Exoplanets and how to find them
- Extra Reading
Each week new observations add to the tally of 'hot-Jupiters' and 'super-Earths' found in orbit around nearby stars. Other planetary systems appear commonplace in our Milky Way, but how closely do they resemble our own Solar System, and what are the implications for life elsewhere?
6 March 2013
Exoplanets and Where to Find Them
Professor Carolin Crawford
Exoplanetis the term used for a planet that’s not part of our own Solar System, but one that is instead in orbit around a star other than our Sun. It’s not a term that was used much – even by astronomers – twenty years ago; today it’s a word that is in common parlance. This change reflects how we have moved rapidly into an exciting phase of discovering entire new worlds every week. And we are not just finding new planetary systems, but also discovering how different and unfamiliar they are compared to our own. There is no shortage of possible locations for these exoplanets. After all, the stars you see in the night sky are all part of the Milky Way, and many of these are not so different from our Sun; each and every one of these 100 billion stars or so is potential host for exoplanets.
The Galaxy is continually evolving and changing, albeit on the astronomical timescale of millions of years. Within the disc, the spiral arms show where diffuse hydrogen gas clouds have been compressed by density waves, triggering the process of gravitational collapse that leads to the formation of stars. The young stars are most clearly seen where they are grouped together to form bright open clusters that line the spiral arms. They are often accompanied by the remnants of their natal cloud which surround the cluster, and glow a characteristic bright pink colour due to the hydrogen atoms they contain being warmed and excited by the energy emitted by the hot young stars.
The Proplyds of the Orion Nebula
The Great Nebula in Orion’s sword is (at a distance of 1,500 light-years from Earth) one of the nearest regions of active star formation, where thousands of young stars – most of which are less than a million years old – are buried deep in the swirls of dusty gas. In 1994 high-resolution images of the core of the nebula taken with the Hubble Space Telescope showed that some of the newly formed young stars were still each embedded in a thick cocoon of gas and dust known as a protoplanetary disc, shortened to proplyd. Such proplyds have diameters over a thousand times the Earth-Sun distance (1 astronomical unit, or AU), and can either show up as dark obscuring tori ringing a protostar, or as a bright enveloping cloud when externally illuminated by the light of nearby bright stars. They are found around many of the stars forming in the Orion nebula, and are the precursor of any planetary systems that will eventually form around these stars. As such, they were the first real indication that planets around stars other than the Sun could be widespread in the Galaxy.
An individual star forms from a small pocket of gas contained within a diffuse cloud. Slightly colder and denser than its surroundings, it collapses under self-gravity slightly more readily than the rest of its immediate environment. The proplyd is the remainder of that pocket, comprised of all the material that didn’t manage to accrete to the centre of the cocoon before it reached temperatures of millions of degrees, sufficient for nuclear fusion to begin at the core – at which point a star is officially born. Any inherent rotation flattens the disc, which surrounds the proto-star in a doughnut shape. The gas and dust within the disc have been compressed, rendering it opaque to external view. Over the next few hundred million years or so, it is this raw material that collects and settles to form a planetary system … if it can survive being burnt away the harsh winds and radiation from nearby stars in the cluster! Meanwhile the protostar at the centre of this cocoon itself is gradually evolving from being powered by the release of gravitational energy to the main sequence process of hydrogen fusion, and will at first pass through a fairly unstable phase (when it is often known as T-Tauri stars, after the prototype) with a very active and variable output luminosity.
Most of the mass of the original gas pocket contracts into the central condensation that becomes the star, with only a few percent remaining to form the proplyd. But as both have formed from the same small cloud, they will have the same elemental composition: around 75% hydrogen, around 24% helium, with a smattering of other elements. These heavier elements are mostly locked into compounds with the hydrogen or with each other, making up the common ices (such as water and methane) and the silicates and carbonates in the dust grains spread throughout the disc. Only the chemically unreactive ones, such as neon and argon, remain as gases.
The Nebular Hypothesis of Planetary System Formation
Our own Solar System has a fairly clear pattern in the distribution of the planets. The inner (terrestrial) planets (Mercury, Venus, Earth and Mars)are all relatively small in both mass and size and made primarily of rock and metal. The gas giants (Jupiter, Saturn, Uranus and Neptune) orbit much further from the Sun, and have enormously deep gaseous envelope surrounding a small solid rocky core. Our understanding of how planets form originally grew from the need to be able to reproduce something similar to the only planetary system we knew about, our own.
The planets build up within the disc in a slow process of accretion. Only about 1% of the mass in the cloud is in the form of particles of dust, but these are essential as they act as nuclei around which more volatile compounds can condense. When the star forms it heats the contents of the disc, giving rise to a strong radial gradient of both the temperature and the composition of the proplyd. Close to the star, the strong UV radiation breaks apart molecules in the inner part of the disc, leaving their lighter, constituent parts to be blown to the outer cooler regions of the disc by the stellar winds and radiation. Any dust grains are also completely evaporated very close to the star. Slightly further out there is a region where the most condensable compounds - such as metals and silicates – start to dominate the disc content. We encounter an important boundary (at about 4 AU in our Solar System) known as the snow line. Here the temperatures falls sufficiently that water and the other ices can condense and add to the mass of the dust grains. This temperature gradient means that it’s the central, warmer parts of the disc that condense to form the smaller rocky/metal terrestrial planets; further out these rocky cores still form but are then in position to accrete icy and gaseous material to grow rapidly in mass to form gas giants.
In the inner parts of the disc, the dust particles undergo low velocity collisions, allowing them to start to stick together, first through electrostatic interactions and then as they grow in mass, through gravitational pull. They coagulate to form increasingly larger (km-sized) bodies known as planetismals, which in turn build up through collisions to make iron-silicate proto-planets with masses up to about that of the Earth.A similar process continues beyond the snow line, but with the added abundance of ices, the rocky protoplanets can build to higher mass (around 10 Earth masses). At this mass they can gravitationally capture the hydrogen and helium gas that makes up the bulk of the proplyd, enabling them to build up the large atmospheres of a gas giant… as well as a few more icy/rocky planetismals that might stray too close. The process of gas accretion – and thus planetary formation – is only halted when the light-weight gas in the disc is blasted out and dissipated into space during the T-Tauri phase of the central star. In the outer (cooler and less dense) parts of the disc, the accretion of gas onto the rocky core is much slower – Uranus and Neptune thus weren’t able to accrete as much matter as Jupiter and Saturn before the opportunity to do so vanished.
During this process, the disc evolves beyond the proplyd phase. As the planets grow, they carve out gaps in the material in the disc, and between this and the stellar activity the disc grows a large doughnut-like hole at its centre. The star is no longer obscured, and has settled to be a young main sequence star. Gradually the surrounding disc has become much wider, flatter, and comprised predominantly of dust grains with little gas.
Huge dusty discs are seen around nearby stars with typical ages of around ten million years. The way that the starlight is reflected by material in the disc can be analysed with polarising filters to deduce information about the size, shape, and other physical properties of the dust particles. They are found to be in the form of tiny and fragile grains, easily destroyed by collisions or radiation pressure and thus not expected to survive for longer than a few thousands of years. The material of the debris discs thus has to be continually replenished, and so the dust is thought to be generated as the debris from the collisions of the planetismals/protoplanetary bodies, still undergoing consolidation to make the final terrestrial-type planets. The stellar wind then disperses these dust particles well beyond where the collisions are taking place and the planets are coalescing, right to the outer reaches of any planetary system.
Directly detecting the faint light reflected by debris discs against the light of the hot and bright central star is difficult; either one reduces the contrast by observing at infra-red wavelengths (where the star is less luminous and the disc is an emitter), or one masks out the glare of the star through the technique of ‘coronography’.
HD 15115 is an example of a relatively nearby star with a debris disc. It is much younger than our Sun, and slightly more luminous. It has a wide, flat debris disc tipped nearly edge-on to our line of view. The actual planets creating the dust supplying the ring can’t be seen as they are too close to the star. The debris disc does, however, appear lopsided and misshapen, suggesting that material within it is being pulled into structures under the influence of the gravitational pull of the young planets forming further in.
The bright southern star Fomalhaut lies 25 light-years distant and is also surrounded by a debris disc - but one that is particularly immense, spanning an annulus from about 24 to 30 billion km around the star. This is a case, however, where we can observe not just the debris disc but also a newly-formed planet around the same star. The exoplanet Fomalhaut b is a tiny point source of light seen at about 3 billion km inside the disc, and its gravity is probably shaping the sharp inner edge of the disc. The planet has a mass of no more than 3 times that of Jupiter, and takes 2,000 years to follow a highly elliptical orbit around its parent star, taking it on a path between orbital extremes of 7.4 and 43 billion km out. Projections of its current trajectory suggest it will encounter the debris disc in about twenty years’ time; if its orbit lies in the same plane as the disc, this could prove potentially destructive.
β Pic is a young star 50 light-years away which is another good example of a system containing both at least one planet and a debris disc of cool matter. The disc is seen edge-on, and extends some 3,000 AU across. The distribution of its material appears asymmetrical: rings mark structures with different concentrations at 500-800 AU away from the star, and the inner parts of the disc are inclined at about 5° relative to the outer regions. When first discovered, the disturbances (particularly the clear dust-free gap) were attributed to the presence of one or more exoplanets in the system, and in 2008 infrared images finally detected a point source in this clear zone. Over time this was observed to move relative to the disc, but aligned within it at a projected distance of 8-15 AU from its star (comparable to the orbit of Saturn in our solar system). This has been interpreted as a giant planet orbiting β Pic, with a mass around 8 times that of Jupiter and taking 17-35 years to complete an orbit.
The detection of a debris disc can thus be used as an indicator for the presence of one or more expolanets in orbit around the same central star, particularly if its shows structures within the disc. The exoplanet(s) help clear the material in a region around their orbit – either by gravitational capture of the matter, enabling it to grow in mass, or by gravitationally perturbing the orbits of particles in the disc so that they migrate away from the region in which they were formed.
Detection of Exoplanets
The only way that astronomers can study any distant cosmic object is through the photons of light that happen to travel our way. Exoplanets present a particularly difficult challenge in this regard, as we are searching for small, inherently cold and dim objects in orbit around a very bright star.
Direct Detection by Imaging – reflected and emitted light
Although one of the most straightforward means at our disposal, direct imaging of nearby stars is not a particularly effective way to discover exoplanets. Planets only shine at visible wavelengths by reflecting the light from their host star; the contrast in brightness between the star and the planet is huge, from a million to 10 billion times different from each other. In addition, planets are generally located extremely close to their parent stars, so the faint reflected light of the planet is completely lost in the glare of the starlight. They are almost impossible to separate, even with some of today’s most powerful telescopes - think of how it may be easy to detect the light of a single firefly on a dark night, but not if that firefly is just sitting right next to the powerful beam of a car’s headlight! Long integration times are required, using the largest telescopes with good, stable and adaptive optics, and high spatial resolution. It is not enough to be able to just collect sufficient light to detect the planet in an image, but the image has to resolve the star and its planet as two separate objects.
Although planets only shine in the visible by reflected light, their temperature (typically a few hundred degrees K) means that the radiation they do emit emerges most strongly in the mid-infrared waveband. It is thus possible to reduce the contrast between a star and any orbiting exoplanet(s) by observing in the infrared rather than the optical. The contrast drops to a million to one or better, particularly if you also target the dimmer host stars, such as brown dwarfs. The drawback is that the ability of telescopes to resolve two sources spatially is worse in the infrared; the minimum separation angle of two objects that can be resolved is proportional to the wavelength, so any telescope will have poorer spatial resolution in the infrared than in the visible. So while successful imaging of exoplanets around nearby stars has been achieved, it is biased to detecting the larger exoplanets (stronger emitters and larger reflectors) that orbit a long way out from their host stars, at distances typically several thousand times further than the bulk of exoplanets found by other means. The matter of the angular resolution also means that its use is restricted to finding such planets around nearby rather than more distant stars.
The first proper image of an exoplanet was obtained in 2004 in the infrared waveband. It was discovered as a companion to the brown dwarf star 2M1207, a star only 0.2% as bright as the Sun, and itself only about 25 Jupiter masses. The fact that it appears only about a hundred times brighter than its planet made the discovery comparatively easy. The planet has a mass of at least 4 times that of Jupiter, and a radius around 1.5 times that of Jupiter; it orbits its host at a distance of about 50 AU (just slightly further from the brown dwarf as Pluto is from the Sun) with an estimated orbital period of 2,500 years.
HR8799 Planetary System
Coronography can be used to cover up the glare of the parent star, and when used with along high spatial resolution imaging (eg with adaptive optics) it is possible to separate the light from the planet from that of the host. This technique was used to obtain the first image of a multiplanet system in 2007. Four planets are seen to orbit HR 8799, an ordinary star fairly similar to our Sun and some 130 light-years away. It’s slightly hotter, brighter and more massive; but with an age of only 30 million years it is much younger, and is still surrounded by a debris disc that lies just beyond the outermost planet. The planets have masses that are approximately between 7 and 10 times that of Jupiter, and of comparable size. They have orbital radii of tens of AU, and orbital periods of hundreds of years.
Detection by Gravitational Microlensing
The gravity of a large mass curves the shape of space around it, which can cause refraction (ie bending) of the path of light as it travels through the vicinity. This is known as gravitational lensing, and is most apparent in the creation of the arced mirages of distant galaxies when they are viewed through a massive cluster of galaxies. Even the gravity from the mass of an individual star will produce this refraction; after all, observations of the minute displacement of the positions of stars around the Sun during a total solar eclipse led to the worldwide acceptance of Einstein’s theory of relativity.
The lensing produced by a star produces too small a distortion of the image of any background source to be resolved spatially. However, a further consequence of gravitational lensing is a focussing effect; more rays are directed towards the observer than there would otherwise be, which amplifies the brightness of the background source over its normal appearance. This can be observed as a microlensing event when one star passes directly in front of another, causing a temporary brightening of the background source. As both the foreground and background stars are moving relative to each other, their mutual alignment is always changing; the brightness will show a smooth rise and fall over a period ranging from a day or so to a month or two.
If the lensing star is accompanied by a planet in orbit around it, the additional mass of the planet will influence the lensing event in a predictable way. It makes a smaller – but still detectable –contribution to the brightening effect on the background star, which is revealed as a characteristic deviation from the brightness increase that would be expected if due to the foreground star alone. A sudden spiked increase in brightness is now superimposed on the underlying variation in intensity. A comparison of the observed light curve to theoretical models of microlensing can be used to find the physical parameters of the system, such as the relative masses of the foreground planet and star, the mass of the planet and its orbital distance.
In practice, this method is very limited. Only a very precise, and incredibly rare, geometry of alignment will produce an observable effect from an exoplanet; indeed, the lack of any addition spike in the light curve of a microlensing event does not necessarily mean that there is no planet present, but just that the exoplanet was not in the right place in its orbit to contribute to the event. The rarity means that to observe any exoplanet events one has to continuously monitor a very large number of stars, and with instrumentation that can record very subtle changes in the brightness of a star. The biggest problem is that each microlensing event is produced by a unique, one-off alignment of the planet, stars and Earth. As these are all moving with respect to each other, the alignment happens only briefly with no chance of a repeat ‘follow-up’ observation to provide confirmation of the result.
So microlensing has not proved an efficient way to detect exoplanets, although at least 12 have been found by this technique. There is a slight bias towards finding the more massive exoplanets (which will produce the greater microlensing effect), but this is less than in other exoplanet detection techniques. The probability of the two stars (lensing and lensed) being along the same line of sight increases with distance, so there is also a bias to finding remoter exoplanets. Despite these difficulties, microlensing still has the potential to discover types of exoplanets that the other methods (that we’re just about to discuss) would miss, such as a lower-mass planet in a wide orbit.
OGLE 2003-BLG-235 is an example of an exoplanet about 1.5-2.5 times the mass of Jupiter, in an orbit 5 AU from its star, some 17,000 light years away from us. It was found from a microlensing event that lasted about 80 days.
Detection of Planets by Gravitational ‘wobble’ – radial velocity method
The simplified picture of planets in orbit around a star is not entirely correct. The star also follows its own small orbit in response to the planet’s gravity. This tiny ‘wobble’ movement is far too tiny to be seen directly as changes in the position of the star, even with the largest telescopes. But it can be detected through the Doppler effect, where features in the star’s spectrum are shifted to red or blue wavelengths according to the relative motion of the star either towards or away from us (its radial velocity).
Two objects of equal mass form a very symmetric orbital situation, moving equally around the midpoint of a line joining their centres; this is seen in the motion of similar mass double systems such as Pluto and its largest moon Charon. Both objects move in orbits around the centre of gravity of the system. As the ratio of the masses of the objects changes, the centre of the orbits moves away from the midpoint and towards the more massive object. The centre of gravity, and the orbital motions, is thus much closer to a star than it is to its planet. In our own solar system, the centre of gravity between the Sun and Jupiter (alone) lies only just outside the Sun’s surface, with Jupiter producing a pull of about ± 13 m/s on the Sun. The real situation is, of course, complicated by the tugs from all the other planets present in the Solar System; on its own the Earth pulls the sun around at only 1/10th of walking speed.
This technique has been one of the most productive ways to discover exoplanets, although it is most sensitive to finding very massive planets in orbits very close to their host. Although it can in principle be used on stars at any distance, in practice the need for high precision spectral measurements limits its use to relatively nearby stars, out to about 160 light-years away from Earth. In the same way that the speed with which the Earth orbits the Sun depends on the mass of the Sun and their separation, the amount of Doppler shift invoked by an exoplanet in its host can reveal how massive it is, and how far from its star. A ‘wobble’ detection of an exoplanet can thus determine: the relative mass of the planet to the star; the distance between them; and the period of its orbit (from the rate of repetition of the signal). Realistically, as we can’t know the exact angle of the exoplanet’s orbit to our line of sight, the mass obtained is only an estimate. The radial velocity we measure through the Doppler effect is only the horizontal component of the true orbital motion of the star; if it were inclined by 45 ° to the line of sight, then its true orbital velocity would be 44% higher than that measured, and its true mass would also be 44%. On average, this effect means we can expect to underestimate the real radial velocity and mass of the exoplanet by about 25%. And the more planets in orbit around a star, the trickier it is to isolate the parameters of each individual object.
51 Pegasi b
The first exoplanet ever discovered around a normal star was found using this method in 1995 by Michel Mayor & Didier Queloz. The host star is 51 Pegasi, which lies some 50 light-years away from Earth. The big surprise was how different the planet was from anything in our Solar System. From the amount of gravitational pull the planet produced on the star, we can deduce its mass to be between 0.5-1 that of Jupiter, making it a gas giant. The problem was that the variations in the host star’s radial velocity took place only every 4.3 days, meaning that it orbits its host enormously fast and so (from Kepler’s laws of planetary motion) it must lie extraordinarily close to the star. Jupiter takes 12 years to make a single orbit of the Sun at a radius of 750 million km; by comparison, 51 Peg b must be some fifty times further in, at an inferred distance of only 16 million km from its star. This means it is a very different world from Jupiter, and incredibly hot, with temperatures at its cloud tops of around 1,200°C. 51 Peg b became the prototype for a particular kind of extrasolar planet we now refer to as a Hot Jupiter.
The very first extra solar planets were also found by a slight variant on this Doppler shift technique.
Although the properties of 51 Peg b were unexpected, they weren’t anywhere near as startling as the very first exoplanets discovered, as they were found round a pulsar. A pulsar is what is left after an extreme stellar explosion known as a supernova destroys a star several times more massive than the Sun. The core of the star collapses down to form an tiny and ultradense neutron star, spinning many times a second, and producing a very regular ‘pulse’ of radiation each time it does so, as jets entrained by its magnetic field sweep across our view. Every time the pulsar spins, it sends out a pulse, and the ticking of this ‘clock’ is accurate to better than one part in 100 thousand million. But if the pulsar is itself under the influence of a gravitational pull, the motion produces an observable change in the timing, or frequency of the pulses as the star moves to and from the observer. Calculations based on the timing of the observed pulses can again reveal the parameters of that orbit. Although the pulsar timing observations were not originally intended to detect planets, the method is so sensitive that it is capable of detecting planets far smaller than through any other technique, and this is indeed the case with the first pulsar planets found. The main drawback of the pulsar-timing method is that pulsars are relatively rare, and it is thus unlikely that a large number of planets will be found this way.
The PSR 1257+12 Planetary System
The very first confirmation of any planets anywhere outside our Solar System came in 1992 using the pulsar timing method, and they were discovered by Alexsander Wolszczan and Dale Frail. PSR 1257+12 is a pulsar about 980 light-years from Earth which rotates around 160 times a second. Deviations in the pulsar timing revealed the presence of two tiny rocky planets, with minimum masses of 3.4 and 2.8 times that of Earth, and lying in nearly circular orbits similar to that of Mercury, at radii of 0.36 and 0.46 AU and with orbital periods of 66.5 and 98.2 day. A third, even lower mass, planet in the same system was found a couple of years later; it has a minimum mass only twice that of the Moon (0.015 Earth masses), and circles the pulsar in a 0.19 AU orbit every 25.3 days.
The very existence of planets around a pulsar was certainly not expected, as it is unclear how they could have survived the supernova explosion that produced the pulsar. It may be safest to assume that they were once Jupiter-sized gas giants whose fragile outer layers were stripped away during the explosion, leaving only their central rocky cores to survive. The alternative is that they could have been formed from the debris left after the supernova explosion. PSR 1257-12 is not the only pulsar found to host planets.
PSR B1620-26 b – the ‘Methuselah’ planet
An even more complicated system is associated with the pulsar PSR B1620-26, located 7000 light-years away in a globular cluster known as Messier 4. The pulsar is itself part of a binary system with a white dwarf star, and a 1.5-3.5 Jupiter-mass planet orbits around the pair of them at a distance of around 23 AU, and with an orbital period of around a century. The stars in the globular cluster are all measured to be around about 12.5 billion years old, and given the expected lifetime of the massive precursor to the pulsar, the exoplanet is expected to be some eight billion years (and 3 times) older than Earth. Again, how such a system came to be is not well understood. The best model so far suggests that the planet originally existed in orbit around the lower-mass star which later burnt itself out to become the white dwarf; then both the white dwarf and its companion exoplanet could later have been captured by the gravity of the neutron star − globular clusters are tightly packed environments, and interactions between the objects in them are likely to be common.
The final – and increasingly most successful – way of detecting exoplanets is through the way that they periodically block the light from their host.
A transit occurs when a planet passes between the observer and its host star, and its silhouette blocks a tiny fraction of the starlight, causing a slight reduction in the overall brightness of the star. We observe transits in our own Solar System, when the two inner planets occasionally can be seen in projection against the Sun’s disc; a transit of Venus decreases the apparent brightness of the Sun by about 0.1% as seen from Earth. An exoplanet can thus be revealed by a repeated pattern of dimming in a star’s brightness, and where each dip in star’s light-curve (the plot of its brightness with time) has a characteristic level-bottomed shape. A minimum of 3 transit events separated by the same time intervals is required to feel confident of a potential detection of an exoplanet, and a better signal can be obtained only with more transits. The first transit dimming caused by an exoplanet was observed in 1999, which was within a planetary system that had already been discovered by the radial velocity method. The first new exoplanets discovered from transits were found in 2002.
A well observed lightcurve generates a lot of information about the planetary system. The orbital period of the planet around the star is revealed from the regular interval between successive transits; from Kepler’s laws of planetary motion this gives you an estimate of the size of its orbit and its speed; and the depth and width of the dip in brightness of the star allows you to estimate the size of the planet relative to that of the star. The time taken for the planet to get fully in front of the star can also inform you about its diameter.
The one thing you can’t estimate from a detection of a transit alone is the mass of the planet. However, if a transiting system also displays an observable Doppler shift signal in its radial velocity, then the combination of both datasets allows the observer to get more physical properties of the system. The size of the planet from the transit, along with the mass from the wobble motion enables an estimate of a planet’s density. In addition, you know this is a good estimate of the mass, as the exoplanet’s orbit must be inclined very close to edge-on for you to be able to see a transit at all. The range of densities inferred is very wide, ranging...
From Corit-3 b...
COROT-3 b is over two and a half times denser than lead; its surface gravity is expected to be over fifty times that felt at on Earth;
...To Tres-4 b
At around 1.7 times the size of Jupiter, Tres-4 b is one of the larger planets found. However, when this is combined with its mass of about 90% that of Jupiter, it is inferred to have the lowest known density – at 0.333 g/cc it is equivalent to the density of cork or balsa wood. Consequently its gravity has only a relatively weak pull on the upper atmosphere, and we expect that much of it will be escaping to both puff up the planet in size, and be dragged around behind the planet almost like a comet’s halo and tail.
The spectrum of the host star during a transit can tell the observer about the relative rotational directions of the star and its planet. The star that is rotating, so the limb that is approaching us contributes a blue-shifted wing to the absorption lines in its spectrum, and the other limb that is receding contributes a redshifted wing; thus the lines in a rotating star appear wider than they would be if it wasn’t rotating. During a transit, when the planet obscures one and then the other limb of the star either the red or blue wing of the absorption line disappears first, depending on the relative direction of rotation. From such observations it appears that exoplanets do not always orbit in the same direction as their host star spins, contrary to what happens in our own Solar System and our theories of planet formation. The planets should inherit the rotation of the original disc they form from; yet several exoplanets are found to be completely retrograde, with orbital motion in the opposite sense to the direction of the rotation of its host star.
Even in the simplest system you have uncertainties in the results introduced by whether or not the exoplanet transits across the centre of the star’s disc, whether the orbit is circular and then whether you know the star’s radius and mass accurately. It becomes more complicated to disentangle the signal if it is due to a multi-planet system. Not only do you have a series of differently timed transits of different amplitudes, but as the planets exert a gravitational pull on each other, they can at times be slightly speeded up or slowed down in their orbit, resulting in the transits occurring very slightly sooner or later than expected! Like the gravitational wobble method, the transit technique is most efficient at discovering the Hot Jupiters - large planets obscure more of the starlight, and if they orbit very close to the host star the period between transits is short.
Planetary transits are only observable for systems where the orbits happen to be perfectly aligned from astronomers' vantage point; assuming random alignments we can expect only one in 200 planetary systems to be at the right angle to show transits. Even the baseline brightness of the star is not necessarily constant: stars can show intrinsic variation in their brightness; and any regions of stellar activity manifested in the form of cooler, darker "starspots" on the surface of the star (similar to the sunspots on our Sun) can also mimic a transit dimming if they are long-lived and repeatedly come into view. Any atmospheric scintillation or ‘twinkling’ caused by turbulence in the atmosphere (see my Atmospheric Phenonena talk for details) can also confuse the signal. The most modern transit projects avoid the problems of geometry and the atmospheric blurring by using dedicated space telescopes to monitor hundreds of thousands of stars at once.
The CoRoT (COnvection ROtation and planetary Transits) is an ESA satellite that houses a 10” telescope that was launched into orbit 2006 by ESA and which remained operational till the end of 2012. Many of the CoRoT discoveries have been Hot Jupiters, although it was also able to find some much smaller systems.
CoRoT-7 b was for a long while the smallest exoplanet known. With a radius only 70% larger than that of the Earth, and a mass around 5 times greater, its average density implies a rocky planet with a metal core. It is in orbit around a star that is only slightly smaller and cooler than our Sun and quite a bit younger, with an age of about 1.5 billion years. But that’s where the similarity ends – CoRoT-7 b orbits its home star 23 times closer than Mercury orbits our Sun, and has a year that lasts only twenty of our hours. The probable temperature on its "day-face" could be as high as 2,500°C, so its surface is most likely molten rock. The data also reveal the presence of a sister planet, with a mass about eight times that of Earth, and which circles in a wider orbit lasting 3 days and 17 hours. Unlike CoRoT-7 b, its orbit doesn’t take it between the star and Earth to create a transit, so we can’t measure its radius and thus its density.
CoRoT-9 b was the first planet found likely to have a temperate climate. It’s much closer to Jupiter in terms of its size and mass, and it follows an orbit slightly larger than that of Mercury, with an orbital period of 95 days. The planet has a density of 0.90 g/cc, and so is mostly likely made of hydrogen and helium; although a gas giant, the temperatures in its upper atmosphere are expected to be between −20 and 160°C.
The Kepler Space Observatory
The NASA/ESA Kepler spacecraft carries a 37” space telescope, which was launched in 2009 and which is still operative. It was designed specifically with the aim of discovering Earth-size planets within the habitability zones of Sun-like stars: the habitability or ‘Goldilocks’ zone around a star marks the range of orbital distance where a planet would be at the right temperature for any water at Earth-like pressures to be liquid. Kepler is continually monitoring the brightness of more than 150,000 pre-selected stars for the signals of transits, and its observations are extending the search radius for exoplanets from hundreds to thousands of light years out into the Galaxy. The telescope is sensitive to changes in brightness to a precision of 0.002% - for comparison, we can expect the transit of an Earth-sized planet around these stars to decrease the star’s brightness by around 0.008-0.009%.
So far Kepler has already discovered thousands of candidate exoplanets from signals that look like planetary transits. These remain only candidates, however, until their detection is confirmed by additional and independent observation, preferably with one of the other techniques (usually studying the star’s radial velocity), so each has to be followed up with ground based telescopes.
HAT-P-7 b and its Modulation
Kepler’s first detection was of the giant extrasolar planet HAT-P-7 b, one already known from previous ground-based observations from the transit method; with an orbital period of only just over two days, its transit signal showed up in only ten days' worth of Kepler data. HAT-P-7 b is comparable to Jupiter in size, and it is in orbit around a star analogous to our Sun. The Kepler lightcurve shows a classic pattern of two eclipses – a primary where the planet blocks the light of the star, and a much smaller secondary eclipse when planet moves behind the star and we lose its contribution to the total light.
There is an even more interesting modulation to its light curve: Kepler detects the light from the exoplanet directly, and is sensitive to the small but regular alteration in brightness of the planet due to changes its phase. Different amounts of light are seen from the planet as it passes from new through to crescent/gibbous before it is occulted by the star – and then in reverse as it emerges again. How much light the planet reflects again depends on how large it is, and how close it is to the star, and the colour of the reflected light allows an estimate of the temperature.
During the transit, some of the stellar radiation will be absorbed and then re-emitted by the planet’s atmosphere, if it has one. While this will have only a negligible effect on the overall light from the star, it may introduce some absorption lines from the atmospheric gases to the spectrum of the star. One can also isolate the light of the planet alone by subtracting the light received during the ‘secondary eclipse’ (when the planet is behind its parent star) from the light received when both are visible. This light is again not just starlight by the planet, but some of it is filtered through the planet's atmosphere before re-emission, leaving spectral features stamped on it, introduced by absorption by molecules in its atmosphere. Thus we can determine what mix of gases are present in an exoplanet’s atmosphere, which teaches us something about its chemistry and the physical conditions. Astronomers are particularly interested in detecting the signal of organic molecules, which might trace the presence of habitable atmosphere. Molecules of water vapour, methane and carbon dioxide have all been detected.
HD 189733 b – an evaporating hot Jupiter
HD 189733 b is a Jupiter-sized (only slight more massive and larger) gaseous exoplanet that circles its host at a distance about 30 times closer than Earth is from the Sun to complete an orbit every 2.2 days. It is so close and hot that the planet's deep atmosphere reaches a temperature over 1000°C. Hubble Space Telescope observations have shown it to have methane, carbon dioxide, carbon monoxide, and water vapour in its atmosphere from the imprint on the light due to absorption by molecules. However, the spectral observations also give information about the motion of the gases within the planet's atmosphere, and suggest that matter is continually escaping from the planet's upper atmosphere to evaporate into space at a rate of over 1000 tons of gas/second, and at a speed of around 500,000 km/hour. These are significant changes in its atmosphere of a planet, and could well be triggered as a response to a powerful outburst on the planet's host star.
The Types of Exoplanets detected So Far
Over the last twenty years we have found that our Galaxy is chock-full of planets.
As you might have noticed, exoplanets take the name of their parent star with the addition of a lowercase letter. ‘b’ is used to denote the first planet discovered in any system, ‘c’ the next and so on; if several planets are found in simultaneously in one system, then the convention is that they ordered ‘b’, ‘c’, etc with distance from their star. The letter ‘a’ is not used to avoid the confusion that could possible arise in binary/multiple star systems which potentially also use ‘a’ to denote one of the component stars.
Most of the exoplanets are found around stars that are similar to the Sun, ie main sequence stars of spectral class F, G and K. Not much significance should perhaps be attached to this yet, as it is in part a selection effect, as such stars attracted much of the initial search effort, for obvious reasons. These stars are very suitable for the Doppler shift method as they have plenty of sharp and deep absorption line sin their spectrum.
The vast majority of the planets found so far are the Hot Jupiters, closer to their host stars than the Earth is to the Sun, with very high cloud-top temperatures. This is completely a selection effect – the larger and more massive planets will have produce a larger gravitational tug on their star or block out most of its light, and the initial stages of any search will begin with picking up those objects that orbit most rapidly round their star. Extreme ‘super-Jupiters’ may reach a mass of 5 times that of Jupiter or even more; but by ~13 Jupiter masses they are regarded more as brown dwarfs, or ‘failed’ stars.
Moving our own Jupiter from its usual location to instead within Mercury’s orbit about the Sun would dramatically increase the amount of energy it receives by a factor of a few thousand. This would raise the cloud-top temperatures from about −150°C to +750°C; in addition the tidal gravitational forces in an orbit so close to the Sun would cause the planet’s rotation to become locked so that it would always keep the same face towards Sun, leading to extreme differences of temperature from one side of the planet to another. Along with the contribution of any internal heating from the gravitational tidal stretching and squeezing of the planet core, we might expect colossal winds racing round the cloud tops at thousands of km/hour – far higher than the hundreds of km/hour out in the cold version of Jupiter. The heating might cause the planet’s radius to expand by about 20-40%, in turn reducing the average density. The planet would still have enough gravity to retain its atmosphere – only if it orbits incredibly close to its host star it run the risk of having the gas blown away into space to leave only the rocky core.
An extreme case of this is WASP-17 b, one of the largest exoplanets known: about twice the size of Jupiter, yet with only about half its mass. It is a light-weight planet (its density is similar to that of expanded polystyrene foam) and it orbits a star hotter than the Sun every 3.7 days at an average distance of 0.05 AU. It follows quite an elliptical orbit and as a consequence is subjected to intense gravitational tides that heat up its interior. Along with the proximity to the star this results in a bloated planet. WASP-17 b was the first exoplanet discovered to have a retrograde orbit.
‘Hot Neptunes’ are simply gas giants which are lower-mass versions of Hot Jupiters: the transition mass is around 20% that of Jupiter, and they range down to 2%, which is still sufficient to retain a substantial hydrogen atmosphere.
Many of the planets found, however, bridge a gap that is not found in our Solar System. These ‘super Earths’ are lighter than the gas giants like Neptune, but up to ten times more massive than Earth. They could be gaseous, although the smaller ones are more likely to be rocky in composition; whether or not such an object has an atmosphere will depend on its temperature (ie proximity to its host star) and its evolutionary history.
Gliese 667C c
With a mass ~ 4.5 times that of Earth, Gliese 667C c is a super-Earth that takes 28 days to complete a single orbit of its star. As its star is much cooler than the Sun, the exoplanet probably absorbs about as much energy from its star as Earth does, and estimates of its surface temperature suggest it could be similar to Earth’s.
HD85512 b is another super Earth (its mass is over 3.6 times that of Earth) that orbits on the inner margins of its star’s habitable zone. As in all these examples, the surface temperature will greatly depend on how much of the incident energy is reflected by cloud cover, but it could be comparable to Earth.
Gliese 163 c
One of two inner planets in orbit around a red dwarf star some 40 light-years distant, Gliese 163 c has a mass of 6.9 times that of Earth and an orbital period of 26 days. Situated right at the inner edge of its star's habitable zone, it receives on average 40% more light from its parent star than Earth from the Sun, making it only slightly hotter (for comparison, Venus receives 90% more Sunlight than Earth…).
Not all super-Earths are anywhere near that temperate. 55 Cancri e is a super-Earth that is the innermost planet of a system of five planets, lying about 25 times closer to its star than Mercury is to the Sun; it is so close that it should be tidally locked with one face forever towards the heat of its sun. It is only 1.6 times as big and eight times as massive as Earth, and its density is about twice that of Earth, suggesting it is rocky. The drop in total brightness observed when it passes behind its host star suggests that the sun-facing side is a blistering 1,700°C.
Many of the planets discovered lie in multiple planetary systems (with 127 having been discovered to date).
The Sunlike star Kepler-11 lies 2,000 light-years away, and has six worlds in orbit around it: all are larger than the Earth, with the largest about Neptune’s size, and their orbital periods of 10-47 days mean that they are all closer to their parent star than Mercury is to the Sun.
Planets around Double Stars
Many stars live in binary systems, but that does not seem to preclude them still hosting exoplanets. The frequent transits and eclipses between both the planet and the two stars can often lead to a very good determination of the physical parameters of the entire system.
Kepler-34, Kepler-35 and Kepler-16
For example the Kepler mission has discovered several cases of Saturn-sized gas giants that orbit around binary star systems. Kepler-34 b orbits its two Sun-like stars every 289 days, while the two stars themselves orbit and eclipse each other every 28 days; similarly Kepler-35 b revolves about a pair of smaller stars (80 and 89 percent of the Sun’s mass) every 131 days, where the stars then orbit and eclipse one another every 21 days. At distances of 4,900 light-years (Kepler-34) and 5,400 light-years from Earth (Kepler-35), these are among some of the most distant exoplanets discovered. Kepler-16 b is a similar-sized planet, occupying Venus-like orbit at the outer edge of the habitable zone around another binary star system. In each of these systems, the distances between the planet and stars will be continually changing due to their orbital motion, dramatically varying the kind and amount of sunlight the planet receives. It’s possible that the planets have really crazy climates not found elsewhere, perhaps undergoing a whole range of rapid seasonal cycles and huge temperature changes, along with the associated effects of such a dramatic climate on the atmospheric dynamics,
…and even quadruple stars.
PH1 /Kepler-74 b
The first exoplanet discovered by amateurs as part of the ‘Planet Hunters’ citizen science project (www.planethunters.org) was found to lie in a quadruple star system. A gas giant, just slightly larger than Neptune and about 170 times more massive than the Earth, it completes an orbit every 138 day around two stars (1.5 and 0.4 times the mass of our Sun) that revolve each other every 20 days. A second binary pair of stars then orbit around the whole system well beyond the planet’s orbit. It is amazing to consider how such a planet can remain in any stable orbit without having long since been pulled apart by the combined and varying gravitational tidal forces between the four stars!
It is only now, a number of years into the Kepler mission that we are systematically beginning to find greater numbers of exoplanet candidates which are smaller in size and mass. Extrapolations from the first detections suggest that lightweight, rocky planets will eventually be found in much greater numbers than the gas giants.
Just last month (Feb 2013), the discovery was announced of the tiniest exoplanet yet found around a normal star (ie not a pulsar). Kepler-37 b is smaller than Mercury, and only slightly larger than our Moon. It orbits its star Kepler-37 in just 13 days, so even though it will be a rocky planet, it will be blisteringly hot on its surface. It’s part of a system with two other planets, one three-quarters Earth's size, and one twice as large as Earth, which orbit their star in 21, and 40 days respectively − all the orbits are within just 20% of the Earth-Sun distance.
Ultimately, we want to find planets that are Earth-like – finding not just Earth-mass and Earth-size planets, but ones that lie in the habitable zone of stars that resemble the Sun; planets where liquid water and possibly life might exist.
The best technique for finding them in any sufficient number is through transit dimming, but even with Kepler this is still challenge. To quote one of the scientists engaged on the project: “Trying to a Jupiter-sized planet crossing in front of its star is akin to measuring the effect of a mosquito flying by a car’s bright headlight. Finding Earth-sized planets is like trying to detect a very tiny flea in front of that same headlight.”
Four super Earths orbit the low-mass red dwarf Gleise 581 20 light-years from Earth. Planets ‘e’ and ‘b’ lie outside the habitable zone (with masses 2 and 16 times that of Earth, respectively), which lies relatively close in around such a cool star. Planet ‘c’ (5 Earth masses) occupies the warm inner edge of the habitable zone, and is probably too hot to sustain water; however ‘d’ (7 earth masses, and twice Earth size) lies well within the habitable zone.
600 light-years away, the planet Kepler-22 b is about 2.4 times the diameter of Earth, and lies 15% closer to its parent star than the Earth is to our Sun, completing an orbit in 290 days. As Kepler-22 is slightly cooler and smaller than the Sun, it emits about 25% less light, suggesting that surface temperatures on the planet might around a balmy 22°C. Don’t book your holiday there yet, though, as we have still to determine whether Kepler-22 b is mostly rock, gas or liquid..
Whilst only a potential planet detection awaiting confirmation, the most recently announced ‘twin’ to Earth is KOI 172.02, anticipated to be only half as big again as the Earth, and in orbit in the habitable zone around a G-type star that is only slightly cooler than our own Sun. It’s the best candidate yet for an Earth-like world beyond our solar system, and one that may well have a similar amount of gravity as Earth and with liquid water.
revisiting the process of planetary formation
Just how many planets are there out there?
Scaling from just our own solar system you might naively expect a ratio of 8 planets around each star, giving a ballpark figure of some 800 billion planets in the Milky Way. Of course, many of the massive stars in our Galaxy are too short-lived to ever host planetary systems, or may live in too crowded environments. But initial results from the exoplanet surveys (such as those using Kepler) agree that planets are indeed commonplace, with the average number of planets around a star expected to be greater than one. The most recent estimates are that: one in six stars hosts a planet the size of Jupiter, half have Neptune-like planets and two thirds have super-Earths. The statistics are much vaguer on how many Earth-like planets there are, but recent computer models indicate that at least one in ten stars are orbited by an Earth-sized planet – though the number of Sun-like stars hosting properly Earth-like planets in Earth-like orbits is surely much smaller. What is clear, however, is that our own Solar System is not obviously the norm; exoplanets such as Hot Jupiters present particular difficulties of understanding the process of planetary formation.
Migration of Hot Jupiters
Gas giants have to be formed out beyond the snow line, otherwise it would be too hot for the volatile materials needed to build gas giant planets to condense. This is, of course, much further out than many Hot Jupiters are observed to lie around their star, so an additional evolutionary process is required within a newly formed exoplanetary system to shrink the orbits of newly-formed cold Jupiters.
A planet forms and grow while still it is embedded in a gaseous protoplanetary disc. But as the mass of the giant planet grows, its gravitational field will set up disturbances that bunch the disc material nearby into asymmetric spiral structures around the protoplanet. These in turn exert a gravitational influence on the planet, leading to a net exchange of angular momentum: the interior spiral structure tends to try to push the young planet outwards, while the exterior structure has an inwards, a slightly stronger, push. The result is a net inwards migration, at a rate depends on the density of material in the disc and the mass of the young planet. When the planet has accreted sufficient mass that it opens up a gap in the disc around it, this migration will slow dramatically, as does the accretion of material from the disc to the planet. The planet formation only really comes to a halt when all the disc material is removed by through the outflowing stellar wind that pushes the matter in the disc out into space. This migration takes place over a few million years, and will results in a Hot Jupiter, but one that is an orbit still aligned with the rotation axis of the parent star. It is also possible that such migrating giants may inhibit the growth of any terrestrial objects interior to the snow line unless the rocky planets form later.
If there is more than one gas giant out beyond the snow line, they can each gravitationally interact not just with the disc, but also between each other in a more gradual and cumulative tug-of-war − or even feel the influence of nearby stellar companions. Such interactions can produce highly eccentric orbits over hundreds of millions of years. Close encounters also have the potential to be far more destructive: one partner could be scattered out into interstellar space and other retained only in a high eccentricity orbit. This is a slower evolution process, but this kind of migration seems to be required to account for the Hot Jupiters observed to lie in retrograde orbits.
Habitability and Comets
An Earth-like rocky planet might still not be habitable – even if it is located in the habitable zone around its star, it may have none of volatiles (including water and carbon compounds) needed to form an atmosphere and liquid water at the surface. We suspect that much of the Earth’s water was delivered from the outskirts of the Solar System by comets and other volatile-rich bodies in the late-heavy bombardment period. Such an event may not occur for any given planet – whether such volatiles are available and deliverable in other planetary systems may be very fragile balance that depends not just on the presence of a cometary cloud, but also whether or not there are cold giant planets between the comet cloud and the Earth-like planet. In our own Solar System Jupiter’s gravity has deflected much debris, and shielded us from an excessive number of impacts; but it could be that some gravitational interplay between gas giant and a once more highly populated Kuiper Belt is needed to provide the deluge of comets that delivered our oceans. Without giant planets to fulfil this role, the evolution of a rocky planet may be a very different story.
GJ 581 and 61 Vir
Infra-red observations have revealed the presence of vast comet belts surrounding at least two nearby planetary systems, GJ 591 and 61 Vir, which are both known to host only super-Earths. The observations detected the signatures of cold dust at −200°C, in quantities that mean these systems must have at least ten times more comets than are found in our own Solar System’s Kuiper Belt. This contrast is suggestive of a correlation: in the Solar System we have giant planets and a relatively sparse Kuiper Belt, but both these systems have the rich dense Kuiper belts but only have low-mass planets. It seems the absence of a cold Jupiter has allowed them to avoid a dramatic heavy bombardment event in their evolution; perhaps they instead experience a much more gradual and continual rain of comets over billions of years.
But the one thing that all the range of planets I’ve talked about so far have in common is that they are all found in orbit around a star. There is further population of planets expected that are ‘rogue’, floating freely through space and unattached to any star. There are at least two possible ways to create a planet that is free of a host’s gravity. They could coalesce from the protoplanetary disc exactly alongside the planets that remain bound to stars, but they have been ejected from the system, perhaps through the inter-planet gravitational interactions that allow cold Jupiters to migrate in. Alternatively, they could be ‘failed’ stars, forming in the same way as stars, but never reaching sufficient mass to spark nuclear fusion; at the top end of the mass range these could be considered as brown dwarfs, although some super-Jupiters could also be expected. Any rogue planets are without an external source of energy input, and can only keep warm by retaining the internal energy from their initial collapse. This might sustain them for a few to tens of millions of years, depending on whether or not they have an insulating atmosphere. Given that most planets we know of are found through the effects they have on their host star's light, pinning down rogue planets has proven incredibly difficult… but not completely impossible.
One international project recently surveyed hundreds of millions of stars and planets in the near-infrared, looking for rogue planets; the result was a single candidate about 100 light-years away. CFBDSIR2149-0403 appears to be moving alongside a group of about 30 stars which are all of roughly the same composition, and believed to have formed from the same gas cloud at about the same time. The physical association suggests that CFBDSIR2149-0403 too formed with the stars, about 50-120 million years ago; but with a mass between four and seven times that of Jupiter, it falls well short of the mass limit to be even a brown dwarf.
And the rest of the story?
It is still early days for exoplanet science, with much remaining yet to be discovered and investigated. For example, how do the different kinds of planetary systems and their formation processes depend on the host star’s mass or metallicity? How do we characterise the Earth-like planets and their atmospheres to assess whether they are habitable?The idea of discovering a twin of Earth is no longer an elusive concept, but is quickly becoming an observable reality. Understanding ‘twin’ Earths and super-Earths and their probable evolution will require an interdisciplinary approach that combines astronomy, chemistry, geophysics and biology, and will put our understanding of the formation, evolution of our own Earth and Solar System into a new context. The operation of the Kepler spacecraft has been extended to 2016, and it will continue to amass thousands of candidates for follow-up study, and candidates of increasingly a wider variety of physical situations. And after Kepler, the NASA James Webb Space Telescope will feature an even more advanced suite of scientific instruments ideal for continuing this study. It seems that the study of many more, and very very different, worlds will soon be within our grasp.
© Professor Carolin Crawford 2013
This event was on Wed, 06 Mar 2013
Gresham College has offered an outstanding education to the public free of charge for over 400 years. Today, Gresham plays an important role in fostering a love of learning and a greater understanding of ourselves and the world around us. Your donation will help to widen our reach and to broaden our audience, allowing more people to benefit from a high-quality education from some of the brightest minds.