Radial velocity follow-up of CoRoT transiting exoplanets

We report on the results from the radial-velocity follow-up program performed to establish the planetary nature and to characterize the transiting candidates discovered by the space mission CoRoT. We use the SOPHIE at OHP, HARPS at ESO and the HIRES at Keck spectrographs to collect spectra and high- precision radial velocity (RV) measurements for several dozens dif- ferent candidates from CoRoT. We have measured the Rossiter- McLaughlin effect of several confirmed planets, especially CoRoT- 1b which revealed that it is another highly inclined system. Such high-precision RV data are necessary for the discovery of new tran- siting planets. Furthermore, several low mass planet candidates have emerged from our Keck and HARPS data.


Introduction
Transiting exoplanets are unique targets for which we can measure the planetary radius by high accuracy photometry when the planet passes 2 A. Santerne et al.
in front of its host star and its mass and orbital characteristics (eccentricity, alignment with the spin of the star) by Doppler measurements of the host star. Thus it is possible to compute the mean density of the planet and to model its internal structure or to explore the composition and characteristics of its atmosphere (albedo, temperature, and atmospheric composition) by photometry or spectroscopy observations during the transit or the eclipse.
CoRoT (Baglin et al. 2006;Deleuil et al., this book) is the first space-based photometric survey dedicated to finding transiting exoplanets. CoRoT finds about 250 objects per run whose light curves show transit-like events. Most of them are clear eclipsing binaries (EB), but when a target shows periodic single transits (i.e. no secondary transits), no ellipsoidal variations and a shape, duration, and depth compatible with a transiting exoplanet, we consider it as a transiting exoplanet candidate. But these planetary candidates could still be mimicked by a transiting low-mass star, a grazing EB, a main sequence star eclipsing a giant star, or by an EB diluted by a third star. These EB scenarii (about 50% of candidates) could be resolved by high-resolution spectroscopy observations in order to discard all binary scenarii (SB1, SB2, SB3, ...). For example, Fig. 1. shows the result of the CCF * of a transiting candidate followed-up with SOPHIE. When EB scenarii are discarded, precise RV observations are required to measure the mass of the transiting object and characterize its orbit.

RV follow-up facilities
Precise RV follow-up observations are obtained using a network of 3 spectrographs that share candidates in a optimized way depending on the brightness of the host star and the shallowness of the candidate: 2.1 SOPHIE spectrograph SOPHIE (Bouchy et al 2009a) is a high-accuracy fiber-fed echelle spectrograph mounted on the 1.93-m telescope in Haute-Provence Observatory, France. SOPHIE has a resolution of R HR ∼ 75, 000 (in High-Resolution mode) or R HE ∼ 39, 000 (in High-Efficiency mode) at 550nm. Due to the faintness of the CoRoT targets, we used only the High-Efficiency mode of SOPHIE which has an intrinsic stability of about 10m s −1 . We observe with SOPHIE all transiting planetary can-

HARPS spectrograph
HARPS (Mayor et al. 2003) is a high-accuracy fiber-fed echelle spectrograph mounted on the ESO 3.6-m telescope in La Silla Observatory, Chile. HARPS has a resolution of R HAM ∼ 110, 000 (in High-Accuracy mode) or R EGGS ∼ 70, 000 (in EGGS mode) at 550nm. We can follow targets up to V-magnitude of 16 but we focus on the shallowest candidates or on candidates with the longest period around the brightest targets. A large program of 16 nights per semester is dedicated to this program on HARPS and permitted to establish or confirm the planetary nature of most of the CoRoT exoplanets discovered so far (e.g. Fig. 2)

HIRES spectrograph
The Keck/HIRES observations were obtained as part of NASA's key science project to support the CoRoT mission. HIRES is a high-resolution, optical spectrograph (Vogt et al. 1994) mounted on the 10 m Keck 1 telescope on the summit of Mauna Kea, Hawaii. We use HIRES in 4 A. Santerne et al.
combination with an iodine cell to obtain highly precise RV measurements. We use HIRES with the 0.86 arcsec slit that yields a resolving power of ∼ 45, 000 at 550nm. The 10-m aperture of Keck permits us to follow-up some candidates that are too faint for HARPS (V-magnitude ∼ 16). In some cases, Keck/HIRES could observe Rossiter-McLaughlin (RM) anomaly with a better time resolution than HARPS or to obtain a high signal-to-noise (S/N) spectrum of the host star needed to characterize the star parameters. About 5 nights per semester are allocated on Keck for CoRoT targets permit to confirm and characterize some of the CoRoT exoplanets discovered so far (e.g. Fig. 2 on right).

Observations strategy
Due to the large CoRoT PSF † , the first step of follow-up process is to reobserve the transit with a higher spatial resolution in order to discard all background eclipsing binaries (BEB) that could cause the transit (Deeg et al. 2009). Then, two high-resolution RV observations are scheduled at the extrema phase of the planetary orbit (assuming a circular orbit), corresponding to T 0 + P n − 1 / 4 and T 0 + P n + 1 / 4 , where P and T 0 are, respectively, the period and the epoch of the transit determined from the CoRoT LC analysis and n is the number of orbits since T 0 . These two RV measurements are sufficient in most cases to estimate the nature of the transiting object: large RV variations of several km s −1 in phase with the CoRoT ephemeris indicate an EB (SB1) as shown in † Point Spread Function Fig. 1.. Small RV variations of less than a few km s −1 are compatible with a planetary nature and require more observations to be confirmed.

Photon noise
Mass characterization of low-mass planets strongly depends on the star brightness and its rotational velocity. Table 1 indicates photon noise uncertainties on SOPHIE and HARPS in one hour exposure time for 3 different v sin i and different V-magnitudes. Transiting Super-Earths (K < 10m.s −1 ) can only be characterized around low-rotating stars brighter than mv=13, while transiting hot-Neptune (K < 30m.s −1 ) can be characterized with HARPS on stars up to mv=14. For stars fainter than mv=14, or fast-rotating stars, only giant planets and brown dwarves can be characterized.

Contamination by Moon Background Light
Due to the faintness of the CoRoT target, observations are mostly scheduled in dark time (i.e., when the Moon is set). The Moon background light (MBL) that is blended the target spectrum can affect the RV measurement of up to several km s −1 as one can see in Fig. 3. The correction consists in subtracting the CCF of the sky observed simultaneously to the CCF of the target (for more details, see Bonomo et al 2010).

Blended eclipsing binary
In the case of a triple system or an unresolved BEB, the spectrum of the main star could be blended with the spectrum of the eclipsing binary and it can affect the RV measurements of the main star (Bouchy et al 2009b). This kind of scenario can mimic the RV signature of a 6 A. Santerne et al. planetary companion, but produces asymmetries in the CCF in phase with the orbital period (see Fig. 4), or different RV amplitudes when computing the CCF with masks from different spectral types which can detect blending stars with different T ef f (see Fig. 4). In some cases, RV variations are not in phase with the transit seen by CoRoT. This is mostly caused by blended binaries : a contaminating EB at the CoRoT period and a binary with a different period. Fig. 5 shows two of these cases where the transit is explained by a BEB. For the first one (left panel), the background binary is separated by a few arcsec (within the CoRoT photometric mask) and for the second, the two binaries are located inside the seeing disk (Tal-Or et al., in prep).

Stellar activity : the case of CoRoT-7
CoRoT-7 is an active star which hosts at least 2 super-Earths. Stellar activity induces RV signatures (see Fig. 6) at the level of a few m.s −1 . To dissociate stellar activity from planetary signature, it is necessary to have ancillary measurements like simultaneous photometry, bisector, FWHM or CaII measurements. Right now, characteristics of the CoRoT-7 system is still under intense discussion (see Queloz et al. 2009;Hatzes et al. 2010;Lanza et al. 2010;Pont et al. 2010b;Boisse et al. subm., Ferraz-Mello et al, subm.).

8
A. Santerne et al. Figure 6.: All HARPS observations of the CoRoT-7 host star  showing (from top to bottom) RV, FWHM, bisector span and CaII activity level (R'(HK)) measurements as a function of time. One can see that FWHM, bisector span and CaII activity level present significant variations that could be explained by stellar activity which affects also RV measurements.

About follow-up statistics
CoRoT so far has discovered 15 new transiting exoplanets and brown dwarves (Deleuil et al. this book). These discoveries are the fruit of CoRoT high-accuracy space-based photometry combined with an intensive ground-based photometric and spectroscopic follow-up of about 200 transiting exoplanet candidates for the first 3 years of CoRoT. Run report papers (IRa01: Moutou et al. 2009, LRc01: Cabrera et al 2009 present the observation and analysis sub-sampled of these candidates including about 50% of binary (SB1, SB2, BEB, blended EB), about 10% of hot stars for which we cannot measure RV with enough precision to characterize the candidate's mass (and could still be planetary), and about 10% of confirmed transiting exoplanets. The remaining 30% are the faintest stars or too low-priority candidates with poor planet likelihood which were not followed up or unsolved candidates due to photon noise limitation.

Rossiter-McLaughlin (RM) observations
To complete the characterization of the system, we observe Rossiter-McLaughlin RV anomaly (Winn et al., this book; Triaud et al., this book) during the transit. It permits us to measure the sky-projected angle between the spin of the star and the orbit of the planet. CoRoT-2b (Bouchy et al. 2008) and CoRoT-3b (Triaud et al. 2009) are spin-orbit aligned exoplanets while CoRoT-1b was revealed to be a misaligned planet (Pont et al. 2010). Observation of part of the transit of CoRoT-11b indicates another misaligned exoplanet (Gandolfi et al. 2010), although observations did not cover the complete transit. Part of the transits of CoRoT-9b and CoRoT-6b were observed by SOPHIE and HARPS and require more observations to conclude.

Conclusion
Transiting exoplanet surveys need RV follow-up in order to determine the nature and the characteristics of the exoplanets candidates. Using the facilities of an optimized network of 3 high-resolution spectrographs for follow-up (SOPHIE, HARPS and HIRES) with powerful diagnostics to discard false positives and secure detection, CoRoT is, so far, the photometric survey that has discovered more planets per square degree of observed sky. More than 1000 spectra with signal-to-noise of up to 100 on about 200 transit candidates were taken with SOPHIE, HARPS, and HIRES during the first 3 years of CoRoT. Fifteen new exoplanets and brown dwarves have been discovered and characterized by these high-resolution spectrographs so far. Currently, 6 of these CoRoT planets have been observed in order to measure their RM effect. One planet orbit is is clearly misaligned with the spin of its host star while another one shows strong evidence of a misalignment, but requires more measurements to confirm this.