Transit detections

We have seen in Section 1.2 that the basic drawback of the RV method is the lack of independent information about the orbital inclination, which leads to a fundamental uncertainty in the planet mass. Currently, the most successful means of obtaining this information is via the detection of planetary transits. In the Solar System, transits are a well-known rare phenomenon in which one of the inner planets (Mercury or Venus) passes in front of the solar disk. The most recent Venus transit occurred on 8 June 2004, and attracted considerable public attention and media coverage. Extrasolar transits occur when an extrasolar planet passes in front of its host-star disk. Obviously, we cannot observe extrasolar planetary transits in the same detail as transits in the Solar System. With current technology, the only observable effect would be a periodic dimming of the star light, because the planet obscures part of the star's surface. Thus, transits can be detected by photometry, i.e., monitoring the stellar light intensity.

The probability that a planetary orbit would be situated in such a geometric configuration to allow transits is not very high. For a circular orbit, simple geometrical considerations show that this probability is:

Fig. 1.3. The transit light curve of HD209458, from Charbonneau et al. (2000).

Fig. 1.3. The transit light curve of HD209458, from Charbonneau et al. (2000).

where R+ and Rp are the radii of the star and the planet, respectively (e.g., Sackett (1999)). For a typical hot Jupiter, this probability is about 10%.

The idea of using transits to detect extrasolar planets was first raised by Struve (1952), butthe first extrasolar transit was observed only in 1999. Mazeh etal. (2000) detected a planet orbiting the star HD 209458, using 'traditional' RV methods. Soon after the RV detection, Charbonneau et al. (2000) and Henry et al. (2000) detected a periodical dimming of the light coming from HD 209458, at exactly the predicted orbital phase and with the same period as the RV variation, of 3.52 days. The light dimmed by about 1.5% for about 1.5 hours (Figure 1.3). The two teams detected the transits using small and relatively cheap telescopes, demonstrating that it was realistic to achieve the required photometric precision with ground-based observations.

The depth of the transit (i.e., the amount by which the light intensity drops) depends on the fraction of the stellar disk obscured by the planet. Thus, assuming there is a reasonable estimate of the star's radius, we can use the depth to derive the planet radius. This is the first direct estimate we have of a physical property of the planet itself. Obviously, the detection of transits immediately constrains the orbital inclination (i) to values close to 90° (transits occur only when we observe the orbit edge-on or almost edge-on). Furthermore, the transit duration depends strongly on the orbital inclination, and we can use it to explicitly derive i (e.g., Sackett (1999)). Thus, in combination with RV data, we can finally obtain a measurement of the planet mass, Mp.

Brown et al. (2001) used the Hubble Space Telescope (HST) to obtain a very precise light curve of HD 209458. This light curve led to a very precise estimate of the planet radius: 1.347 Jupiter radii (RJ). Using the inclination and the mass estimate from the RV orbit, the planet mean density could be derived: 0.35 g cm-3.

1.3 Transit detections

The special circumstances of a transiting extrasolar planet were exploited by many more observations of HD 209458, with new clues about its atmosphere. Those observations are reviewed in Chapter 2.

The successful observations of HD 209458 encouraged many teams to try to detect more transiting extrasolar planets. Currently about 25 surveys are being conducted by teams around the world (Horne, 2006). Some of these surveys use small dedicated telescopes to monitor nearby stars (which are relatively bright) in large fields of view, like TrES (Alonso et al., 2004) or HATnet (Bakos et al., 2004). Other surveys, such as OGLE-III (Udalski et al., 2002) or STEPSS (Burke et al., 2004), focus on crowded fields like the Galactic Centre or globular clusters, and monitor tens of thousands of stars.

So far, the only successful surveys have been OGLE, with five confirmed planets, and TrES and XO, with one planet each. Their success highlights the difficulties such surveys face, and the problems in interpreting the observational data. The basic technical challenge is transit detection itself. The transits last only a small fraction of the planet's orbital time around its central star, and the drop in the stellar brightness is usually of the order of 1-2% at most. The first obvious challenge is to reach a sufficient photometric precision. The next challenge is to obtain sufficient phase coverage, on the observational front, and efficient signal analysis algorithms, on the computational front.

The OGLE project has yielded so far 177 transiting planet candidates (Udalski et al., 2004). However, only five thus far have been confirmed as planets. This is due to the fundamental problem in using photometry to detect giant planets. Since the transit light curve alone does not provide any information regarding the mass of the eclipsing companion, we have to rely on its inferred radius to deduce its nature. However, it is known (e.g., Chabrier and Baraffe (2000)) that in the substellar mass regime, down to Jupiter mass, the radius depends extremely weakly on the mass. Therefore, even if we detect what seems to be a genuine transit light curve, the eclipsing object may still be a very low-mass star or a brown dwarf. The only way to determine its nature conclusively is through RV follow-up that would derive its mass. Thus, while only five of the OGLE candidates have been shown to be planetary companions, many others have been identified as stellar companions.

The proven non-planetary OGLE candidates demonstrate the diversity of events that can be mistaken for planetary transits. OGLE-TR-122 is a perfect example of a low-mass star which eclipses its larger companion, like a planet would. Only RV follow-up determined its stellar nature (Pont et al., 2005). Other confusing configurations are 'grazing' eclipsing binary stars, where one star obscures only a tiny part of its companion, and 'blends', where the light of an eclipsing binary star is added to the light of a background star, effectively reducing the measured eclipse depth. In principle, these cases can be identified by close scrutiny of the light curve

(Drake, 2003; Seager and Mallen-Ornellas, 2003), or by using colour information in addition to the light-curve shape (Tingley, 2004). However, the most decisive identification is still through RVs.

Although transit detection alone is not sufficient to count as planet detection, transit surveys still offer two important advantages over RV surveys. First, they allow the simultaneous study of many more stars - the crowded fields monitored by transit surveys contain thousands of stars. Second, they broaden the range of stellar types which we examine for the existence of planets. While obtaining the required precision of RV measurements puts somewhat stringent constraints on the stellar spectrum, transit detection relies much less on the stellar type. The star simply has to be bright enough and maintain a stable enough brightness so that we can spot the minute periodical dimming caused by the transiting planet. Since stellar radii depend only weakly on the stellar mass, we should be able to detect planetary transits even around stars much hotter (and therefore more massive) than the Sun.

Was this article helpful?

0 0

Post a comment