R. Da Silva¹ - S. Udry¹ - F. Bouchy² - M. Mayor¹ - C. Moutou² - F. Pont¹ - D. Queloz¹ - N. C. Santos^3,1 - D. Ségransan¹ - S. Zucker^1,4

1 - Geneva Observatory, 1290 Sauverny, Switzerland
2 - Observatoire de Marseille, France
3 - Centro de Astronomia e Astrofísica da Universidade de Lisboa, Tapada da Ajuda, 1349-018 Lisboa, Portugal
4 - The Weizmann Institute of Science, PO Box 26, Rehovot 76100, Israel

Abstract
We report the discovery of a new planet candidate orbiting the subgiant star HD 118203 with a period of P = 6.1335 days. The best Keplerian solution yields an eccentricity e = 0.31 and a minimum mass $m_2\sin{i}$ = 2.1 $M_{\rm Jup}$ for the planet. This star has been observed with the ELODIE fiber-fed spectrograph as one of the targets in our planet-search programme biased toward high-metallicity stars, on-going since March 2004 at the Haute-Provence Observatory. An analysis of the spectroscopic line profiles using line bisectors revealed no correlation between the radial velocities and the line-bisector orientations, indicating that the periodic radial-velocity signal is best explained by the presence of a planet-mass companion. A linear trend is observed in the residuals around the orbital solution that could be explained by the presence of a second companion in a longer-period orbit. We also present here our orbital solution for another slightly evolved star in our metal-rich sample, HD 149143, recently proposed to host a 4-d period Hot Jupiter by the N2K consortium. Our solution yields a period P = 4.09 days, a marginally significant eccentricity e = 0.08 and a planetary minimum mass of 1.36 $M_{\rm Jup}$ . We checked that the shape of the spectral lines does not vary for this star as well.

Key words: stars: individual: HD 118203 - stars: individual: HD 149143 - planetary systems - techniques: radial velocities

1 Introduction

Stars hosting planets are significantly metal rich in comparison to field stars in the solar neighbourhood (Fischer & Valenti 2005; Santos et al. 2003,2005,2001; Gonzalez 1997). These authors have also shown that the probability of hosting a giant planet is a strongly rising function of the star metal content. According to their estimate, we can expect that up to 25-30% of the more metal-rich stars ([Fe/H] > 0.2-0.3) host a giant planet.

On the basis of this argument, a new survey was started in March 2004 at the Haute-Provence Observatory with the high-precision ELODIE fiber-fed echelle spectrograph. The main idea of this new programme is to bias our target sample towards high-metallicity stars, much more likely to host planets. This will strongly increase our probability of finding new planets in a sample of yet non-observed stars. The survey uses the cross-correlation technique for the radial-velocity and metallicity estimates. The programme mainly targets giant planets with short periods (Hot Jupiters). They are the ideal candidates in the search for photometric transits. From this survey, we present here a new short-period planet candidate (P = 6.1335 d) with an eccentric orbit around the star HD 118203.

A similar planet-search programme was simultaneously started by the N2K consortium (Fischer et al. 2004) aiming at the detection of short-period planets orbiting metal-rich stars. Fischer (2005; also in Fischer 2005) recently announced two new Hot-Jupiter detections around HD 149143 and HD 109749. HD 149143 is amongst the stars already observed in our sample and we present here our orbital description of the system.

Together, the N2K and ELODIE metallicity-biased planet-search programmes have detected five new Hot Jupiters in less than one year. One of them, HD 149026, is transiting in front of its parent star (Sato et al. 2005) allowing for the determination of the planet radius and mean density. The planet is found to have an unexpectedly large core. This result clearly illustrates the importance of such programmes for our understanding of planet interiors. However, when examining possible statistical trends between orbital and stellar parameters to derive constraints for planet formation models, we have to keep in mind the built-in bias of this subsample of exoplanets. In particular these planets must be removed when considering correlations with the star metallicity.

The sample selection and observations of the new ELODIE programme are described in Sect. 2. Stellar parameters, radial-velocity measurements of HD 118203 as well as the orbital solution derived for the new Hot Jupiter candidate are presented in Sect. 3. This section also provides information on stellar activity and in particular the results of the bisector analysis for the star. Section 4 reports our results for HD 149143 and conclusions are presented in Sect. 5.

2 Sample and observational strategy

Observations were conducted with the high-precision ELODIE fiber-fed echelle spectrograph (Baranne et al. 1996) mounted on the 1.93-m telescope at the Haute-Provence Observatory (France). The spectra have a resolution ( $\lambda/\Delta\lambda$ ) of about 42 000. Typical signal-to-noise ratios obtained in 20 min exposures range from $\sim$ 20 to 100 for the programme stars, corresponding to photon-noise errors between 5 and 20 ms^-1 on individual measurements. The data reduction is performed on-line during the observations by the automatic reduction software (see Baranne et al. 1996, for a detailed description).

After a single radial-velocity measurement, it is possible to obtain a very good estimate of the star metallicity by measuring the surface of the cross correlation function (CCF) of the ELODIE spectra (Naef 2003; Santos et al. 2002). The typical uncertainty of the resulting metallicities is $\sim$ 0.05 dex compared to values obtained from a high-resolution spectroscopic analysis. The expected percentage of >10-30% of giant planets orbiting stars with [Fe/H] $\geq$ 0.10 (Fischer & Valenti 2005; Santos et al. 2001,2004) should lead to the discovery of a few tens of new planets in our programme, about ten of them in short-period orbits.

Table 1: Observed and estimated parameters for HD 118203. (See text for references on the quoted values.)

3 A short-period planet orbiting HD 118203

3.1 Stellar characteristics of HD 118203

HD 118203 (HIP 66192) is listed in the HIPPARCOS catalogue (ESA 1997) as a K0 dwarf in the northern hemisphere, with a visual magnitude V=8.05, a colour index B-V = 0.699, and an astrometric parallax $\pi$ = 11.29 $\pm$ 0.82 mas, setting a distance of 88.6 pc from the Sun. The corresponding absolute magnitude M_V = 3.31 is too high for a K0 dwarf, indicating that the star is already slightly evolved and is in a subgiant stage. This is confirmed by the astrometric surface gravity $\log{g}$ = 3.87 dex estimated from the effective temperature and stellar mass (see below) and Hipparcos precise parallax (Santos et al. 2004).

From calibrations of the width and the surface of the ELODIE cross-correlation functions (Naef 2003; Santos et al. 2002), we have estimated a projected rotation velocity $v\sin{i}$ = 4.7 km s^-1 and a metallicity [Fe/H] = 0.10 for the star. Comparing these observable stellar characteristics with the stellar evolution models of Girardi et al. (2002), we can infer the following intrinsic properties for HD 118203: a coarse estimate of the effective temperature $T_{\rm eff}$ = 5600 $\pm$ 150 K, a mass $M_\star$ = 1.23 $\pm$ 0.03 $M_{\odot}$ and an age of 4.6 $\pm$ 0.8 Gyr. A similar value of the effective temperature $T_{\rm eff}$ = 5695 $\pm$ 50 K is also derived from the calibration in Santos et al. (2004) using the color index and metallicity of the star. The stellar parameters are gathered in Table 1.

$\begin{figure} \par\includegraphics[width=8cm,clip]{4116fig1.eps} \end{figure}$	Figure 1: $\lambda 3968.5$ Å Ca II H absorption line region of the summed ELODIE spectra for HD 118203. No clear emission feature is observed. For clarity, spectral features due to pollution by the thorium spectrum have been removed (e.g. around $\lambda 3966.5$ Å).
Open with DEXTER

A check for stellar chromospheric activity was also performed by looking at the Ca II H absorption line in the spectra. Figure 1 shows the co-added ELODIE spectra where the clear absence of an emission feature indicates a low activity level, as expected for slightly evolved subgiants (Wright 2004). The activity-induced radial-velocity jitter is thus also expected to be low for this moderately rotating star.

3.2 Orbital solution for HD 118203 b

ELODIE observations of HD 118203 were conducted from May 2004 (JD = 2 453 151) to July 2005 (JD = 2 453 553), stimulated by the difference in radial velocity found among the first three observations. A set of 43 precise radial-velocity measurements were then gathered. They are provided in Table 2.

A clear 6.1-d periodic variation is seen in this data set. The residuals around a Keplerian fit with this period are however unexpectedly large (>40 ms^-1) and shows a clear additional radial-velocity drift as a function of the Julian date. A simultaneous Keplerian + linear-drift adjustment then provides a very satisfactory solution for the system with a period P = 6.1335 $\pm$ 0.0006 days and an eccentricity e = 0.309 $\pm$ 0.014. The slope of the linear drift is found to be 49.7 $\pm$ 5.7 m ${\rm s}^{-1}~{\rm yr}^{-1}$ . It is most probably accounted for by the presence of a second companion in the system, on a longer-period orbit. This could explain the significantly non-zero value found for the eccentricity at such a short period.

A plot with the last-season radial-velocity measurements of HD 118203 is shown in Fig. 2, together with the derived solution. The residuals around the solution are displayed as well in the bottom panel of the figure. The phase-folded radial-velocity curve with the complete set of data points corrected for the observed linear drift is displayed in Fig. 3. The weighted rms around the solution is $\sigma$ (O-C) = 18.1 ms^-1, slightly larger than the individual photon-noise errors ( $\sim$ 15 ms^-1). The orbital elements are listed in Table 3 with the inferred planetary parameters.

Table 2: ELODIE radial velocities of HD 118203. All data are relative to the solar system barycentre. Given uncertainties correspond to photon-noise errors.

3.3 Low chromospheric activity for HD 118203

The bisector inverse slope (BIS value) computed from the HD 118203 spectra are plotted in Fig. 4 in comparison to the corresponding radial-velocities. If the line-shape and the radial-velocity variations share the same origin, the BIS and radial velocities are expected to be correlated. This is not the case. Furthermore, no coherent signal is observed when the BIS are phased with the orbital period of 6.1335 days.

$\begin{figure} \par\includegraphics[width=8cm,clip]{4116fig2.eps} \end{figure}$	Figure 2: Top: last 5-months radial-velocity measurements of HD 118203 superimposed on the best Keplerian + linear drift solution indicative of a second longer-period companion in the system. Bottom: residuals around the solution. Error bars represent the photon-noise errors.
Open with DEXTER

Moreover, data gathered during the still negative search for a photometric transit (presented elsewhere) show that the star is constant at a 0.0047 mag level and thus further support its low-activity level. Consequently, the explanation of the observed radial-velocity variations due to stellar activity can be discarded.

$\begin{figure} \par\includegraphics[width=8cm,clip]{4116fig3.eps} \end{figure}$	Figure 3: Top: phase-folded ELODIE radial-velocity measurements for the star HD 118203 after correction for the linear radial-velocity drift observed in the data. Bottom: corresponding residuals around the solution. Error bars represent the photon-noise errors.
Open with DEXTER

Table 3: ELODIE best Keplerian orbital solution obtained for HD 118203 as well as the inferred planetary parameters. For eccentric orbits T is defined as the time of the peri-astron passage.

$\begin{figure} \par\includegraphics[width=6.5cm,clip]{4116fig4.eps} \end{figure}$	Figure 4: Radial velocities (RV, upper panel) and inverse bisector slope (BIS, middle panel) phased with the orbital period P = 6.1335 d for HD 118203. The independence of the two quantities is shown by the radial velocity vs. BIS plot ( bottom panel).
Open with DEXTER

4 A Hot Jupiter around HD 149143

During the preparation of this publication, another new Hot Jupiter candidate was identified in our programme, orbiting the star HD 149143. While gathering more points to characterize the system, we learned that the planet had just been announced by the N2K consortium (Fischer et al. 2005; Fischer 2005). We thus present here our ELODIE solution for the system, relying on the N2K photometric measurements for the transit non-detection.

Table 4: Observed and estimated parameters for HD 149143. (See text for references on the quoted values.)

Table 5: ELODIE radial velocities of HD 149143. All data are relative to the solar system barycentre.

Table 6: ELODIE derived Keplerian orbital solution obtained for HD 149143 as well as the inferred planetary parameters. Both the quasi-circular and circular solutions are given because the derived eccentricity is only marginally significant. For the eccentric orbit T is defined as the time of the peri-astron passage whereas for the circular orbit T indicates the maximum of radial velocities.

4.1 HD 149143: stellar characteristics

HD 149143 is listed in the Hipparcos catalogue (HIP 81022) with a G0 spectral type, a visual magnitude V = 7.90, and a colour index B - V = 0.68. From calibrations of the width and the surface of the ELODIE cross-correlation functions we have estimated a projected rotation velocity $v\sin{i}$ = 3.9 km s^-1 and a metallicity [Fe/H] = 0.2.

$\begin{figure} \par\includegraphics[width=8cm,clip]{4116fig5.eps} \end{figure}$	Figure 5: ELODIE radial velocities of HD 149143 superimposed on the derived Keplerian quasi-circular model. Residuals around the solution are displayed in the bottom panel. Error bars represent the photon-noise errors.
Open with DEXTER

$\begin{figure} \par\includegraphics[width=8cm,clip]{4116fig6.eps} \end{figure}$	Figure 6: Phase-folded ELODIE radial velocities of HD 149143 superimposed on the derived quasi-circular Keplerian solution. Error bars represent the photon-noise errors.
Open with DEXTER

The measured Hipparcos parallax of HD 149143 is $\pi$ = 15.75 $\pm$ 1.07 mas setting the star 63.5 pc away from the Sun. The inferred absolute magnitude is then M_V = 3.88, also slightly overluminous for a metallic G0 dwarf. This is confirmed by the astrometric estimate of $\log{g}$ = 4.10 dex. From the models of Girardi et al. (2002) we obtain an effective temperature $T_{\rm eff}$ = 5730 $\pm$ 150 K, a primary mass M₁ = 1.1 $\pm$ 0.1 $M_{\odot}$ and an age of 7.6 $\pm$ 1.2 Gyr. A calibration of the stellar effective temperature based on the stellar colour and metallicity (Santos et al. 2004, only slightly sensitive to the gravity of the star) leads then to $T_{\rm eff}$ = 5790 $\pm$ 50 K, compatible with the value given above. The stellar parameters are given in Table 4.

4.2 Orbital solution for HD 149143 b

Although the number of ELODIE observations is small for this star (Table 5), thanks to the large amplitude of the radial-velocity variation (K = 163 ms^-1), we can easily derive a quasi-circular Keplerian solution (e = 0.08 $\pm$ 0.04) with a period of 4.088 days, in agreement with the Fischer (2005) and Fischer et al. (2005) announcement. The eccentricity of the orbit is small and marginally significant, we thus also provide the corresponding circular solution fixing e=0 (Table 6). The derived orbital elements coupled with the above estimate of the primary mass of 1.1 $M_{\odot}$ lead to a minimum mass $m_2\sin{i}$ = 1.36 $M_{\rm Jup}$ and a separation of 0.052 AU for the planetary companion.

Figures 5 and 6 present the temporal and phased ELODIE radial-velocity measurements superimposed on the derived Keplerian model as well as the residuals around the solution. The inferred orbital and planetary parameters are given in Table 6.

As for the case of HD 118203, the bisector inverse slope of the cross-correlation function has been calculated. No correlation is found between these bisector slopes and the radial velocities, excluding activity-induced variations of the shape of the spectral lines as the source of the radial-velocity variations.

5 Summary and concluding remarks

We have presented the characteristics of a new planet candidate in orbit around the subgiant star HD 118203, detected by the new ELODIE planet-search programme biased towards metal-rich stars. The planet is in a rather eccentric orbit (e = 0.31), with a period of P = 6.1335 days, and is close to its parent star (a = 0.06 AU).

An additional trend of the radial-velocity measurements increasing as a function of Julian date with a slope of 49.7 ms $^{-1}~{\rm yr}^{-1}$ is observed, suggesting the presence of a second companion around the main star on a longer-period orbit. This additional companion could explain the somewhat high eccentricity of the orbit of this new planet at short period.

We have also reported the ELODIE solution for another Hot Jupiter in our programme, HD 149143, recently announced by the similar-goal N2K project. The derived best Keplerian quasi-circular solution presents a period of 4.09 days and the inferred planetary mass is 1.36 $M_{\rm Jup}$ .

By selection these planets are orbiting metal-rich stars. They increase to five the number of Hot Jupiters detected in less than one year by dedicated metallicity-biased programmes. One of these candidates, HD 149026 (Sato et al. 2005), transits in front of its parent star and thus allows the determination of its radius and mean density when combining photometric and radial-velocity measurements. This demonstrates the efficiency of such approaches to find candidates suitable for constraining planet-interior models. However, the built-in biases of the sample have to be kept in mind when examining possible statistical relations between the star metallicity and other orbital or stellar parameters.

Elodie metallicity-biased search for transiting Hot Jupiters

I. Two Hot Jupiters orbiting the slightly evolved stars HD 118203 and HD 149143