THE ULTRA COOL BROWN DWARF COMPANION OF WD 0806–661B: AGE, MASS, AND FORMATION MECHANISM

The first examples of the low-temperature L- and T-spectral classes were discovered as companions to stars (Becklin & Zuckerman 1988; Nakajima et al. 1995). The upper temperature limit for a Y-type object is not known; though it is expected to be <500 K, when water vapor is predicted to condense (see Figure 2 in Burrows et al. 2003). Wide-field searches for low-temperature T (and Y) dwarfs are currently in progress and several very cool objects have been found at temperatures of ∼570 K (Burningham et al. 2009), 550–500 K (Leggett et al. 2009), 500–400 K (Lucas et al. 2010), and <500 K (Eisenhardt et al. 2010). However, the spectra of those objects (when available) do not differ substantially from late T dwarfs. A ∼300 K object comoving with WD 0806–661 could represent the coolest substellar companion observed to date (Luhman et al. 2011). While no spectrum has yet been obtained, this could very well be the first example of the Y-spectral class.

This Letter presents an upper limit at J-band to the substellar companion to WD 0806-661 (GJ 3483). Because the 'WD' designation is used exclusively for white dwarfs (E. Sion 2011, private communication), we suggest the companion be referred to as GJ 3483B rather than WD 0806-661B, as adopted in the discovery paper (Luhman et al. 2011). We will use GJ 3483 and GJ 3483B throughout this Letter. We discuss what constraints our upper limit places on the object's mass and temperature as well as how GJ 3483B ties in to other imaged low-mass companions. This object, and similar low-mass companions, probably formed at large separations from their primary star via Jeans-mass fragmentation (Low & Lynden-Bell 1976) rather than been scattered out via planet-planet interactions.

As part of an ongoing imaging campaign to search for wide separation low-mass companions to nearby, old white dwarfs, we observed GJ 3483 using the Infrared Side Port Imager (ISPI; van der Bliek et al. 2004) on the CTIO 4 m Blanco telescope. Observations were carried out on UT 2009 March 16–17 and 2010 April 3 & 5, each lasting 3240 s (3 sets of 3 × 3 dithers with individual exposure times of 120 s) and covering a field of view of approximately 10' × 10'. Conditions were generally good with seeing ∼1'' at J band (∼13 on 2009 March 17) and clear (except for thin clouds on 2009 March 16). Data reduction was carried out in the usual manner using standard IRAF routines.⁵ The brown dwarf companion observed by Spitzer (Luhman et al. 2011) was not detected in any of the individual nights.

To obtain a deeper J-band limit, we co-added data from the four nights together. Because the white dwarf has proper motion of (+340.3, −289.6) mas yr⁻¹ (Subasavage et al. 2009), we shifted the 2009 data by 1 pixel in both R.A. and decl. to match the white dwarf's motion (the ISPI plate scale is 03 pixel⁻¹). Individual nights were weighted by the square of the signal-to-noise of well-detected Two Micron All Sky Survey (2MASS) stars. Our co-added, 3.5 hr image is displayed in Figure 1 alongside the 2009 4.5 μm Spitzer field from program 60160 (M. Burleigh) and it is clear that we do not detect the companion brown dwarf. To estimate a limiting magnitude, we generated synthetic stars in 0.1 mag bins ranging from J = 20.6–22.6 using IRAF/DAOPHOT's addstar routine with a point-spread function (PSF) derived from stars in our field. PSF fitting was performed on these synthetic stars, which allows us to estimate a 3σ limit of J = 21.7 for this co-added exposure, 1.7 mag fainter than the limit provided by Luhman et al. (2011).

Zoom In Zoom Out Reset image size

With our upper limit of J = 21.7, we estimate J − [4.5] > 4.95, nearly 2 mag greater than the value quoted in Luhman et al. (2011) and redder than any other T dwarf currently known. Figure 2 illustrates the trend of decreasing absolute 4.5 μm magnitude for known T dwarfs as a function of J − [4.5] color. Our upper limit for GJ 3483B is consistent with the T-dwarf trend, though it could be redder (and, thus, as faint as J ∼ 23–24). The initial–final mass relationships in Catalán et al. (2008, and references therein) along with the estimated white dwarf mass of 0.62 ± 0.03 M_☉ (Subasavage et al. 2009) suggest the progenitor mass was 2.0 ± 0.3 M_☉ with a main-sequence lifetime of 1.3^+0.7 _{− 0.5} Gyr (Catalán et al. 2008, and references therein; S. Leggett 2011, private communication). With the cooling age of 670 Myr derived in Subasavage et al. (2009), this implies a total age of 1.5–2.7 Gyr. Based on our 3σ limit, an age of 1.5 Gyr, and theoretical evolutionary models from Burrows et al. (2003) and Baraffe et al. (2003), we estimate a mass of <10 M_Jup and temperature <410 K. However, with an age of 2.7 Gyr, we find the companion has a mass of <13 M_Jup and temperature <400 K.

Zoom In Zoom Out Reset image size

The IRAC 4.5 μm photometric magnitude of 16.72 (Luhman et al. 2011) can also be used to derive the mass and temperature of GJ 3483B. With an age of 1.5 Gyr, we find the Burrows et al. (2003) models predict a mass of ∼6 M_Jup and temperature ∼330 K. With the upper age limit of 2.7 Gyr, we find mass and temperature of ∼10 M_Jup and ∼350 K. These model temperatures are somewhat warmer than the value Luhman et al. (2011) present (∼300 K). While the uncertainty in age and thus the mass of GJ 3483B are substantial, the range of implied masses is consistent with those anticipated from Jeans-mass fragmentation (see Section 4). That is, the mass of GJ 3483B is unlikely to be much smaller than ∼7 M_Jup. The range in mass and temperature (6–10 M_Jup, 330–350 K) for GJ 3483B is comparable to that of CFBDSIR J1458+1013B (6–15 M_Jup, 370  ±  40 K), another recently discovered ultra cool companion (Liu et al. 2011).

Companions with masses less than ∼20 M_Jup have been imaged in only a handful of systems (listed in Table 1) and most have masses between 7 and 15 M_Jup. As described in Zuckerman & Song (2009), objects of mass ∼15 M_Jup would require at least 2 Gyr to cool down to ∼500 K, which is usually taken as the temperature in which the Y-dwarf spectral type will appear. Binaries with brown dwarf secondaries having masses larger than 15 M_Jup are somewhat more common than the few listed in Table 1 (see Table 2 in Zuckerman & Song 2009), but for a ∼25 M_Jup object to cool to 500 K it must be at least 7 Gyr old, according to the Burrows et al. (2003) models. Hence, the coolest companions are more likely to be found in relatively old systems.

A handful of white dwarfs are known to host widely separated brown dwarf companions (see Table 2). While a few other WD+BD pairs are known, these are unresolved systems whose separations are a few AU or less (GD 1400, Farihi & Christopher 2004; WD 0137–349, Maxted et al. 2006). The separations among the resolved WD+BD pairs are generally on the order of ∼1000's of AU, albeit with low number statistics. Companions around white dwarfs are expected (and observed; see Farihi et al. 2010); to have a bimodal distribution in their orbital semimajor axes (Nordhaus et al. 2010). Those orbiting close to the main-sequence star will spiral in via tidal dissipation (Hansen 2010). More widely separated companions (such as most of those in Table 1) will undergo orbital expansion following post-main-sequence mass loss from the white dwarf progenitor. Assuming the mass loss was adiabatic, which is almost certainly true (the orbital period is short compared to the time frame of mass loss), the ratio of semimajor axes (a) between the white dwarf and main-sequence phases should be identical to the inverse ratio of white dwarf and main-sequence masses, or a_wd = a_msM_ms/M_wd (Jeans 1924). As previously mentioned, GJ 3483 has a final mass of 0.62 M_☉ and initial mass of 2.0 M_☉ . If we assume the 2500 AU projected separation corresponds to a semimajor axis, then while on the main sequence this binary would have had a ∼ 780 AU, consistent with the values in Table 1.

The commonly accepted mass limits for brown dwarfs are 13 and 75 M_Jup, where an object would fuse deuterium, but not hydrogen. In this case, planets (in addition to orbiting stars) would have masses below ∼13 M_Jup. However, Jeans fragmentation can produce gravitationally collapsing clumps with masses as low as ∼7 M_Jup (Low & Lynden-Bell 1976). Typically, these fragments will have average separations of a few 100 AU. As seen in Table 1, there are several imaged systems with masses lower than ∼15 M_Jup that orbit their host stars at these wide separations. Giant planets, on the other hand, can form via core accretion in a disk fairly close to the star (Pollack et al. 1996) or via gravitational instability at larger separations so long as the disk is sufficiently massive (Boss 1997). Numerical simulations, however, have shown that planet formation via core accretion does not generally occur beyond 35 AU (Dodson-Robinson et al. 2009; but see also Currie et al. 2011 for contrasting arguments). Gravitational instability could create massive planets at larger semimajor axes, but only for the most massive disks would one expect planets to form at ∼100 AU (see Rafikov 2005, and references therein).

Figure 3 shows the separations and masses of low-mass companions from Table 1 and extrasolar planets from the Exoplanet Orbit Database (Wright et al. 2011). The population of exoplanets extends to lower masses and shorter separations (semimajor axes), but is not displayed here. As mentioned in the preceding paragraphs, one would not expect giant planets to form at separations larger than ∼100 AU. However, it appears that for objects ∼7–17 M_Jup, companions are distributed in a wide range of separations, reaching as far as ∼1000 AU. Planet–planet scattering could conceivably eject objects out to these large separations (Veras et al. 2009; Dodson-Robinson et al. 2009, and references therein). Since lower mass planets are more readily scattered out to larger orbits than more massive objects, one would expect to see a larger number of low-mass planets at large separations. However, numerous adaptive optics searches for companions have been performed, yet very few low-mass objects have been found despite sensitivities to objects with masses ≳ 3 M_Jup at separations ≳40 AU (Chauvin et al. 2010; Nielsen & Close 2010, and references therein). Additionally, while many giant planets have been detected with radial velocity searches, there is a lack of >13 M_Jup substellar companions within ≲ 5 AU, despite being sensitive to such objects (the well-known brown dwarf desert; see Marcy & Butler 2000). The paucity of low-mass companions at large separations contrasts with both theoretical expectations and detected giant planets at a ≲ 5 AU and suggests that planet–planet scattering is not a common mechanism to move large, ∼7–17 M_Jup mass companions, such as GJ 3483B and those listed in Table 1, to large separations.

Zoom In Zoom Out Reset image size

The low-mass, directly imaged companions with separations ≲ 100 AU are worth mentioning, as they may be notable exceptions to Jeans fragmentation: HR 8799b,c,d,e; β Pic b; and 2M1207b. In the case of HR 8799 and β Pic, the planetary mass companions orbit relatively close to their A-type stars, which possess extended circumstellar disks (see Rhee et al. 2007). The brown dwarf 2M1207 also possesses a disk close to the primary (R ≲ 0.2 AU; Riaz et al. 2006), but the low-mass companion orbits beyond it at 46 AU. While the disks around A-type stars can be massive enough to form giant planets at large separations, it is unlikely that a brown dwarf can possess a similarly massive disk. This suggests that, while low-mass brown dwarf formation and planet formation operate on different scales, formation mechanisms for companions with masses ≲ 13 M_Jup begin to overlap (see discussions in Currie et al. 2011 and Kratter et al. 2010). The discovery of additional low-mass, widely separated companions will play a key role in our understanding of giant planet and low-mass brown dwarf formation.

We thank Adam Burgasser, Kevin Luhman, and Patrick Dufour for useful discussions. We thank our referee, Sandy Leggett, for a prompt review and useful suggestions that strengthened this Letter. We would also like to thank Jay Farihi for bringing to our attention the non-standard use of WD designations for non-white dwarf companions to white dwarf stars. This research has made use of the Exoplanet Orbit Database and the Exoplanet Data Explorer at exoplanets.org. This research was supported by NASA Astrophysics Data Analysis Program grant NNX09AC96G to RIT and UCLA.