RESOLVED SPECTROSCOPY OF M DWARF/L DWARF BINARIES. IV. DISCOVERY OF AN M9 + L6 BINARY SEPARATED BY OVER 100 AU

2MASS J0130−4445 was imaged with the 3 m NASA Infrared Telescope Facility (IRTF) SpeX spectrograph (Rayner et al. 2003) on 2009 December 7 (UT) as part of a program to identify unresolved M/L dwarf plus T dwarf spectral binaries (e.g., Burgasser et al. 2008). Conditions were clear but with poor seeing, 12 at K band, due in part to the large airmass of the observation (2.34–2.37). These images revealed a faint point source due east of the primary target at a separation of roughly 3''. Four dithered exposures were obtained of the pair in each of the MKO⁷ J, H, and K filters, with individual exposure times of 45 s, 30 s, and 30 s, respectively. The field rotator was aligned at a position angle of 0°, i.e., north up and east to the left.

Imaging data were reduced in a standard manner using custom IDL routines. Raw images were mirror flipped about the y-axis to reproduce the sky orientation and pair-wise subtracted to remove sky contributions. The difference images were divided by normalized flat-field frames, constructed by median combining the imaging data for each filter after masking out the sources. Subsections of each image, 10'' (83 pixels) on a side and centered on the target source, were extracted from these calibrated frames. A final image for each filter/target pair (Figure 1) was produced by averaging the registered subframes together, rejecting 5σ pixel outliers. The two sources of 2MASS J0130−4445 are well resolved along a nearly east–west axis. The brighter western component is hereafter referred to as 2MASS J0130−4445A and the eastern component as 2MASS J0130−4445B.

Zoom In Zoom Out Reset image size

Component magnitudes and the angular separation of the 2MASS J0130−4445 pair were determined through point-spread function (PSF) fits to the reduced imaging data, following the prescription described in McElwain & Burgasser (2006). The PSF models were derived from Gaussian fits to the primary component in the individual subimage frames. For each filter, four distinct PSF models were produced, each of which were fitted to the individual images, resulting in a total of 16 independent measures of the relative component magnitudes and 48 independent measures of the separation and orientation of the pair, in each of the JHK filters. However, as the secondary was undetected in one of the four J-band images, four measures of the relative J-band flux and separation were discarded before computing mean values and standard deviations. Separation measurements were converted from pixels to arcseconds assuming a plate scale of 0120 ± 0002 pixel⁻¹ (J. Rayner, 2005, personal communication) and no distortion. The position angle (set at 0°) was assumed to be accurate to within 025 (ibid.).

Results are listed in Table 1. The angular separation of the pair was measured to be 3282 ± 0047 at a position angle of 873 ± 09, i.e., along an east–west line. The secondary is both considerably fainter and significantly redder than the primary. We derived relative magnitudes of ΔJ= 3.11 ± 0.06 and ΔK= 2.34 ± 0.04. Using the combined-light 2MASS photometry for the system,⁸ this translates into J − K_s colors of 1.13 ± 0.04 and 1.94 ± 0.08 for the primary and secondary, respectively.

The two components of 2MASS J0130−4445 were observed on separate nights with the prism-dispersed mode of SpeX, the primary on 2009 December 7 (the same night as the imaging observations) and the secondary on 2009 December 28 (UT). Conditions on the latter night were clear with a stable seeing of 06 at K band. The SpeX prism mode provides 0.75–2.5 μm continuous spectroscopy with resolution λ/Δλ ≈120 for the 05 slit employed (dispersion across the chip is 20–30 Å pixel⁻¹). Both components were observed separately, with the slit oriented north–south, roughly aligned with the parallactic angle and perpendicular to the separation axis. For the primary, eight exposures of 90 s each were obtained at an average airmass of 2.33, while guiding on spillover light from the slit. For the secondary, eight exposures of 150 s each were obtained at an average airmass of 2.34, while guiding on the primary off-slit. For both sources, the A0 V star HD 8977 was observed immediately before the target for telluric and flux calibration while the quartz and Ar arc lamps were observed for flat-field and wavelength calibration, respectively. Data were reduced using the SpeXtool package, version 3.4 (Vacca et al. 2003; Cushing et al. 2004) using standard settings; see Burgasser et al. (2007b) for details.

Figure 2 shows the spectra of the two components of 2MASS J0130−4445AB; signal-to-noise at the JHK flux peaks was 100–150 and 25–35 for the A and B components, respectively. Both spectra show the characteristic near-infrared (NIR) features of late-type M and L dwarfs (e.g., Reid et al. 2001; McLean et al. 2003; Cushing et al. 2005): steep red optical slopes (0.8–1.0 μm) from the pressure broadened wing of the 0.77 μm K i doublet; molecular absorption bands arising from H₂O (1.4 and 1.9 μm), CO (2.3 μm), and FeH (0.99 μm); and an overall red spectral energy distribution (SED), consistent with the photometric colors. 2MASS J0130−4445A also exhibits additional absorption features in the 0.8–1.2 μm region arising from TiO, VO, FeH, and unresolved K i and Na i lines, all typical for a late-type M dwarf. The corresponding region in the spectrum of 2MASS J0130−4445B is considerably smoother, albeit more noisy, suggesting that many of these gaseous species have condensed out (e.g., Tsuji et al. 1998; Ackerman & Marley 2001). The appearance of weak H₂O absorption at 1.15 μm and the very red SED of the NIR spectrum all indicate that 2MASS J0130−4445B is a mid- to late-type L dwarf with relatively thick condensate clouds at the photosphere.

Zoom In Zoom Out Reset image size

To determine spectral types we compared our NIR spectra of 2MASS J0130−4445AB with 463 spectra of 439 M7 or later dwarfs from the SpeX Prism Spectral Libraries.⁹ All templates were chosen to have median S/N > 10 and could not be binaries, giants, subdwarfs, or spectral classifications that were peculiar or uncertain. Best matches were determined by finding the minimum χ² deviation between component spectra and templates in the 0.95–1.35, 1.45–1.80, and 2.00–2.35 μm regions (i.e., avoiding telluric bands), following the procedure of Cushing et al. (2008) with no pixel weighting. The two best matching templates to both components of 2MASS J0130−4445AB are shown in Figure 3. For 2MASS J0130−4445A, these are the optically classified M8 dwarf 2MASS J05173729−3348593 (Cruz et al. 2003) and L0 dwarf DENIS–P J0652197−253450 (Phan-Bao et al. 2008); for 2MASS J0130−4445B these are the optically classified L5 dwarf 2MASSW J1326201−272937 (Gizis 2002) and L7 dwarf 2MASS J03185403−3421292 (Kirkpatrick et al. 2008). Note that despite the differences in optical type, these spectra provide equivalently good fits—the two fits were different only by 1.7σ for the primary and 1.1σ for the secondary based on the F-test which gauges whether two different fits to data are significantly distinct based on the ratio of χ² values and degrees of freedom (Burgasser et al. 2010)—to the components of 2MASS J0130−4445AB, a reflection of the discrepancies between optical and NIR spectral morphologies for late-type M and L dwarfs (e.g., Geballe et al. 2002; Kirkpatrick 2005). A χ²-weighted mean of all the SpeX templates (e.g., Burgasser et al. 2010) indicates component types of M9.0 ± 0.5 for 2MASS J0130−4445A and L6 ± 1 for 2MASS J0130−4445B.

Zoom In Zoom Out Reset image size

We also derived classifications using a suite of spectral indices and spectral index/spectral-type relations from Tokunaga & Kobayashi (1999), Reid et al. (2001), Geballe et al. (2002), Burgasser et al. (2006), and Burgasser (2007). Table 2 shows the measured values and the inferred spectral subtypes of 2MASS J0130−4445A and 2MASS J0130−4445B for each of the indices. The mean and scatter from these indices yields classifications of L0.5 ± 1.0 for 2MASS J0130−4445A and L7.0 ± 1.5 for 2MASS J0130−4445B. These are consistent with, but less precise than, the types inferred from spectral template matching, so we adopt the latter for our subsequent analysis.

To assess whether two stars comprise a physical binary or are just a chance alignment of random stars, the most reliable method used is to check for a common systemic velocity. However, 2MASS J0130−4445B is very faint, even in the infrared, and has not been detected in any earlier epoch; hence, we do not have proper motions for the secondary nor radial velocities for either component. In the absence of kinematic information, we employed two other tests to examine whether 2MASS J0130−4445AB is a physical pair: (1) the heliocentric distances of the two components and (2) the probability that the sources are a chance alignment based on the surface distribution of stars on the sky.

The spectrophotometric distances to each component of 2MASS J0130−4445AB were derived using the M_J/spectral-type relations of Cruz et al. (2003) based on the combined-light 2MASS photometry and our relative SpeX photometry (Table 3). The derived distances are 34.5 ± 3.2 pc for the primary and 45.8 ± 13.6 pc for the secondary, where the errors are from the uncertainties in the NIR spectral types (see Section 3.2). These distances are consistent with each other within their associated errors.

Next, we calculated the probability that 2MASS J0130− 4445AB is a random chance alignment along our line of sight based on its three-dimensional position in the Galaxy. We followed Dhital et al. (2010), constructing a three-component Galactic model with the thin disk, thick disk, and halo, constrained with empirical stellar density profiles (Jurić et al. 2008; Bochanski et al. 2010). The model recreates a 30' × 30' region in the sky, centered around the coordinates of the given binary system and out to heliocentric distances of 2500 pc. As all the simulated stars are single and non-associated, any visual binary is a random chance alignment. In 10⁷ Monte Carlo realizations, we found, on average, 0.0285 chance alignments per realization on the sky, within the 33 angular separation of 2MASS J0130−4445AB. More importantly, none of these chance alignments was within the range of spectrophotometric distances (40 ± 14 pc) estimated for 2MASS J0130−4445AB. As such, we conservatively infer a ≲10⁻⁷ probability of positional coincidence. We therefore conclude that 2MASS J0130−4445AB is a physically bound binary and not a chance alignment of two unassociated stars.

The NIR spectral types of 2MASS J0130−4445A and 2MASS J0130−4445B correspond to effective temperatures, T_eff, of 2400 ± 110 K and 1450 ± 100 K, respectively, based on the T_eff/spectral-type relation of Stephens et al. (2009). Uncertainties include scatter in the T_eff relation and uncertainties in the classifications (Section 3.2) added in quadrature. Figure 4 shows the Burrows et al. (1993, 1997) evolutionary models, displaying T_eff as a function of mass and age; the Burrows mass tracks and the observed T_eff ranges are shown as dotted and dashed lines and gray boxes, respectively. As the masses of the two components vary quite a bit with assumed age, it is imperative to constrain the age of 2MASS J0130−4445AB.

Zoom In Zoom Out Reset image size

The absence of Li i in 2MASS J0130−4445A indicates that it is more massive than the predicted lithium depletion boundary (LDB) mass, ∼0.06 M_☉ (Chabrier et al. 1996; Burrows et al. 2001) for field stars of solar metallicity. Using the T_eff for the primary based on its spectral type (including uncertainties) and assuming a mass ≳0.06 M_☉, the evolutionary models of Burrows et al. (1997) indicate an age ≳250 Myr (Figure 4). We note that recent work by Baraffe & Chabrier (2010) has suggested that episodic accretion during the pre-main-sequence stages causes the central temperature of a star to increase up to 1 dex, with a sharp dependence on the frequency and magnitude of the episodic accretion. This serves to deplete Li i earlier than in non-accreting stars of the same mass and effectively reduces the inferred LDB mass and, thus, the minimum allowable age. Here, we have not taken episodic accretion into account. While we do not have an optical spectrum of the secondary, the presence (absence) of Li i in the spectrum would set a upper (lower) limit on the mass and, hence, the age of the system, in this case ≲1.8 Gyr (≳1.1 Gyr). The likely proximity of the mass of the secondary to the LDB is motivation to obtain an optical spectrum of this component.

The absence of Hα emission in the optical spectrum of 2MASS J0130−4445A and lack of UV or X-ray flux—the system is not detected in the Galaxy Evolution Explorer (Martin et al. 2005) or the ROSAT (Voges et al. 1999) All-Sky Surveys—indicate that the system is not particularly active and, hence, not likely to be a very young system. This absence indicates that 2MASS J0130−4445AB is probably older than 1–100 Myr, as such emission has been detected in BDs in the Orion Nebula Cluster (isochronal age ∼1 Myr; Peterson et al. 2008), Taurus (∼3 Myr; Guieu et al. 2006), σ Orionis (2–7 Myr; Zapatero Osorio et al. 2002), α Persei (∼80 Myr; Stauffer et al. 1999), Pleiades (∼100 Myr; Stauffer et al. 1998; Martín et al. 2000), and Blanco 1 (∼100 Myr; Cargile et al. 2010). However, activity signatures might not be reliable age indicators in the VLM regime. Both Hα and X-ray emission drop precipitously across the M dwarf/L dwarf transition (e.g., Kirkpatrick et al. 2000; Gizis et al. 2000; West et al. 2004; Stelzer et al. 2006), likely the result of reduced magnetic field coupling with increasingly neutral photospheres (e.g., Gelino et al. 2002; Mohanty et al. 2002).

Extreme youth can also be ruled out based on the NIR spectra of these sources, which do not exhibit the triangular H-band peaks seen in ∼100 Myr Pleiades M and L dwarfs (Bihain et al. 2010) and young field L dwarfs (e.g., Kirkpatrick et al. 2006). It is notable that 2MASS J0130−4445B is somewhat red compared to typical L6 dwarfs (〈J − K_s〉= 1.82 ± 0.07; Schmidt et al. 2010), as red sources have been shown to exhibit smaller velocity dispersions and, hence, younger ages (Faherty et al. 2009; Schmidt et al. 2010). However, 2MASS J0130−4445A is not unusually red for its spectral type (〈J − K_s〉= 1.12 ± 0.10), and the red color of the secondary may reflect an unusually dusty atmosphere (e.g., Looper et al. 2008b). Also, neither NIR spectra nor the optical spectrum of 2MASS J0130−4445A show high gravity signatures, i.e., unusually blue colors form enhanced H₂, or evidence of the system being metal-poor, making it unlikely that the system is as old as ∼10 Gyr (Burgasser et al. 2003b; Reid et al. 2007).

Considering the kinematics of the system, the tangential velocity of 2MASS J0130−4445A, 19 ± 3 km s⁻¹ is similar to the median velocities of the L dwarfs in the Sloan Digital Sky Survey (SDSS) sample (28 ± 25 km s⁻¹; Schmidt et al. 2010), the M9 dwarfs in the BDKP sample (23 ± 23 km s⁻¹; Faherty et al. 2009), and the M7–L8 dwarfs in the 2MASS sample (25 ± 21 km s⁻¹; Schmidt et al. 2007), with the quoted errors being the 1σ dispersions. The low tangential velocity suggests that 2MASS J0130−4445AB is part of the thin disk, although we note that we cannot rule out a higher space velocity for the binary system. Kinematic studies have found that late-M and L dwarfs with average kinematics are typically ∼2–4 Gyr old (Wielen 1977; Faherty et al. 2009).

In conclusion, based on the absence of Li i in the primary, we can place a (model dependent) hard limit on the minimum age of 2MASS J0130−4445AB of ∼250 Myr while its kinematics indicate a preferred age of ∼2–4 Gyr. Spectral features for both components are in agreement with these ages. For the ages of 0.25–10 Gyr, based on the Burrows et al. (1993) and Burrows et al. (1997) models, the estimated masses for 2MASS J0130−4445A and 2MASS J0130−4445B are 0.055–0.083 and 0.032–0.076 M_☉, respectively, and the mass ratio is 0.57–0.92. For kinematics-based age limits of 2–4 Gyr, the estimated masses for 2MASS J0130−4445A and 2MASS J0130−4445B are 0.082–0.083 and 0.066–0.073 M_☉, respectively, and the mass ratio is 0.81–0.89 (Table 4). Hence, the components straddle the hydrogen-burning mass limit, and this system is likely composed of a VLM star and (massive) BD pair.

With a projected separation of 130 ± 50 AU, 2MASS J0130−4445AB is one of only ten VLM systems wider than 100 AU, with six of them in the field. All of these systems have been identified relatively recently; prior to their discovery, it was believed that VLM field systems were nearly all tight, a possible consequence of dynamic ejection early on. Based on this idea and the VLM binary population known at the time, two relations to define the largest possible separation of VLM binaries were proposed. First, Burgasser et al. (2003a) suggested that the maximum separation of a system was dependent on its mass: a_max (AU) = 1400 (M_tot/M_☉)². Second, Close et al. (2003) proposed that the stability of binary systems was contingent on their binding energy—a criterion based on the product rather than the sum of component masses; thus, only systems with binding energy ⩾10^42.5 erg would exist in the field.¹⁰ For the (age dependent) estimated mass of 2MASS J0130−4445AB, the Burgasser et al. (2003a) relation equates to maximum physical separations of only 9.2 AU and 35.4 AU for ages of 0.25 and 10 Gyr, respectively, which are both much smaller than the physical separation we have measured for 2MASS J0130−4445AB. Similarly, the binding energies for the system are 10^41.61 and 10^41.97 erg for the same ages (see Table 4). Both the Burgasser et al. (2003a) and Close et al. (2003) relations are definitively violated by 2MASS J0130−4445AB, for all ages and mass ratios. Assuming these limits emerge from dynamical scattering processes, this binary seems unlikely to have formed via the ejection of protostellar embryos.

More recently, Zuckerman & Song (2009) have argued that fragmentation, rather than dynamical, processes are more likely to describe the boundary for the lowest binding energy systems. A protostellar cloud can only fragment if it is more massive than the minimum Jeans mass (∼7 M_J; Low & Lynden-Bell 1976). Assuming fiducial separations of 300 AU for the fragments, they derived a cutoff for binding energy as a function of the total systemic mass. Finding that this disfavors the formation of very wide and/or high mass ratio binaries, Faherty et al. (2010) used the Jeans length, instead of the fiducial separation, and mass ratio of the system. For 2MASS J0130−4445AB at 0.25 Gyr (M_tot ≈ 0.1 M_☉), the Zuckerman & Song (2009) and Faherty et al. (2010) relations suggest minimum binding energies of 10^40.5 and 10³⁹ erg, respectively; if the system were older, they would be even more stable due to the higher masses. 2MASS J0130−4445AB is well within the bounds of both the Zuckerman & Song (2009) and Faherty et al. (2010) formation criteria for all ages and mass ratios. Hence, the observed wide, low binding energy VLM binaries could have formed from small protostellar clouds, with masses close to the local Jeans mass.

Current numerical simulations have suggested an alternative mechanism to form wide binaries: N-body dynamics in small clusters disrupt the VLM pairs wider than ∼60 AU, and very wide systems (>10⁴–10⁵ AU) can then be formed when stars are ejected into the field in the same direction (Moeckel & Bate 2010; Kouwenhoven et al. 2010). While this provides a mechanism to create the most fragile VLM pairs identified to date, it does not aid in the formation of 100–1000 AU pairs like 2MASS J0130−4445AB. Two other VLM systems in this separation range are known: 2MASS J1623361−240221 (220 AU; Billères et al. 2005) and SDSS J141623.94+134826.3 (100 AU; Burningham et al. 2010; Scholz 2010).

The components of 2MASS J0130−4445AB straddle a spectral-type range that is particularly interesting for three reasons. First, condensate clouds become an important source of photospheric opacity and thermal evolution starting at the end of the M dwarf sequence and peaking in influence in the middle of the L dwarf sequence (Ackerman & Marley 2001; Kirkpatrick et al. 2008; Saumon & Marley 2008). With a component on either end of this regime, 2MASS J0130−4445AB is a particularly useful coeval laboratory for studying the emergence and dispersal of these clouds. Second, both components straddle the hydrogen-burning mass limit, and the secondary is very near the LDB. Detection of Li i in the optical spectrum of the secondary could provide a relatively precise constraint on the age of this system (0.25–1.8 Gyr) and thereby make it a useful benchmark for studies of BD thermal evolution and atmospheric models (e.g., Pinfield et al. 2006). Third, the M dwarf/L dwarf transition exhibits a steep decline in magnetic activity metrics, including Hα, UV, and X-ray emission (e.g., Gizis et al. 2000; West et al. 2004) but notably not radio emission (e.g., Berger 2006; Berger et al. 2010). This is believed to be due to the decoupling of magnetic fields from an increasingly neutral photosphere (Gelino et al. 2002; Mohanty et al. 2002) but does not rule out the presence of significant magnetic fields (Reiners & Basri 2007). While Hα is absent in the spectrum of 2MASS J0130−4445A, examination of field strengths and radio emission in this coeval pair may facilitate understanding of how magnetic fields evolve across the stellar/BD transition. As one of only three binaries spanning the M dwarf/L dwarf transition whose components are easily resolvable from ground-based facilities (the other two are the 12 L1.5+L4.5 2MASS J1520022−442242 and the 10 M9+L3 2MASS J1707234−055824; Burgasser 2004; Burgasser et al. 2007a; Folkes et al. 2007), 2MASS J0130−4445AB is an important laboratory for studying how condensate clouds, lithium burning, and magnetic activity trends vary across this transition.