atmosphere. three hot Jupiters — TrES-1, HD 209458 b and times that of Jupiter (150–1,000 Earth masses). HD 189733 b. But they have an unexpectedly wide range of Is there evidence for water on any sizes, with radii between 0.8 and 1.7 times exoplanet? We seem to know of many larger, Jupiter’s. Jupiter is thought to have a small core It would be surprising not to find some water Jupiter-sized planets — what about of heavy elements, surrounded by envelopes in the atmospheres of hot Jupiters: water is Earth-like planets? of hydrogen with some helium intermingled. very common in the low-temperature environ- First off, the rich diversity of giant exoplan- Explaining an exoplanet with a similar mass to ments where planets form and evolve, owing to ets came as a surprise to astronomers — we Jupiter but a smaller size (higher mean density) the large cosmic abundance of both hydrogen expected to find a greater variety of ‘terres- 30 NATURE|Vol 451|3 January 2008 NEWS & VIEWS Q&A

trial’ planets. But to date we know of only a Distance Density Mean surface Mass (Earth handful. None of them is transiting, so we from star (Earth temperature Notable features masses) don’t know their mean densities or even their (au) densities) (K) exact masses. And ‘terrestrial’ is perhaps a mis- Venus 0.8 0.7 0.95 735 Sweltering CO2 atmosphere nomer: the masses of such planets discovered so far are between 5 and 10 times Earth’s mass Earth 1 1 1 287 Life (ME). Such ‘super-Earths’ do not feature in Gliese 581 c > 5 0.07 > 0.7 > 400 Hot super-Earth our Solar System, but are seemingly common Gliese 581 d > 8 0.25 > 0.8 < 280 Cool super-Earth — habitable? elsewhere. The limit of 10 ME is theoretically inspired: when planets form in a typical proto- Neptune 17 30 0.30 72 Vivid blue colour — cold Neptune-like, but orbiting close planetary disk of gas and dust, 10 ME is roughly Gliese 436 b 23 0.03 0.32 712 the critical mass above which hydrogen can be around a red dwarf accreted and retained by the growing planet, Hydrogen, sodium and possibly HD 209458 b 219 0.05 0.07 1,600 turning it into a Neptune-like giant. Given the water vapour diversity of models for predicting the composi- Largest known planet in size — TrES-4 267 0.05 0.04 2,100 tion of the planet interiors at 10 ME, we need a but would float on water large sample of planets and their mean densi- The ‘Great Red Spot’ — Jupiter 318 5.2 0.24 165 ties to check this. a violent, centuries-old storm Efficient heat transfer in a Why do astronomers expect a great HD 189733 b 366 0.03 0.18 1,200 diversity among smaller planets? turbulent atmosphere Most massive planet known — Because smaller, solid planets can have more XO-3 b ~4,400 0.05 0.4–2.0 Unknown varied material compositions. Depending on a planet, or a failed star? how its material compresses, cools and mixes, Figure 3 | A world of worlds. Our own Solar System is home to an amazing variety of planets and what it weighs, a planet’s interior can (examples in red), from the small, rocky sisters Earth and Venus to the gas giants Neptune and, come to be dominated by an iron or iron-alloy largest of all, Jupiter. But that is nothing compared with the diversity that is emerging among core, a silicate mantle, (mostly water) ice, or a exoplanets, as this selection of particularly notable examples shows. Limitations on current hydrogen–helium envelope. A rare fifth type observational capabilities favour the discovery of exoplanets that are particularly large (and are of interior material is a carbon-rich mantle. In therefore gas giants), and those orbiting very close to their stars. 1 astronomical unit, au, is the general, we expect two distinct families: ‘ocean Earth–Sun distance; data for exoplanets can by their nature be highly uncertain. planets’, with water making up more than 10% of their mass; and rocky, Earth-like planets of gravitational forces that caused them to cycle, which stabilizes Earth’s climate over long (water makes up only about 0.05% of Earth by migrate from an initial location farther out in timescales, for example), and perhaps let us mass) that may still have oceans (Fig. 2). the protoplanetary disk. The two super-Earths take the first steps in the search for markers of are thus probably ocean planets, as they must biological entities. The essential precondition So should we expect other solar have formed outside the disk’s snow line, where is that we know something of the make-up of systems just as diverse as our own? water freezes and accretes easily; some of this the planet in question — that is, its mass and Yes indeed — there’s no reason to think that surface water would have liquefied when the radius. Kepler is designed to find Earth-sized our Solar System is special in any way; and planets were pushed closer in. planets by 2014 using transits. Following this equally, we are beginning to realize that there up with spectroscopy will sort out the planets’ is a great diversity of planetary types not rep- What will be the next stage of true nature. resented in our own backyard (Fig. 3). Since exoplanet discovery? early 2007, for example, we have spotted at It will be to match recent observational break- And would another Earth necessarily least three planets orbiting the star Gliese 581. throughs with breakthroughs in our under- be habitable? Two of them are super-Earths, but they do not standing. For that, we need a large enough The first terrestrial planets that we will be able transit, so we know only their minimum pos- sample to encompass the full diversity of plan- to study in detail will be super-Earths that sible mass through the term Mpsin(i) — 5 and ets out there — meaning many more hundreds, transit small stars. We are likely to learn a lot 8 ME — and their orbital distances, which are if not thousands. That will be achieved quickly of geophysics from them, and, who knows, 7% and 25% of the distance from Earth to the in the coming years with results from dedicated they might turn out to have excellent habitable Sun, respectively. Because Gliese 581 is small, space missions such as the French-led COROT potential as well. But ultimately, we’ll have to cool and faint (it is 75 times less luminous than (for ‘COnvection, ROtation and planetary anchor our theoretical models to the only true our Sun), the tiny orbits of these planets mean Transits’), which was launched in December reference — Earth. Exploring planets down to that their surface temperatures might range 2006, and NASA’s Kepler, scheduled to launch Earth’s size remains a priority of enormous from that of Venus (around 460 °C) down to in early 2009. In the meantime, with advances resonance, both scientific and emotional. ■ that of Mars (a mean temperature of –55 °C). in remote-sensing techniques, exoplanet sci- Dimitar D. Sasselov is in the Department of One of the planets, Gliese 581 d, might have ence is already shifting from mere discovery Astronomy, Harvard University, Cambridge, a temperature similar to that of Earth’s polar to exploration. As planet hunters hone their Massachusetts 02138, USA, and the Harvard regions. tools to discover smaller and smaller planets, Origins of Life Initiative. more surprises should be in store. e-mail: [email protected] Could any of these planets be habitable? When will we find another Earth? FURTHER READING Probably not: the inner, 5-ME super-Earth is Burrows, A. Nature 433, 261–268 (2005). probably too hot, and the outer one is prob- The goal of characterizing Earth-like plan- Knutson, H. A. et al. Nature 447, 183–186 (2007). ably too cold, although still potentially hab- ets — even true analogues of Earth — is very Selsis, F. et al. Astron. Astrophys. 476, 1373–1387 itable. But because they do not transit, we exciting, and is feasible before 2020. By that (2007). Tinetti, G. et al. Nature 448, 169–171 (2007). have no constraints on their bulk composi- time, spectroscopy at very low resolutions Udry, S. et al. Astron. Astrophys. 469, L43–L47 (2007). tion — except that the existence of a third, could allow us to see signatures of familiar Valencia, D., Sasselov, D. D. & O’Connell, R. J. Neptune-mass planet implies the presence global planetary cycles (the carbonate–silicate Astrophys. J. 665, 1413–1420 (2007). 31