A&A 466, 1179–1183 (2007) Astronomy DOI: 10.1051/0004-6361:20066835 & c ESO 2007 Astrophysics No evidence of a hot Jupiter around HD 188753 A, A. Eggenberger1,S.Udry1, T. Mazeh2,Y.Segal2, and M. Mayor1 1 Observatoire de Genève, 51 Ch. des Maillettes, 1290 Sauverny, Switzerland e-mail: [email protected] 2 School of Physics and Astronomy, Raymond and Beverly Sackler Faculty of Exact Sciences, Tel Aviv University, Tel Aviv, Israel Received 28 November 2006 / Accepted 25 January 2007 ABSTRACT Context. The discovery of a short-period giant planet (a hot Jupiter) around the primary component of the triple star system HD 188753 has often been considered as an important observational evidence and as a serious challenge to planet-formation theories. Aims. Following this discovery, we monitored HD 188753 during one year to better characterize the planetary orbit and the feasibility of planet searches in close binaries and multiple star systems. Methods. We obtained Doppler measurements of HD 188753 with the ELODIE spectrograph at the Observatoire de Haute-Provence. We then extracted radial velocities for the two brightest components of the system using our multi-order, two-dimensional correlation algorithm, TODCOR. Results. Our observations and analysis do not confirm the existence of the short-period giant planet previously reported around HD 188753 A. Monte Carlo simulations show that we had both the precision and the temporal sampling required to detect a planetary signal like the one quoted. Conclusions. From our failure to detect the presumed planet around HD 188753 A and from the available data on HD 188753, we conclude that there is currently no convincing evidence of a close-in giant planet around HD 188753 A. Key words. techniques: radial velocities – stars: binaries: spectroscopic – stars: individual: HD 188753 – stars: planetary systems 1. Introduction maintained its current orbital configuration ever since the planet formed, explaining the existence of a hot Jupiter around the pri- In a recent paper, Konacki (2005a) has reported the discovery mary component of HD 188753 remains a challenge (Nelson of a 3.35-day radial-velocity modulation of HD 188753 A with −1 2000; Mayer et al. 2005; Boss 2006; Jang-Condell 2007) and a semiamplitude of 149 m s , and attributed this signal to the might indicate that the core accretion mechanism is not able to presence of a planet with a minimum mass of 1.14 MJup in or- account for the existence of all the planets discovered so far. bit around the star. Interestingly, HD 188753 A is the primary An alternative way to explain the existence of a close-in component of a triple star system, while the visual companion, Jovian planet around HD 188753 A and also to circumvent the HD 188753 B, is itself a spectroscopic binary with a period of theoretical problem is to assume that the current orbital con- 155 days (Griffin 1977; Konacki 2005a). The visual orbit of the figuration is the result of a dynamical process (Pfahl 2005; AB pair is characterized by a period of 25.7 years, a semimajor Portegies Zwart & McMillan 2005). According to this approach, axis of 12.3 AU (0.27 separation), and an eccentricity of 0.5 HD 188753 A was a single star or had a more distant stellar com- (Söderhjelm 1999). HD 188753 is therefore a hierarchical sys- panion at the planet formation phase, and this initial state was tem, with the primary component (A) hosting the giant planet later transformed through a dynamical encounter into the triple and the secondary (Ba) and tertiary (Bb) components forming a system we presently observe. Pfahl & Muterspaugh (2006) esti- close pair in orbit at a distance of 12.3 AU from the planet-host mate that dynamical interactions could deposit giant planets in star. about 0.1% of the binaries with a semimajor axis a < 50 AU. The The discovery of a giant planet around HD 188753 A has at- existence of a short-period Jovian planet around HD 188753 A tracted much attention, as the proximity of HD 188753 B poses might thus be the result of a dynamical process, rather than of a serious problem for planet formation theories. Indeed, the pe- an in situ formation (Pfahl 2005; Portegies Zwart & McMillan riastron distance of the visual binary is only 6.2 AU, so that any 2005; Jang-Condell 2007). protoplanetary disk around HD 188753 A would be truncated at Among the ∼200 extrasolar planets discovered so far by ∼2 AU (Konacki 2005a; Jang-Condell 2007), i.e. probably inside Doppler spectroscopy, 33 are known to orbit the components the snow line. The favored core accretion model, which stipu- of binaries or triple stars (Eggenberger et al. 2004; Mugrauer lates that the cores of giant planets form beyond the snow line, et al. 2005; Raghavan et al. 2006). However, unlike the giant might thus have a problem accounting for the existence of a gi- planet around HD 188753A, most of these planets reside in sys- ant planet around HD 188753 A. Assuming that the system has tems with separations larger than 100 AU. Since double stars ffi Based on observations collected at the Observatoire de Haute- closer than 2–6 present observational di culties for Doppler Provence with the ELODIE echelle spectrograph mounted on the studies, they have commonly been left out of radial-velocity 1.93-m telescope. planet searches. The occurrence of planets in binaries with semi- Table 1 is only available in electronic form at major axes below 100 AU is therefore still largely unprobed. http://www.aanda.org Despite this strong bias and prior to the discovery by Konacki, Article published by EDP Sciences and available at http://www.aanda.org or http://dx.doi.org/10.1051/0004-6361:20066835 1180 A. Eggenberger et al.: No evidence of a hot Jupiter around HD 188753 A three planets were found in stellar systems with separations correlation features. This leads to a situation where the two cor- close to 20 AU: Gliese 86 (Queloz et al. 2000; Els et al. 2001; relation features are always strongly blended, the blend changing Mugrauer & Neuhäuser 2005; Lagrange et al. 2006), γ Cephei continuously due to the 155-day modulation. (Hatzes et al. 2003; Neuhäuser et al. 2007; Torres 2007), and Given the double-lined nature of HD 188753 and the strong HD 41004 (Zucker et al. 2003, 2004). Two features render the line blending, we derived radial velocities using a multi-order planet around HD 188753A particularly interesting. First, both version (Zucker et al. 2003) of the two-dimensional correla- the planetary and the binary orbital parameters are known (only tion algorithm TODCOR (Zucker & Mazeh 1994). This algo- γ Cephei shares this property). Second, the binary periastron dis- rithm uses two templates with a given flux ratio and unknown tance may be small enough to preclude giant planet formation Doppler shifts to compute the two-dimensional correlation func- according to the canonical models. tion, whose maximum simultaneously gives the radial veloc- Doppler searches for planets in close binaries have recently ity of both components. Our version of TODCOR uses as tem- proven feasible using dedicated reduction techniques based plates high signal-to-noise stellar spectra built up using spectra on two-dimensional correlation (Zucker et al. 2003; Konacki from our planet search programs with CORALIE and ELODIE 2005b). In order to probe the occurrence of planets in close (Queloz et al. 2000; Perrier et al. 2003), along with spectra from stellar systems, two surveys searching for planets in spec- the surveys for low-mass companions to M dwarfs by Delfosse troscopic binaries are currently underway (Eggenberger et al. and coworkers (Delfosse et al. 1998, 1999). Each template can 2003; Konacki 2005b), the planet around HD 188753A being furthermore be convolved with a rotational broadening profile, a product of Konacki’s survey. Yet, deriving the velocity of thus allowing for a better fit to the observed spectrum. In the HD 188753 A to the precision needed to reveal the presence multi-order version of TODCOR, the flux ratio is a function of of a planet is particularly challenging, as the radial velocity of wavelength and is calculated for each order according to the the secondary, HD 188753 Ba, varies with a semiamplitude of spectral types of the two templates using the library of spec- ∼13 km s−1 over a timescale of 155 days. tral energy distributions by Pickles (1998). To insure fine-tuning In order to study this intriguing system further, we moni- with the observed system, the table of flux ratios can also be tored HD 188753 during one year with the ELODIE spectro- multiplied by a global normalization factor. graph (Baranne et al. 1996). We then used our multi-order To derive the radial velocities of HD 188753 A and Ba, we TODCOR algorithm (Zucker et al. 2003) to derive the radial ve- ran TODCOR for a variety of different pairs of templates from locities of the two brightest components, namely HD 188753 A our library, searching for the pair that best matches our observed and HD 188753 Ba. Unfortunately, our data and analysis do not composite spectra. The two templates finally retained were the confirm the existence of the planet reported by Konacki (2005a) spectrum of HD 224752 (a G6 dwarf) for HD 188753 A and the around HD 188753 A. We present our observations and data re- spectrum of HD 225208 (a K0 dwarf) for HD 188753 Ba. This duction technique in Sect. 2. Our results and the lack of evidence template configuration was chosen because (i) it gave the low- of a hot Jupiter around HD 188753 A are described in Sect.