In the course of our ongoing multiplicity study of exoplanet host stars we detected a faint companion located at ∼43 arcsec (480 au physical projected separation) north-west of its primary – the exoplanet host star HD 3651 at 11 pc. The companion, HD 3651 B, clearly shares the proper motion of the exoplanet host star in our four images, obtained with the European Southern Observatory New Technology Telescope and the United Kingdom Infrared Telescope, spanning three years in epoch difference. The magnitude of the companion is H= 16.75 ± 0.16 mag, the faintest co-moving companion of an exoplanet host star imaged directly. HD 3651 B is not detected in the POSS-II B-, R- or I-band images, indicating that this object is fainter than ∼20 mag in the B- and R-band and fainter than ∼19 mag in the I-band. With the Hipparcos distance of HD 3651 of 11 pc, the absolute magnitude of HD 3651 B is about 16.5 mag in the H band. Our H-band photometry and the Baraffe et al. (2003) evolutionary models yield a mass of HD 3651 B to be 20 to 60 MJup (Jupiter masses) for assumed ages between 1 and 10 Gyr. The effective temperature ranges between 800 and 900 K, consistent with a spectral type of T7 to T8. We conclude that HD 3651 B is a brown dwarf companion, the first of its kind directly imaged as a companion of an exoplanet host star, and one of the faintest T dwarfs found in the solar vicinity (within 11 pc).
During the last decade, high-precision radial-velocity studies revealed almost 200 exoplanets around mainly solar-like stars located in the solar neighbourhood. The stellar multiplicity of these planet host stars was already investigated by several groups using imaging data from visible and infrared all-sky surveys such as the Palomar Observatory Sky Survey (POSS) or the Two Micron All Sky Survey (2MASS), see e.g. Bakos et al. (2006) or Raghavan et al. (2006). With seeing-limited, near-infrared imaging (see e.g. Mugrauer et al. 2005, 2006) as well as high-contrast, diffraction-limited Adaptive Optics (AO) observations (e.g. Els et al. 2001; Patience et al. 2002; Luhman & Jayawardhana 2002; Chauvin et al. 2006), further companions of exoplanet host stars have been detected during the last few years. Up to now, more than 30 companions of exoplanet host stars are known, suggesting that the multiplicity of these stars is at least 20 per cent.
Faint companions of exoplanet host stars were detected close to HD 114762 (Patience et al. 2002) and Gl 86 (Els et al. 2001) with angular separations smaller than 3 arcsec, which corresponds to physical projected separations of 130 and ∼20 au, respectively. In addition to these faint close objects, Wilson et al. (2001) reported a faint, wide companion to HD 89744, detected in the 2MASS point source catalogue (Skrutskie et al. 2006). HD 89744B is located 63 arcsec (∼2500 au physical projected separation) north-east of HD 89744 A, and shares the proper motion of this exoplanet host star (Mugrauer et al. 2004). While the faint companions HD 114762 B (MH∼ 10.4 mag) and HD 89744 B (MH∼ 11.1 mag) are both very low-mass objects with masses of about 0.080 M⊙, Gl 86 B (MH∼ 14.2 mag) turned out to be a white dwarf (see Mugrauer et al. 2005, for further details).
So far, substellar companions with masses significantly below the stellar/substellar mass border of 0.075 M⊙ have not been directly detected as companions to radial-velocity planet host stars, neither with AO nor seeing-limited imaging. In contrast, radial-velocity studies have already revealed a number of brown dwarf companions close to the exoplanet host stars e.g. HD 38529 (Fischer et al. 2003a), HD 168443 (Marcy et al. 2001), and HD 202206 (Correia et al. 2005), with minimum masses of m sin (i) = 12.7, 17.2 and 17.4 MJup, respectively. While the brown dwarf companions of HD 38529 and HD 168443 both revolve around the planet host star on orbits with semi-major axes which are at least 10 times wider than those of the detected exoplanets, the brown dwarf companion of HD 202206 moves around the planet host star inside the planetary orbit. All of these systems are interesting examples which might point out that planet and brown dwarf formation processes can occur around the same star at comparable distances (see Correia et al. 2005, for further discussion).
In this Letter we present the first directly imaged brown dwarf companion of an exoplanet host star which we detected in the course of our multiplicity study. We show the results of our near-infrared observations in Section 2 and discuss the properties of this new very faint companion in Section 3.
HD 3651 is a high proper motion nearby K0 dwarf, located in the northern constellation, Pisces. Its proper motion and parallax [μα cos (δ) =−461.09 ± 0.75 mas yr−1, μδ=−370.90 ± 0.61 mas yr−1, and π= 90.03 ± 0.72 mas] are determined by Hipparcos (Perryman et al. 1997), yielding a distance of ∼11 pc. According to Fischer et al. (2003b) HD 3651 is chromospherically inactive and photometrically stable, as expected for a middle-aged early K dwarf. The stellar radial velocity shows a periodical modulation with a period of 62.23 d, indicating the star is orbited by a sub-Saturn mass planet [m sin (i) = 0.20 MJup] with an eccentric (e= 0.63) orbit and a semi-major axis of a= 0.284 au. According to Santos, Israelian & Mayor (2004) HD 3651 exhibits an effective temperature of 5173 ± 35 K and a surface gravity of log (g) = 4.37 ± 0.12 cm s−2, as it is expected for an early K dwarf. Its metallicity is slightly enhanced compared to that of the Sun ([Fe/H]= 0.12 ± 0.04), typical for a planet host star. Santos et al. (2004) also derive the mass of HD 3651 to be 0.76 M⊙, consistent with the mass estimate of 0.79 M⊙ from Fischer et al. (2003b).
The age of this exoplanet host star has been estimated by several groups. Valenti & Fischer (2005) report an age between 3 and 12.5 Gyr, derived with isochron fitting. This range is consistent with the chromospheric age estimates of 5.9 Gyr from Wright et al. (2004) and 2.1 Gyr from Rocha-Pinto et al. (2004).
HD 3651 is a target of our imaging search campaign for visual companions of northern exoplanet host stars, which was carried out with the 3.8-m United Kingdom Infrared Telescope (UKIRT), located at Mauna Kea (Hawaii). We observed HD 3651 in the H-band with the UKIRT Fast Track Imager (UFTI), a 1024 × 1024 HgCdTe infrared detector with a pixel scale of ∼91 mas pixel−1 and a 93 × 93 arcsec field-of-view. The first epoch imaging is carried out in 2003 June. In order to reduce strong saturation effects of the bright exoplanet host star, the shortest available integration time (4 s) was used, and six of these 4-s integrations were averaged to one image. The standard jitter/dither technique was applied to minimize the sky background.
In total 22 jitter positions were chosen, resulting in a total integration time of 8.8 min. The European Southern Observatory (ESO) package eclipse (Devillard 2001) was used for background estimation and subtraction as well as the flat-fielding of all images which where finally combined by shift+add.
In addition to HD 3651 we observed many other exoplanet host stars in that run. With all these images we can derive the astrometric calibration of the UFTI detector using the 2MASS point source catalogue (Skrutskie et al. 2006), which contains accurate positions of objects brighter than 15.2 mag in H (signal-to-noise ratio S/N > 5). The derived pixel scale and detector orientation of all observing runs presented here are summarized in Table 1. The achieved detection limit (S/N > 10) of the UFTI observations is H∼ 18.4 mag, which is reached in the background noise limited region at angular separations larger than 15 arcsec around the bright exoplanet host star.
Five faint objects around HD 3651 are detected in our UFTI image, all of which might be real companions of the exoplanet host star. With a follow-up second epoch observation we tested the companionship of the detected companion-candidates. Real companions of the exoplanet host star would share the proper motion of their primary, as their orbital motion is much smaller than the proper motion of the exoplanet host star. Such co-moving companions can therefore be easily distinguished from non- or slowly moving background stars by comparing two images taken with sufficiently long time-difference.
The second epoch UFTI observation of HD 3651 was obtained in 2003 October. Again we averaged six 4-s integrations per jitter position but this time chose a 24-position jitter pattern, yielding a total integration time of 9.6 min.
Although the time-difference between the two UFTI observations is only 4 months, the astrometric companion search is already feasible because of the high proper motion of HD 3651, which amounts to 0.6 arcsec yr−1. All candidates except one significantly change their separations and position angles relative to the exoplanet host star. Only one candidate does not show a significant variation in its position angle and separation. Its relative astrometry is summarized in Table 1, and is illustrated in Fig. 2, later. This object might be a real companion of the exoplanet host star, and therefore is being denoted as HD 3651 B henceforth.
Due to the proximity of HD 3651, which is located only at a distance of 11 pc, the UFTI field-of-view allows only the detection of companions with physical projected separations up to ∼500 au from the exoplanet host star. In order to reach wider companions we include HD 3651 in our companion search program of southern exoplanet host stars and nearby stars, both being carried out at La Silla observatory with the ESO 3.58-m New Technology Telescope (NTT) and its infrared camera SofI, a 1024 × 1024 HgCdTe-detector. All observation were obtained again in the H-band using the SofI small-field objective with its pixel scale of 144 mas pixel−1, yielding a field-of-view of 147 × 147 arcsec. This allows the detection of wide companions around HD 3651 with physical projected separations up to ∼800 au. Our first epoch SofI observation of HD 3651 was carried out in 2004 July. The second epoch follow-up imaging was recently taken in 2006 June. In both observing epochs we used the jitter technique and obtained 10 images, each the average of 50 1.2-s integrations, i.e. 10 min of total integration time. Our second epoch SofI image is shown in the left-hand panel of Fig. 1.
Both SofI images are calibrated with the 2MASS point source catalogue (see Table 1). HD 3651 B and the four objects already detected in the UFTI images as well as several more wider companion candidates, are found in the SofI images. However, all of these additional companion candidates exhibit only negligibly small proper motions, determined by the comparison of both SofI images. In contrast, HD 3651 B clearly shares the proper motion of HD 3651 A, and its separation and position angle does not change during three years of epoch difference between the first UFTI observation (epoch 06/03) and the second epoch SofI imaging (epoch 06/06), hence this is a real companion of the exoplanet host star (see Fig. 2).
The reached detection limit of our SofI observations for a range of separations to the planet host star is shown in Fig. 3. We are sensitive to companions with apparent H-band magnitudes of 17.5 mag in the background noise limited region, i.e. at angular separations larger 16 arcsec (∼180 au) around the bright exoplanet host star. According to the Baraffe et al. (2003) models and the magnitude–mass relation therein, the achieved limiting magnitude allows the detection of brown dwarf companions with a mass m≥ 37 MJup for an assumed age of the primary of 5 Gyr. Additional stellar companions (m≥ 75 MJup) can be ruled out beyond 4.5 arcsec (∼50 au) up to separations of ∼760 au. All objects within 69 arcsec around the exoplanet host star are imaged in both SofI observing epochs but, except HD 3651 B, no further co-moving companions are detected.
We determine the apparent H-band magnitude of the co-moving companion HD 3651 B in all UFTI and SofI images, all of which are listed in Table 1. We measure the companion magnitude with aperture photometry using the data reduction package eso-midas. The photometric zero-points of the observations are derived with sources detected in our UFTI and SofI images that have H-band magnitude listed in the 2MASS point source catalogue. The averaged apparent H-band magnitude of the companion is H= 16.75 ± 0.16 mag. The Hipparcos parallax π= 90.03 ± 0.72 mas of the exoplanet host star finally yields an absolute magnitude of MH= 16.52 ± 0.16 mag.
On the Nature of HD 651 B
Our H-band multi-epoch UFTI and SofI observations of the exoplanet host star HD 3651 revealed the faint co-moving companion HD 3651 B, separated from its primary by 43 arcsec, corresponding to a physical projected separation of ∼480 au. With an apparent H-band magnitude of H= 16.75 ± 0.16 mag, HD 3651 B is the faintest co-moving companion of an exoplanet host star detected so far. Up to now the faintest known companion of an exoplanet host star is Gl 86 B, whose absolute H-band magnitude was measured by Els et al. (2001) to be MH= 14.2 ± 0.2 mag. With its absolute H-band magnitude of MH∼ 16.5 mag, HD 3651 B is 2.3 mag fainter than Gl 86 B.
HD 3651 B is detected in our near-infrared UFTI and SofI images but we do not find a visible counterpart on the photographic B-, R- and I-band plates of the second Palomar All Sky Survey (POSS-II). The POSS-II R-band plate is shown in the right-hand panel of Fig. 1. According to Griffin (2002) the detection limit of the POSS-II plates are 22.5 mag in B-band, 20.8 mag in R-band and 19.5 mag in I-band. Hence, HD 3651 B has to be fainter than ∼20 mag in R- and B-band and fainter than ∼19 mag in the I-band to remain undetectable on all POSS plates.
HD 3651 B is a faint source in the near-infrared H-band and it is not detectable in the B-, R- and I-bands. Due to its faintness in the optical spectral range compared to its near-infrared H-band photometry, HD 3651 B is not a cool white dwarf that is expected to be comparably bright in all photometric bands (see Bergeron, Saumon & Wesemael 1995), i.e. it should be detectable on the POSS plates. In contrast, according to the evolutionary models of low-mass substellar objects from Baraffe et al. (2003) all these photometric results are fully consistent with a cool brown dwarf located at the distance of the exoplanet host star. With this models we can also derive the physical properties of HD 3651 B from the measured H-band absolute magnitude assuming different system ages. In Table 2 we have summarized the derived companion masses and effective temperatures as well as the expected absolute R- and I-band magnitudes for assumed ages of 1, 5 and 10 Gyr (B-band magnitudes not given in theses models). The mass ranges from 22 up to 57 MJup for an age of 1 and 10 Gyr, respectively. The expected effective temperature of the brown dwarf ranges between 800 and 900 K. According to Golimowski et al. (2004), the derived temperature range is consistent with a spectral type T8. Furthermore, if we use the MH versus spectral type relation, derived by Vrba et al. (2004), the absolute magnitude of HD 3651 B is consistent with a brown dwarf of spectral type T7.9. Follow-up spectroscopy would be interesting to confirm the spectral type.
After Gl 229 B (Nakajima et al. 1995), ε Ind Bb (McCaughrean et al. 2004), Gl 570 D (Burgasser et al. 2000) and SCR 1845-6357 B (Biller et al. 2006), HD 3651 B is a further late T dwarf companion of a star located in the solar neighbourhood. Its companionship to an exoplanet host star is especially remarkable and puts news constraints to the formation theory of brown dwarfs and extrasolar planets. This finding is a further evidence that both formation process can occur around the same object.
We would like to thank the technical staff of the ESO NTT and Joint Astronomy Center, Hawaii, for all their help and assistance in carrying out the observations. We made use of the Simbad database operated at the Observatoire Strasbourg. This work was partly supported by the Israel Science Foundation (grant no. 233/03).