Detection of Extrasolar Planets

Directly observing extrasolar planets is extremely difficult given the large brightness contrast between a star and its planets and also the small angular separation. For example, if our own Solar System were observed at a distance of, say, 5 parsecs, the greatest angular separation of the Sun and Jupiter would be just 1 arcsecond with the Sun appearing 109 times brighter at visible wavelengths. Under these conditions it would be impossible to pick Jupiter out from the Sun's glare (Lewis, 2004). One possible solution to this problem is to search for planets around dimmer stars such as white and brown dwarfs. Searches for extrasolar planets around white dwarfs have so far been unsuccessful (e.g. Burleigh et al., 2003; Friedrich et al., 2006), but four planets/brown dwarfs (Sect. 1.4.2) have now been directly imaged about brown dwarfs, the first being imaged by the Very Large Telescope (VLT) orbiting a brown dwarf, situated 200 light years away, at a distance of ~ 60 AU (Chauvin et al., 2005a). Another strategy is to attempt to detect the planet at wavelengths near the peak of the planet's Planck function. Observing at 50 ¡m rather than 0.6 ¡m reduces the flux ratio to 104 for the Sun-Jupiter system, but at these longer wavelengths the diffraction-limited angular resolution of any achievable telescope would be insufficient.

Although direct optical detection of extrasolar planets initially appeared very difficult, it was realised that it might be possible to indirectly detect them through their influence on the motion of the central star. There are two ways of doing this: 1) by observing the radial velocity of the star as the planetary system rotates about its centre-of-mass and 2) by observing the actual reflex motion1 of the star against the heavens (astrometry). In addition, it also came to be realised that there was a chance that an extrasolar planet could be detected if it transited in front of its star, while other detection methods, such as gravitational lensing, revealed themselves serendipitously. There are now numerous methods of detecting extrasolar planets, which will be briefly summarised.

1.2.1 Radial Velocity Detections

For a planet of mass Mp in a circular orbit of radius a about a star of mass M*, the star and planet will both orbit about their centre-of-mass, situated at a distance 2aMp/ (Mp + M*) from the star. Equating the gravitational force with the centripetal force acting on the star, and assuming that M* ^ Mp, the maximum velocity of the star v in the line of sight of an observer may be shown to satisfy v2 = G (Mp sin i)2 /2M*a, where i is the inclination of the planet's orbit with respect to the observer, i.e. the angle between the normal to the orbital plane of the planet and the line from the star to the observer on the Earth. The radial velocity method can determine both Mp sin i and also, from the shape of the variation of v with time, the eccentricity, e, of the planet's orbit. It is worth noting that unless the inclination can be determined from other methods such as astrometry (Sect. 1.2.2), this method only provides a lower limit on the planet's mass. The technique is most effective for larger mass planets orbiting close to the lower mass stars (i.e. G and K type) since this gives the largest line-of-sight stellar velocity and it is crucial to

1 The reflex motion of the star is caused by both it and the planet orbiting their common centre of mass.

be able to distinguish the radial velocity of the star due to the orbit of a planet from the naturally occurring turbulent velocities present in a stellar photosphere. An example of a measured radial velocity curve for the star GJ 446 (Butler et al., 2004) is given in Fig. 1.1.

Fig. 1.1. Measured velocities vs orbital phase for GJ 436 (Butler et al., 2004). The dotted line is the radial velocity curve from the best-fit solution: P = 2.644 days, e = 0.12, M sin i = 0.067Mj

Since the discovery of 51 Peg b there have been detections (almost all by the radial velocity technique) of over 200 extrasolar planets. Indeed it is now estimated that more than 6% of sun-like stars have a detectable 'wobble' due to the orbit of at least one Jupiter-mass planet. At the time of writing (28 February 2007), the total number of planets listed in the Extrasolar Planets Encyclopedia ( was 215 in 185 planetary systems (including 21 multiple planet systems). Most of the recent radial velocity planet searches have been able to detect velocity variations as small as 10 m/s (Marcy et al., 2003) and so a Sun-Jupiter system (for which the Sun's radial velocity is 13.2 m/s) should have been just about detectable and, indeed, such planets are now regularly being found. For example (Wittenmyer et al., 2007) report the discovery of 47 UMa c, a planet with mass 1.34 MJ, low eccentricity and an orbital radius a = 7.73 AU. Recent improvements have meant that current observations can now achieve even greater accuracies of 3 m/s and thus the number of planets detectable by this technique is steadily increasing. In addition, the current data sets only last for ~ 10 years. As measurements continue, and the sensitivity improves, the discovery of more Jupiterlike planets orbiting far from their star with longer periods is expected. At the time of writing 26 exoplanets have now been catalogued with an orbital distance greater than 3 AU.

1.2.2 Astrometry

Given a sequence of observations of a star's position of sufficiently high accuracy relative to the celestial sphere, the reflex motion of the star caused by the orbit of a planet around it can be detected. This can be used to determine both the absolute mass and orbital inclination of a planet. Considering the motion of the star and planet about their common centre of mass we can see that the reflex amplitude of the star is a* = apMp/M*, where a* and ap are the distances from the centre-of-mass to the star and planet respectively. Thus, this method is most effective for large mass planets orbiting at some distance from their parent stars. In addition, since what is actually measured is the angular position of the star, the method is clearly best for planetary systems within a few parsecs of the Earth.

The accurate measurement of a star's position over a number of years is a challenging task. Current optical systems have an absolute accuracy of a few milliarc-seconds. However this precision can be improved through the use of long-baseline interferometry. The VLT and Keck currently have programmes to do this and are expected to achieve accuracies of 30 ^as (microarcseconds), which should be sufficient to observe the reflex motion of the stars of several extrasolar giant planets already discovered. In addition, there are two space missions planned to exploit this technique. The NASA SIM (Space Interferometry Mission) is due for launch sometime betwee 2009 and 2015 and will be able to achieve 1 ^as accuracy, while the ESA GAIA spacecraft, which is a follow-up to ESA's Hipparcos mission, is due to launch in 2011. Although not an interferometric instrument, GAIA aims to observe 1 billion stars with magnitude brighter than 20, with an accuracy of 10-20 ^as at magnitude 15.

1.2.3 Transit Detections

For extrasolar planets, there is a small, but finite, chance that the orbital inclination i will be very close to 90° and thus that a planet will periodically pass between the planet's star and the Earth. If the planet is sufficiently large, then the drop of intensity of the starlight can be detected and used to determine both i and also the radius of the planet.

The first published detection of a planetary transit (using the STARE transit camera (Charbonneau et al., 2000)), was of the planet HD 209458 b, which orbits its star at a distance of 0.046 AU in a period of 3.5 days (Henry et al., 2000). The transit was observed the next year with the Hubble Space Telescope (HST) (Fig. 1.2) and Brown et al. (2001) concluded, from the transit depth, that the planet had a radius of 1.35 RJ (where Rj is the radius of Jupiter). This figure has recently been revised to 1.32 RJ (Knutson et al., 2007).

Assuming HD 209458 b to be typical, and until more transits of this type are observed there is no reason to think otherwise, these observations showed that the massive, close-orbiting planets discovered by the radial velocity survey were not just rocky cores, but large Jupiter-sized objects. The radius observed is considerably larger than that expected from a planet cooling in isolation and Burrows et al.


12 0.990


time from center of transit (doys)

Fig. 1.2. HST observation of transit of HD 209458 b (Brown et al., 2001)

(2000) proposed that irradiation from the star inhibits convection and thus cooling/contraction. This idea was developed by Bodenheimer et al. (2001) and Guillot and Showman (2002).

A number of other extrasolar planetary transits have been observed since 1999, using projects such as OGLE (Sect. 1.2.4). Most lead to a dip in intensity of the order of 1%, and at these levels care must be taken to ensure that phenomena such as sunspot variations or isolated or blended eclipsing binary systems are not mistaken for planet detections (e.g. Mandushev et al., 2005; O'Donovan et al., 2006b, 2007).

Transit Spectroscopy

Soon after the first transit of HD 209458 b was observed, it was realised that observations at a number of different wavelengths might be used to infer the atmospheric transmission of the planet's atmosphere, since a planet's effective cross-sectional area will be larger at wavelengths where its atmosphere is more strongly absorbing than at others. Just such a study is reported by Charbonneau et al. (2002) who used HST observations near 600 nm to search for the atmospheric sodium absorption lines predicted for 'hot Jupiters' by radiative transfer models such as Sudarsky et al. (2003). The absorption line was duly detected, the first ever detection of an exoplanetary atmosphere, although the magnitude of the absorption was found to be less than that predicted by cloud-free radiative transfer models suggesting that clouds high in the atmosphere of this planet reduce the absorption band depth. Brown et al. (2002), Richardson et al. (2003a) and Richardson et al. (2003b) extended this campaign to the infrared, searching for CO, H2O and CH4 absorption, and recently Deming et al. (2005) detected a weak absorption due to CO at 4325 cm-1 and also suggested the presence of a high level cloud at, or above, 3.3 mbar.

In addition to direct detection of atmospheric absorption during transits, a gas giant orbiting as close to its star as HD 209458 b will get very hot in its upper atmosphere leading possibly to exospheric loss. Vidal-Madjar et al. (2003) report HST observations of atomic hydrogen absorption of starlight during several transits of HD 209458 b. They interpret this observation as being due to absorption by hydrogen atoms that have exospherically escaped the planet's atmosphere and are now beyond the Hill radius2 of the planet. They further conclude that if the timescale for this evaporation is comparable to the lifetime of the stellar system then it may explain why so few 'hot Jupiters' are found orbiting with periods less than ~ 3 days. More recent HST observations by Vidal-Madjar et al. (2004) have also detected exospherically escaping carbon and oxygen atoms. Such atoms should be too heavy to escape by the Jean's mechanism, responsible for the hydrogen escape, and instead Vidal-Madjar et al. (2004) suggest that hydrodynamic escape (or 'blow-off') is responsible, whereby the outward flow of exospherically escaping hydrogen atoms carry with them heavier atoms such as carbon and oxygen.

1.2.4 Microlensing

For several years now there have been campaigns to observe galactic bulge microlensing events, with a view to searching for dark matter and extrasolar planets. In this technique, light from a distant (source) star is observed as another star at intermediate distance (the lens star) passes close to, or in front of it. Light from the source star is gravitationally bent around the lens star and thus its apparent magnitude changes during the event. Two such campaigns are OGLE (Udalski, 2003) and MOA (Bond et al., 2001). In addition to lensing events, such programmes are also sensitive to planetary transits and to date, OGLE has detected the transits of five previously unknown extrasolar planets.

In 2003, both observatories observed a remarkable microlensing event shown in Fig. 1.3 where, in addition to the central peak in source star brightness due to the gravitational lensing of the lens star, two additional sharp peaks were observed which are interpreted as being due to the microlensing of a planetary companion to the lens star. Bond et al. (2004) conclude, assuming the lens star to be a main sequence M dwarf, that the planet has a mass of 1.5 MJ, and orbits the lens star at a distance of approximately 3 AU.

OGLE has now detected three further planets through gravitational microlens-ing events. For future observations we will see later in Sect. 1.4 that gravitational lensing is the only detection method that is capable of sensing terrestrial planets orbiting some distance from their stars (dubbed 'cool Earths'). In addition to the continuation of the OGLE and MOA campaigns, other ground-based campaigns include PLANET, which is a collaboration of telescopes in the southern hemisphere observing since 1995. The sensitivity of microlensing campaigns to 'cool Earths' would be further advanced by placing the telescope in space and proposed mis

2The Hill radius gives the limit of the gravitational sphere of influence of a body in orbit about another heavier body, in this case the central star.


Fig. 1.3. Observation of gravitational microlensing by a planet by OGLE (Bond et al., 2004). Inset panel shows all OGLE data from 2001 to 2003, while the main figure shows a close-up of the data for 2003 for both OGLE and MOA.

2820 2840 2860 2880 HJD- 2450000

Fig. 1.3. Observation of gravitational microlensing by a planet by OGLE (Bond et al., 2004). Inset panel shows all OGLE data from 2001 to 2003, while the main figure shows a close-up of the data for 2003 for both OGLE and MOA.

sions include GEST (Galactic Exoplanet Survey Telescope) and Microlensing Planet Finder (MPF).

Was this article helpful?

0 0

Post a comment