EDP Sciences
Free Access
Issue
A&A
Volume 522, November 2010
Article Number A110
Number of page(s) 10
Section Planets and planetary systems
DOI https://doi.org/10.1051/0004-6361/201015154
Published online 09 November 2010

© ESO, 2010

1. Introduction

Transiting planets are fundamental objects to our understanding of planetary formation and evolution. Their particular geometrical orientation allows the simultaneous measurement of their mass and radius, permitting a first order study of their internal structure. Moreover, through the careful analysis of their passages in front of (primary transit) and behind (occultation of the planet or secondary transit) their host star, a characterization of the composition and the temperature structure of their atmosphere can be carried out.

CoRoT  is a space telescope dedicated to the study of asteroseismology and the discovery of extrasolar planets by the method of transits (Baglin et al. 2006). The CoRoT survey has previously detected, among other planets, the faint signal of a small rocky planet (CoRoT-7b; Léger et al. 2009; Queloz et al. 2009; Bruntt et al. 2010b) and a giant planet with a temperate surface temperature (CoRoT-9b; Deeg et al. 2010).

This paper presents the discovery of the planet CoRoT-13b, a dense giant planet in a close orbit around a main sequence star. Section 2 describes the observations performed with the satellite. Ground-based observations including spectroscopic characterization of the star and radial velocity measurements are reported in Sect. 3. The derivation of the planetary parameters is described in Sect. 4. A discussion of the results is presented in Sect. 5.

2. CoRoT observations

The observations of the CoRoT field LRa02 started on 16 November 2008; the first transits of CoRoT-13b were discovered by the Alarm Mode and the target was oversampled (the standard 512 s sampling rate was changed to 32 s; see Surace et al. 2008) between 9 December 2008 and the end of the observations in 11 March 2009, gathering in total about 250000 photometric measurements.

The coordinates, identification labels, and magnitudes of CoRoT-13 are given in Table 1. It is a relatively faint object for CoRoT (V = 15.039) and it was therefore assigned a monochromatic aperture mask (for a complete description of the observing modes of the satellite, please refer to Boisnard & Auvergne 2006; Barge et al. 2008; Auvergne et al. 2009). Figure 1 shows the raw light curve (LC) of the target. One of the most significant environmental effects in CoRoT LCs are hot pixels, which are proton impacts that produce permanent damage to the CCD lattice (see Drummond et al. 2008; Pinheiro da Silva et al. 2008). On average, 0.3% of the pixels are affected by one of these events with an impact of more than 1000 electrons in flux during a long run (Auvergne et al. 2009). The mask of CoRoT-13b has only 69 pixels, but unfortunately this particular LC shows at least 3 impacts over the 1000-electron threshold plus some other impacts of lesser importance. These impacts do not significantly affect the characterization of the target as long as the regions affected are properly treated (Pinheiro da Silva et al. 2008).

Table 1

IDs, coordinates, and magnitudes of CoRoT-13.

3. Ground-based observations

3.1. Photometric measurements

A complete description of the photometric follow-up of CoRoT targets is given in Deeg et al. (2009). For the particular case of CoRoT-13b, photometric measurements of the host star were carried out at the 3.6 m Canada-France-Hawaii-Telescope (CFHT) in late November 2009 confirming that the transit was on-target, although with a depth slightly greater than expected because of the contribution of contaminants to the flux in the CoRoT  mask. Later on, in early January 2010, new on-off observations were carried out at the IAC80 telescope, in Tenerife, confirming the previous results.

Figure 2 shows the region of the sky around the target taken with the Wide Field Camera of the Isaac Newton Telescope (INT/WFC) in Roque de los Muchachos. The closest contaminants to the main target, according to the ExoDat database (Deleuil et al. 2009), are shown in Table 2. In particular, the object 110839426 is included completely in the CoRoT mask, diluting the transit signal of CoRoT-13b. An analysis of the point spread function (PSF) of CoRoT reveals that the main target, the planet-hosting star, is responsible for 89 ± 1% of the flux within the mask.

thumbnail Fig. 1

Raw light curve of CoRoT-13b. The target was oversampled from 9 December 2008 (HJD 2454810.20); for visualization purposes, we superimpose the LC binned at a sampling rate of 512 s to the 32 s sampled region. The jumps in the data are caused by hot pixels as described in the text.

Open with DEXTER

3.2. Spectroscopic measurements

The star was observed with HIRES at the Keck on 5 December 2009 as part of NASA’s key science programme in support of the CoRoT mission. We obtained one spectrum without the iodine cell and with an exposure time of 1200 s. The spectral resolution is  ~ 45000. To determine the atmospheric parameters of CoRoT-13, we analysed the spectrum using the VWA software (Bruntt et al. 2004, 2010a). We selected lines in the range 5050–7810 Å. A small section of the spectrum is shown in Fig. 3, where the typical signal-to-noise ratio (SNR) is 55. To determine the atmospheric parameters we used the neutral and ionized Fe lines and also the wide Ca lines at 6122 and 6162 Å (see Bruntt et al. 2010a, for a description). From this analysis we determined the parameters Teff = 5945 ± 90 K, log g = 4.30 ± 0.10,  [ M / H ]  =  + 0.01 ± 0.07 (mean of Si, Fe and Ni), and projected rotational velocity vsini = 4 ± 1 km s-1. The Teff was adjusted by  − 40 K based on the comparison of 10 stars with Teff determined from both interferometric and spectroscopic methods as described by Bruntt et al. (2010a). The abundances of 10 elements are given in Table 3 and shown in Fig. 4. The horizontal yellow bar is the mean metallicity. Although the metallicity of this star is essentially solar, the abundance of Li i is  + 1.45 dex, similar to the case of another planet hosting star, CoRoT-6 (Fridlund et al. 2010).

The mass, radius, and age of the star are calculated using the Starevol evolutionary tracks (Siess 2006, Palacios, private communication). The input values are the spectroscopic parameters derived above (Teff and [Fe/H]) and the proxy for the stellar density () obtained from the modeling of the transit light-curve. The results are M = 1.09 ± 0.02 M  and R = 1.01 ± 0.03 R, which gives a corresponding surface gravity of log g = 4.46 ± 0.05 in perfect agreement with the spectroscopic value. We obtain an interval for the star’s age of between 0.12 and 3.15 Gyr.

Finally, using simultaneously the seven BVriJHKs broad-band magnitudes reported in Table 1 and following the method described in Gandolfi et al. (2008), we derived an interstellar extinction AV = 0.20 ± 0.10 mag and a distance to the star d = 1060 ± 100 pc. Consistent results were obtained using the absolute magnitude (Allen 1973) and the intrinsic near-infrared colours (González Hernández & Bonifacio 2009) tabulated for a G0 V type star.

3.3. Radial velocity measurements

Precise radial velocity measurements of CoRoT-13 were obtained with the HARPS spectrograph between the nights of 22 November 2009 and 15 February 2010 (ESO program 184.C-0639). HARPS is a cross-dispersed echelle spectrograph fiber-fed from the Cassegrain focus of the 3.6 m telescope at La Silla Observatory, Chile (Mayor et al. 2003). Fifteen spectra with a spectral resolution R ≈ 115000 were obtained using exposure times of 3600 s, and setting one of the two available fibers on the sky to monitor the presence of moonlight and help achieve optimal sky background subtraction, which is important for faint targets such as this. The SNRs per pixel of these observations at 5500 Å  range from 7.1 to 11.7. Our Th-Ar calibrations exposures were adquired at the beginning of each night, which has been shown to be enough to obtain the required precision due to the high stability of the instrument.

thumbnail Fig. 2

Region of the sky around CoRoT-13. Left: image taken with INT/WFC with a resolution of 0.3′′. The position of the target is indicated with a white cross. Right: image taken by CoRoT with the same scale and orientation; the resolution is around 2.32′′/pixel. The area inside the black line is the mask used to compute the photometry; the position of the nearest contaminants from the ExoDat database are indicated with different types of crosses.

Open with DEXTER

Table 2

Closest contaminants to CoRoT-13b with their respective magnitudes and relative distances to the target.

The spectra were reduced and extracted using the HARPS pipeline, and the radial velocity was measured on each extracted spectrum by means of a weighted cross-correlation (see Baranne et al. 1996) with a numerical mask corresponding to a G2 star. The resulting cross-correlation functions (CCFs) were fitted by Gaussians to derive the radial velocities. The measured values are listed in Table 4 and shown in Fig. 5, together with the best-fit orbital solution (see below). During some of the observations, the star fiber was contaminated by moonlight. In those cases, if the peak of the CCF produced by moonlight was expected to be close to the measured speed of the target, a correction was applied using data from the fiber recording the sky (see Bonomo et al. 2010, in preparation). Those points are shown as white circles in Fig. 5 and we added 30 m s-1  quadratically to the uncertainty estimated from the CCF to account for possible systematic errors introduced by the moonlight correction.

thumbnail Fig. 3

Detail of the Keck spectrum of CoRoT-13 around the pressure-sensitive Ca ii line at 6162 Å. The synthetic spectrum is shown with a dashed line. Spectral lines used in the abundance analysis are marked with solid vertical lines, while the other spectral lines are marked with dotted lines.

Open with DEXTER

The orbital solution was found by χ2 minimization, with the period and epoch of inferior conjunction (when radial velocity is zero after removal of the systemic velocity) being fixed to the values provided by the CoRoT  ephemeris (which are calculated by fitting a linear regression to the centre position of the individual transits). The eccentricity of the orbit was a free parameter at first, but since the best-fit solution was compatible with a circular orbit at the two-σ level (the three-σ upper limit to the eccentricity is 0.145), we decided to fix it to e = 0 when determining the remaining parameters and their uncertainties. Figure 5 shows the RV measurements, phased to the CoRoT  period, together with the best-fit circular model and the residuals; the obtained parameters are listed in Table 6. The resulting value of χ2 is 8.2 for 13 degrees of freedom, and the rms of the residuals is 20.2ms-1, which is compatible with expectations based on the median of the RV uncertainties, 21.2ms-1. These suggest that the circular model – with the obtained parameters – adequately describes the available data.

Table 3

Abundances of 10 elements in CoRoT-13 relative to the Sun.

thumbnail Fig. 4

Mean abundances of 10 elements in CoRoT-13 Keck spectrum. White circles correspond to neutral lines and red boxes represent singly ionized lines. The yellow area represents the mean metallicity within one σ error bar.

Open with DEXTER

Table 4

Radial velocities measurements and bisector span velocities measured with HARPS.

thumbnail Fig. 5

Upper panel: radial Velocity measurements phased to the orbital period measured by CoRoT. The solid curve represents the best-fit solution. Lower panel: residuals to the fit. In both panels, the white symbols indicate the measurements that have been corrected for moonlight contamination.

Open with DEXTER

With fixed ephemeris and eccentricity set to zero, the fitting problem is reduced to a linear least squares minimization with two free parameters. The uncertainties reported in Table 6 are therefore estimated using the covariance matrix, which has a covariance term of 6.48m2s-2. However, stellar activity and other long-term phenomena can produce correlated noise in the observations and hence render the above estimation of the uncertainties invalid. To explore this, we used the Prayer Bead method (see, for example, Désert et al. 2009; Winn et al. 2009), i.e., we performed a cyclic permutation of the residuals of the best-fit curve and fit the model again. We repeated this for every possible shift and measured the standard deviation of the obtained parameters. We also performed a similar analysis but after randomly re-ordering the residuals rather than shifting them. In this way, we constructed 10000 synthetic data sets that were again used to fit our model. In both cases, the obtained dispersion in the parameters was found to be smaller than the error bars reported in Table 6.

The bisector analysis for these data is shown in Fig. 6, where the uncertainty in the bisector span velocity has been set to twice that of the corresponding radial velocity. The bisector span velocities do not show any clear dependence with radial velocity values and the Pearson correlation coefficient between these two magnitudes is around 0.15, which is a sign of a lack of correlation. This clearly indicates that the measured RV variations do not originate from changes in the shape in the CCF as would be the case if the system consisted of a background eclipsing binary whose light were diluted by the CoRoT  main target. In addition, no significant changes in the measured RVs are observed when different stellar masks are used for the correlation, which more conclusively excludes the background eclipsing binary scenario.

4. Planetary parameters

The transit parameters were determined in the same manner as described in the case of CoRoT-6b (Fridlund et al. 2010). The pre-processed light curve was divided by its median, and a new light curve was constructed by convolving it with a fourth order Savitzky-Golay filter (Press et al. 2002). The standard deviation of the differences between the measured and the convolved light curves was calculated and a 5σ clip applied to remove spurious outliers. This clipping was iterated until no more outliers were found. After these steps, the light curve was folded and a bin average applied forming 5000 bins.

We used the Mandel & Agol (2002) model to perform the fit to the final, phase-folded light curve. We assumed a circular orbit and an absolutely dark planet at the CoRoT wavelength.

The six adjusted parameters were the ratios a / Rs, k = Rpl / Rs, the impact parameter b = acosi / Rs (i is the inclination of the orbit), the two limb-darkening coefficients u + , u − , and the contaminant factor, which was varied across the range 10–12% (corresponding to the measured value of 11 ± 1%) to enable a more accurate estimation of errors. The quadratic limb darkening law is I(μ) / I(1) = 1 − μ + uaμ + ub(1 − μ)2, and we fitted the coefficients u +  = ua + ub and u −  = ua − ub. The best-fit relation was found by the Harmony Search algorithm (Geem et al. 2001), a genetic-type algorithm. The 1σ errors were obtained from the width of the parameter distribution to be between and .

We calculated different solutions for the light curve of CoRoT-13b using different approaches to the treatment of the limb darkening parameters to understand the constraints that this effect places on the determination of the final values. The results are presented in Table 5. In model A, we allowed both u +  and u −  to be free parameters. In models B, C, and D we fixed u +  to be 0.81 (the value that we found in model A), 0.88 (i.e. 0.81 + 1σ) and 0.74 (0.81 − 1σ; where σ means the error bar of the coefficient found in model A), respectively. In model E, we fixed the limb darkening to the predicted values for the star (Sing 2010). The corresponding χ2-values of these solutions did not differ too much from each other, showing that limb darkening is a second order effect in the transit shape. The different assumptions about the limb-darkening yielded quite consistent light curve solutions: the a / Rs, k ratios are in good agreement within their error bars for these five models. When we fixed u + , the uncertainties in the impact parameter increased by a factor of two, despite the error bars in the impact parameters found in the B, C, D, and E solutions are overlapping with the one found in model A. The precise determination of the impact parameter is more complicated when the transit is nearly central. We learned from this experiment that fitting both limb darkening coefficients (or, what we actually did, fitting both of their combinations) yielded more precise values.

Since u +  = 0.81 ± 0.07 (Sing 2010 gives 0.662 ± 0.022 for this temperature, log g, and metallicity) and u −  =  − 0.09 ± 0.09 (Sing 2010 gives 0.156 ± 0.022) the agreement between theoretical predictions and measurements is withing 2σ error bars, which is satisfactory. We accept solution A as the definitive one (the fit is shown in Fig. 7).

Table 5

Values of the adjusted parameters in the modeling of the transit of CoRoT-13b in the different approximations described in the text.

Table 6

Planet and star parameters.

5. Discussion

5.1. Stellar properties

The spectroscopic analysis of CoRoT-13 reveals a G0V star with an age of between 0.12 and 3.15 Gyr, solar metallicity ( [ M / H ]  =  + 0.01 ± 0.07), a high relative abundance of lithium ( + 1.45 dex), and a low activity level according to the analysis of activity indicators such as the H-K Ca ii lines (where no emission is detected).

The rate of lithium depletion of solar-like stars is related to both the age of the star and to the depth of the convective zone, as it is destroyed at a temperature of approximately  ~ 2.5 × 106 K in the radiative region of a star (Chaboyer 1998). Given the spectral type of CoRoT-13, we expect a lower lithium depletion rate than in solar analogs (Castro et al. 2009). Using the value of the Li i abundance, we computed a log n(Li) = 2.55. From Fig. 7 of Sestito & Randich (2005) and with the value of the effective temperature (Teff = 5945 ± 90 K), we estimated the age of the star to be in the range 300 Myr to 1 Gyr, consistent with the range from the evolutionary models. Israelian et al. (2009) claimed a lithium depletion in solar-like stars with orbiting planets, although it is not clear that previous observations support this conclusion (Meléndez et al. 2009b). CoRoT-13 is not depleted in lithium, albeit we emphasize that the effective temperature of this star is slightly higher than the upper limit for depletion given in Israelian et al. (2009).

The v sin i  value indicates a rotational period of the star of around 13 days1. Gyrochronology (Barnes 2007) was able to be used as an age estimator. Using the improved gyrochronology relations from Mamajek & Hillenbrand (2008), we derived a gyrochronologic age of 1.66 Gyr well within the range of age given by evolutionary models. Emission lines are detected neither at the bottom of the Ca ii H and K lines nor in the Hα line implying that the star belongs to the inactive population with . We were thus unable to derive any chromospheric age for the star. Other G0V stars are found with similar rotation rates and low activity levels (Noyes et al. 1984).

thumbnail Fig. 6

Bisector analysis of the HARPS data. The white symbols represent measurements that have been corrected for moonlight contamination.

Open with DEXTER

thumbnail Fig. 7

Folded transit of CoRoT-13b, best-fit and its residuals using the values of solution A in Table 5.

Open with DEXTER

We searched for signs of stellar rotation in the light curve (LC) observed by CoRoT  to compare with the clear signs of spot modulation found in the cases of CoRoT-2b (Lanza et al. 2009b), CoRoT-4b (Lanza et al. 2009a), CoRoT-6b (Fridlund et al. 2010) or CoRoT-7b (Lanza et al. 2010). The Lomb-Scargle periodogram of the LC, once the planetary transits have been removed and the hot-pixel events treated, indeed exhibits a significant broad peak around 77 days, but no significant peak at the expected rotational frequencies. The 77 day period is comparable with the length of the run (115 days), so we may have observed an irregular pattern that, in the observing window of 115 days, has a typical timescale of variation of 77 days that may mimic an harmonic oscillation. The only reliable information about this particular signal is its characteristic amplitude of 0.5% and its characteristic timescale of 77 days; one has to be extremely cautious when interpreting its nature. Two immediate possibilities are stellar activity and an instrumental residual signal. In the latter case, straylight or other instrumental effects such as temperature fluctuations should affect a region of the CCD (if not all), instead of a single target. None of the targets in the neighborhood of CoRoT-13 show a similar pattern. Other environmental features (such as hot pixels), although present, have a completely different behaviour in terms of both amplitude and timescale and are probably not responsible for the signal. The spectroscopic analysis shows that CoRoT-13 is a quiet dwarf star. A G0V star might show spot modulation with a characteristic timescale similar to the rotational period of the star as well as long-term variations in timescales of several years (Baliunas et al. 1995), but not in the timescales considered here. On the other hand, slowly rotating giant stars might show a pattern of variability with similar amplitudes and characteristic timescales as those revealed by the periodogram. We therefore conclude that the modulation measured in the LC of CoRoT-13 is due to a background contaminant and not the main target.

Lanza (2010) found a relation between the presence of giant planets and the angular momentum loss of their host stars, such that giant planet hosting stars with Teff ≳ 6000 K tend to have an n / Ω ≈ 1,2 orbital period/rotational period synchronization with their respective planets. The effective temperature of CoRoT-13 is immediately below this lower limit and the rotational period obtained from the spectroscopic analysis (around 13 days) is indeed longer than the expected 2:1 resonance or the synchronization. However, the rotational period is not sufficiently well constrained (the relative uncertainty is around 35%) and the inclination of the spin-axis of the star is not known. A study of the Rossiter-McLaughlin effect of this planet would infer the relative angle of inclination of the planet and the star, which is an important parameter in the model proposed by Lanza (2010).

thumbnail Fig. 8

Position of CoRoT-13b (square) among the other transiting planets in a mass-radius diagram.

Open with DEXTER

5.2. Planet interior

Figure 8 compares CoRoT-13b to other transiting exoplanets in a mass-radius diagram. Although it is not the densest object known so far (both super-Earths and brown dwarfs may be denser), it is clearly extremely dense for its mass. This is confirmed by a combined modeling of the star and planet evolution (see Guillot & Morel 1995; Morel & Lebreton 2008; Bordé et al. 2010) shown in Fig. 9. The small planetary radius derived from the transit photometry and spectroscopy is consistent with a truly considerable amount of heavy elements in the planet. When fitting the stellar effective temperature and density within their 1σ error bars (see Table 6 and red area in Fig. 9), between about 140 and 300 M ⊕   of heavy elements are required to reproduce the measured planetary size. When fitting the stellar parameters only within 3σ (yellow area), at least 100 M ⊕   of heavy elements is still needed. In our planetary evolutions calculations, we assumed all heavy elements to be grouped into a well-defined central core, surrounded by a solar-composition envelope, with no added sources of heat. The possibility that these heavy elements may be at least partly mixed in the envelope is not expected to change these numbers significantly (Guillot 2005; Ikoma et al. 2006; Baraffe et al. 2008).

This extremely high amount of heavy elements is surprising for several reasons. First, this is probably a record – HD149026b, a Saturn-mass planet, has been known to possess about 60 − 70 M ⊕   of heavy elements (Sato et al. 2005; Ikoma et al. 2006). Other objects, such as OGLE-TR-56b and OGLE-TR-132b, were later found to have close to 100 M ⊕   in heavy elements (Guillot et al. 2006), but no planetary-mass object has yet been shown to possess a larger amount of heavy elements. An exception may be HAT-P-2b, initially understood to have MZ = 200 to 600 M ⊕   (Baraffe et al. 2008; Leconte et al. 2009), but a revision of the stellar parameters (Torres et al. 2008) yields much smaller MZ values for this object.

Second, planet formation models do not predict the existence of such dense objects (Mordasini et al. 2009). This is because a protoplanetary core growing beyond a few tens of Earth masses rapidly captures any surrounding hydrogen and helium, and the accretion of planetesimals is suppressed by the growth of the protoplanet beyond Saturn’s mass. Giant impacts would be necessary to capture a significant mass of heavy elements (Ikoma et al. 2006).

thumbnail Fig. 9

Age (in Ga = 109years) versus transit radius of CoRoT-13b (in Jupiter units, 1   RJup = 71492 km). The coloured area correspond to constraints derived from stellar evolution models matching the stellar density and effective temperature within a certain number of standard deviation intervals of 1σ (red), 2σ (blue) or 3σ (yellow). The curves are evolution tracks for CoRoT-13b (assuming M = 1.308 MJup, Teq = 1700 K), with various models being labelled.

Open with DEXTER

Third, CoRoT-13b is a counterexample outside the correlation between stellar metallicity and planetary mass in heavy elements (Guillot et al. 2006; Burrows et al. 2007; Guillot 2008). One interesting possibility however is related to the high lithium abundance of the star. If the metallicity of stars with planets did provide information about the late accretion of circumstellar gas filtered of their heavy elements by planet formation (Meléndez et al. 2009b; Ramírez et al. 2009; Nordlund 2009), then the CoRoT-13 system may be metal-rich, even though a heavy-element poor, lithium-rich rare last burst of accretion modified the chemical properties of the star’s thin outer convective zone.

5.3. Thermal losses

We estimated the thermal mass loss of CoRoT-13b by using the method and formulae described in Lammer et al. (2009), and found a negligible escape rate that did not influence the planet mass over its history. The reason for the negligible thermal mass loss of CoRoT-13b is the planet compactness and high density of about 2.34 g cm-3.

thumbnail Fig. 10

Expected amplitude of the occultation of the planet as a function of the wavelength for two extreme values for the heat redistribution factor for CoRoT-13b (as defined by Cowan & Agol 2010; ϵ = 1: uniform heat distribution over the planet; ϵ = 0 no distribution) and two values of the Bond albedo. The gray shaded area is the wavelength window observed by CoRoT. The black arrow is the upper limit of CoRoT observations.

Open with DEXTER

5.4. Occultation of the planet by the star

In the visible wavelength range, occultations of planets behind their host stars have been detected in only a few cases: Corot-1b (Snellen et al. 2009; Alonso et al. 2009a) and CoRoT-2b (Alonso et al. 2009b; Snellen et al. 2010) in data from CoRoT, and HAT-P-7 in data from Kepler (Borucki et al. 2009). All of these occultations are related to inflated giant planets (Rp > 1.4 RJup) on very short orbits, which means that they have large surfaces that tend to produce a large amount of reflected light and additionally are very hot (Teff > 2000 K), which is indicative of thermal emission in the visible regime. Using the definition from Cowan & Agol (2010), CoRoT-13b, in the most favourable case (zero albedo, zero heat redistribution factor represented by the upper solid line in Fig. 10), should have an average surface temperature of around 1700 K. The emission of the planet in the visible range observed by CoRoT is therefore dominated by reflection, meaning that eclipse amplitudes larger than 2.5 × 10-5 cannot be expected, even with a very high Bond albedo of AB = 0.5 (Fig. 10, dotted lines). This signal is not however detectable in the current data, as the measured scatter in the folded and binned light curve is on the order of 10-4.

thumbnail Fig. 11

Raw light curve of CoRoT-13b between the observed transits number 8 (HJD = 2454819.06) and 9 (HJD = 2454823.09) where a) describes the 8th transit and b) describes the event at HJD = 2454821.03, which mimics the transit of a long-period planetary companion.

Open with DEXTER

5.5. Constraints on the presence of additional planets.

The region of the light curve (LC) shown in Fig. 11, between the 8th and the 9th transits observed by CoRoT  shows a feature that mimics a single planetary transit that is not related to the pattern of transits of CoRoT-13b. Almost in the middle of the two regular transits, we find a short feature, roughly 1.5 h long and 1% deep, which in principle could be due to a planetary transit from an additional planet in the system. This feature does not repeat in the LC, hence a hypothetical companion would require a period longer than the 80 days that remained until either end of the observations. Nevertheless, the duration of a planetary transit is linked to the period of the orbit (Seager & Mallén-Ornelas 2003) and for an 80 day orbit we expect a transit of around 8 h long. Of course, some corrections have to be done in the case the orbit is eccentric (Carter et al. 2008; Kipping 2008); but in our case, we would have to assume an eccentricity of 0.94 (and a favourable orientation) for the hypothetical planet to account for a 1.5 h long transit in this long period orbit. However, this hypothetical planet would have a periastron passage inside the orbit of CoRoT-13b. There are studies of the planetary three-body problem with high eccentricities (see for example Beaugé & Michtchenko 2003; Michtchenko et al. 2006), but to the knowledge of the authors there is no possible justification of the dynamical stability of one system formed by a close-in Jupiter-like planet in a circular orbit and a highly eccentric Jupiter-sized companion.

thumbnail Fig. 12

Parameter space of the additional planets in the system that can been discarded on the basis of current HARPS measurements. The region above the curve is discarded by RV measurements, while planets located in the region below the curve would produce an undetectable signal.

Open with DEXTER

Moreover, there is no hint of the drift produced by any additional planet in the scatter of the residuals of the circular radial velocity (RV) analysis. The region of the parameter space of an additional planet in a circular orbit that can be discarded based on RV data is shown in Fig. 12. We note that for periods below 85 days (the span of HARPS observations), only planets in circular orbits with masses below  ~ 0.45 MJup are not discarded by RV data. The mass limit for eccentric orbits is even lower in this range of periods, as the amplitude K of the RV movement grows as (1 − e2) − 1 / 2.

Given these findings, the most simple explanation is an environmental effect that mimics a transit.

6. Summary

We have reported the discovery of CoRoT-13b, a transiting giant planet orbiting the star CoRoT 110839339. Our spectroscopic analysis indicates that CoRoT-13 is a G0V star with Teff = 5945 ± 90 K, M = 1.09 ± 0.02 M, R = 1.01 ± 0.03 R, solar metallicity ( [ M / H ]  =  + 0.01 ± 0.07), and a high relative abundance of lithium ( + 1.45 dex). The evolutionary tracks constrain the age of the star to be between 0.12 and 3.15 Gyr. The results of the study of the lithium abundance, the v sin i, and the activity level (which is low in both the spectroscopic and the photometric analyse) are consistent with the spectral type and the age range of the star. However, the lower limit to the interval is not favoured due to the low activity level measured.

CoRoT-13b is the highest-density jovian planet among its relatives with masses between 0.2 and 2.5 Jupiter masses. Its extreme density implies the existence of a significant amount, between 140 and 300 M ⊕ , of heavy elements in the planet, which contradicts the expected link between MZ of a planet and the metalliciy of its host stars (Guillot et al. 2006; Burrows et al. 2007; Guillot 2008). There is no hint of any other massive companion in the system, which is quite a common finding for hot Jupiters.

We have discussed the consequences of this characterization and we presented evidence that this particular target provides important information about the knowledge of the interactions between the structure of stars, the formation of planets, and their joint evolution (Ikoma et al. 2006; Ammler-von Eiff et al. 2009; Meléndez et al. 2009a; Nordlund 2009; Ramírez et al. 2009; Lanza 2010).


1

A lower limit, as the value of sini for the spin axis of the star is unknown.

Acknowledgments

The team at IAC acknowledges support by grant ESP2007-65480-C02-02 of the Spanish Ministerio de Ciencia e Innovación. This research has made use of the ExoDat database, operated at LAM-OAMP, Marseille, France, on behalf of the CoRoT/Exoplanet program. This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation. This research has made use of NASA’s Astrophysics Data System.

References

All Tables

Table 1

IDs, coordinates, and magnitudes of CoRoT-13.

Table 2

Closest contaminants to CoRoT-13b with their respective magnitudes and relative distances to the target.

Table 3

Abundances of 10 elements in CoRoT-13 relative to the Sun.

Table 4

Radial velocities measurements and bisector span velocities measured with HARPS.

Table 5

Values of the adjusted parameters in the modeling of the transit of CoRoT-13b in the different approximations described in the text.

Table 6

Planet and star parameters.

All Figures

thumbnail Fig. 1

Raw light curve of CoRoT-13b. The target was oversampled from 9 December 2008 (HJD 2454810.20); for visualization purposes, we superimpose the LC binned at a sampling rate of 512 s to the 32 s sampled region. The jumps in the data are caused by hot pixels as described in the text.

Open with DEXTER
In the text
thumbnail Fig. 2

Region of the sky around CoRoT-13. Left: image taken with INT/WFC with a resolution of 0.3′′. The position of the target is indicated with a white cross. Right: image taken by CoRoT with the same scale and orientation; the resolution is around 2.32′′/pixel. The area inside the black line is the mask used to compute the photometry; the position of the nearest contaminants from the ExoDat database are indicated with different types of crosses.

Open with DEXTER
In the text
thumbnail Fig. 3

Detail of the Keck spectrum of CoRoT-13 around the pressure-sensitive Ca ii line at 6162 Å. The synthetic spectrum is shown with a dashed line. Spectral lines used in the abundance analysis are marked with solid vertical lines, while the other spectral lines are marked with dotted lines.

Open with DEXTER
In the text
thumbnail Fig. 4

Mean abundances of 10 elements in CoRoT-13 Keck spectrum. White circles correspond to neutral lines and red boxes represent singly ionized lines. The yellow area represents the mean metallicity within one σ error bar.

Open with DEXTER
In the text
thumbnail Fig. 5

Upper panel: radial Velocity measurements phased to the orbital period measured by CoRoT. The solid curve represents the best-fit solution. Lower panel: residuals to the fit. In both panels, the white symbols indicate the measurements that have been corrected for moonlight contamination.

Open with DEXTER
In the text
thumbnail Fig. 6

Bisector analysis of the HARPS data. The white symbols represent measurements that have been corrected for moonlight contamination.

Open with DEXTER
In the text
thumbnail Fig. 7

Folded transit of CoRoT-13b, best-fit and its residuals using the values of solution A in Table 5.

Open with DEXTER
In the text
thumbnail Fig. 8

Position of CoRoT-13b (square) among the other transiting planets in a mass-radius diagram.

Open with DEXTER
In the text
thumbnail Fig. 9

Age (in Ga = 109years) versus transit radius of CoRoT-13b (in Jupiter units, 1   RJup = 71492 km). The coloured area correspond to constraints derived from stellar evolution models matching the stellar density and effective temperature within a certain number of standard deviation intervals of 1σ (red), 2σ (blue) or 3σ (yellow). The curves are evolution tracks for CoRoT-13b (assuming M = 1.308 MJup, Teq = 1700 K), with various models being labelled.

Open with DEXTER
In the text
thumbnail Fig. 10

Expected amplitude of the occultation of the planet as a function of the wavelength for two extreme values for the heat redistribution factor for CoRoT-13b (as defined by Cowan & Agol 2010; ϵ = 1: uniform heat distribution over the planet; ϵ = 0 no distribution) and two values of the Bond albedo. The gray shaded area is the wavelength window observed by CoRoT. The black arrow is the upper limit of CoRoT observations.

Open with DEXTER
In the text
thumbnail Fig. 11

Raw light curve of CoRoT-13b between the observed transits number 8 (HJD = 2454819.06) and 9 (HJD = 2454823.09) where a) describes the 8th transit and b) describes the event at HJD = 2454821.03, which mimics the transit of a long-period planetary companion.

Open with DEXTER
In the text
thumbnail Fig. 12

Parameter space of the additional planets in the system that can been discarded on the basis of current HARPS measurements. The region above the curve is discarded by RV measurements, while planets located in the region below the curve would produce an undetectable signal.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.