THE HYPERACTIVE L DWARF 2MASS J13153094−2649513: CONTINUED EMISSION AND A BROWN DWARF COMPANION^*

Moderate-resolution optical spectra of 2MASS J1315−2649 were obtained on 2011 March 26 (UT) with the Magellan Echellette (MagE; Marshall et al. 2008), mounted on the 6.5 m Landon Clay Telescope at Las Campanas Observatory. Conditions during the observations were clear with 06 seeing. Two exposures totaling 3000 s were obtained at an average airmass of 1.003 using the 07 slit aligned with the parallactic angle; this setup provided 3200–10050 Å spectroscopy at a resolution of λ/Δλ ≈ 4000. We also observed the spectrophotometric flux standard EG 274 (Hamuy et al. 1994) on the same night for flux calibration. ThAr lamps were obtained after each source observation for wavelength calibration, and internal quartz and dome flat-field lamps were obtained during the night for pixel response calibration. Data were reduced using the MASE reduction pipeline (Bochanski et al. 2009), following standard procedures for order tracing, flat-field correction, wavelength calibration (including heliocentric correction), optimal source extraction, order stitching, and flux calibration.

The red portion of the reduced spectrum is shown in Figure 1, flux-calibrated to an apparent i-band magnitude of 20.16 ± 0.14 as estimated from Two Micron All Sky Survey (2MASS) photometry and a mean i − J = 4.97 ± 0.13 color for L5 dwarfs (Schmidt et al. 2010). We confirm the characteristic mid-L dwarf features identified in previous studies, including strong FeH and CrH bands; weak TiO absorption (relative to late-M and early-L dwarfs); line absorption from Na i, Rb i, and Cs i; and the strongly pressure-broadened 7700 Å K i doublet. The overall spectral shape is well matched to the L5 dwarf 2MASS J15074769−1627386 (Reid et al. 2000), consistent with previously reported classifications (Gizis 2002; Kirkpatrick et al. 2008). We confirm the absence of 6710 Å Li i absorption as reported by Kirkpatrick et al. (2008) to an equivalent width (EW) limit of 0.5 Å. This is well below measured EWs for equivalently classified L dwarfs (e.g., Kirkpatrick et al. 2000). We also see no evidence of any peculiar spectral features associated with low surface gravities, such as enhanced VO absorption or weakened alkali lines (Kirkpatrick et al. 2006; Cruz et al. 2009). Line center measurements of the alkali lines indicate a heliocentric radial velocity of −7 ± 9 km s⁻¹.

Zoom In Zoom Out Reset image size

The most striking feature in the optical spectrum of 2MASS J1315−2649 is its pronounced Hα emission. We measure an EW of −58 ± 4 Å in the MagE data, in the middle of prior measurements that span −24 to −160 Å. We also detect Hγ (4342 Å) and Hβ (4863 Å) in emission, and confirm the presence of Na i emission as reported in Fuhrmeister et al. (2005). In addition, we detect weak emission in the cores of the 7700 Å K i doublets, but no emission from the 5877 Å He i D3 (EW < 18 Å) or 8500–8660 Å Ca ii triplet lines (EW < 0.3 Å).⁸ Line profiles of detected emission features (Figure 2) show no appreciable broadening at the 75 km s⁻¹ velocity resolution of MagE, nor do we detect any significant velocity shift between emission and absorption lines (ΔV_rad = 4 ± 16 km s⁻¹). As most of these lines are superimposed on an undetected continuum, we report in Table 1 line fluxes and relative line-to-bolometric luminosities, the latter computed using the bolometric correction (BC)/spectral type relations of Liu et al. (2010). Our measurement of log₁₀L_Hα/L_bol = −4.18 ± 0.06 is similar to the first detection made by Hall (2002b), and is again one to two orders of magnitude greater than measured quiescent or flaring fluxes for equivalently classified L dwarfs. The Balmer decrement F_Hα/F_Hβ = 2.7 ± 0.3 is roughly comparable to the mean values for non-flaring M dwarfs (Gizis et al. 2002). Hα and Hβ emission contribute 61% and 23% of the total measured line flux of log₁₀L_e/L_bol ≈ −4.

Zoom In Zoom Out Reset image size

Low-resolution near-infrared spectra of 2MASS J1315−2649 were obtained on 2009 June 30 (UT) with the 3 m NASA Infrared Telescope Facility (IRTF) SpeX spectrograph (Rayner et al. 2003). Observing conditions were clear and dry with 07 seeing at the H band. We used the prism-dispersed mode of SpeX with a 05 slit (aligned to the parallactic angle) to obtain a continuous 0.7–2.5 μm spectrum with resolution λ/Δλ ≈120. A total of eight exposures of 120 s each were obtained in two ABBA dither pairs, nodding along the slit, at an average airmass of 1.50. We also observed the A0 V star HD 125438 (V = 7.10) for flux calibration and telluric absorption correction, as well as internal flat-field and argon arc lamps for pixel response and wavelength calibration. Data were reduced with the IDL SpeXtool package, version 3.4 (Vacca et al. 2003; Cushing et al. 2004), using standard settings see Burgasser & McElwain (2006) for details.

The reduced spectrum of 2MASS J1315−2649 is shown in Figure 3, compared to the L5 dwarf 2MASS J21373742+0808463 (hereafter 2MASS J2137+0808; Reid et al. 2008). Its near-infrared spectral morphology is consistent with its L5 optical type, with strong H₂O and CO absorption bands, FeH absorption at 1.0 and 1.6 μm, and (unresolved) alkali line absorption in the 1.1–1.3 μm region. The overall spectral shape is again consistent with a normal L5 field dwarf, with no evidence of an unusual surface gravity, metallicity, or cloud content (e.g., McGovern et al. 2004; Allers et al. 2007; Burgasser et al. 2008b; Looper et al. 2008b).

Zoom In Zoom Out Reset image size

There is, however, a subtle "notch" feature present at 1.62 μm that differs from the spectrum of 2MASS J2137+0808 (see inset box of Figure 3). This feature has previously been noted in the combined-light spectra of L dwarf plus T dwarf binaries, arising from the overlap of FeH absorption in the primary and CH₄ absorption in the secondary (e.g., Burgasser 2007b; Burgasser et al. 2008a; Gelino & Burgasser 2010; Geißler et al. 2011). To characterize this feature, we performed a spectral template fitting analysis similar to that described in Burgasser et al. (2008a), using 295 L2–T8 spectral templates from the SpeX Prism Spectral Libraries.⁹ Fluxes of these templates were scaled to the $M_{K_s}$/spectral type relation of Looper et al. (2008a), and all spectra were interpolated onto a common wavelength scale. Restricting potential secondaries to have T spectral types, a total of 17,889 binary templates were constructed and compared to the spectrum of 2MASS J1315−2649 over the wavelength ranges 0.95–1.35 μm, 1.45–1.80 μm, and 2.00–2.35 μm using the χ² statistic.

Figure 4 shows the four best-fitting binary templates from these comparisons, a combination of 2MASS J2137+0808 and either early- or late-type T dwarfs. The addition of a T-type companion fills in the "missing" flux at 1.58 μm, creating the distinct notch feature at 1.62 μm and at 1.27 μm produces a somewhat sharper J-band flux peak. Importantly, all of the binary templates shown in Figure 4 provide statistically superior fits to the spectrum of 2MASS J1315−2649 as compared to 2MASS J2137+0808 alone, based on the F-test statistic (Equations (2)–(6) in Burgasser et al. 2010a). However, we cannot precisely constrain the properties of the secondary from this analysis; the average spectral type of all of the template fits weighted by the F-test probability distribution function (F-PDF) is T3 ± 4.

Zoom In Zoom Out Reset image size

To more accurately characterize this putative companion, high-resolution, near-infrared images of 2MASS J1315−2649 were obtained with the 10 m Keck II laser guide star adaptive optics system (LGSAO; Wizinowich et al. 2006; van Dam et al. 2006) and facility NIRC2 near-infrared camera. Observations were conducted on two runs, 2010 March 24 and 2010 May 13 (UT), both with clear skies and fair seeing (<1'' and 05, respectively). We used the narrow camera with image scale 9.963 ± 0.011 mas pixel⁻¹ (Pravdo et al. 2006) covering a 102 × 102 field of view. Images were obtained through the J, H, and K_s filters, using a three-point dither pattern that avoided the noisy, lower left quadrant of the focal plane array. Exposure times ranged from 30 s with eight co-adds to 120 s with two co-adds per pointing position, with total integrations of 360–720 s in a given filter. The sodium LGS provided the wave front reference source for AO correction, while tip-tilt aberrations and quasi-static changes were measured by monitoring the R = 12.8 field star USNO-B1.0 0631-0348160 (Monet et al. 2003) located ρ = 386 from 2MASS J1315−2649. Images were reduced using custom IDL¹⁰ scripts, as described in Gelino & Burgasser (2010).

Figure 5 displays the reduced NIRC2 images from our May observations. A faint source is clearly present southeast of 2MASS J1315−2649, and was visible during both imaging epochs. Relative astrometry for each epoch was measured on the individual frames using a centroiding algorithm, and these values were then averaged and multiplied by the pixel scale (uncertainties include the standard deviations of the position measurements and 0.1% pixel scale uncertainty). Relative photometry was performed on the co-added mosaics through aperture photometry. As the point-spread function (PSF) of the brighter component contributes significant flux (≈25%) to the brightness of the fainter object, photometry for the latter was extracted from a primary PSF-subtracted image, constructed by rotating the frame 180° about the centroid of the primary and differencing. Systematic errors in the primary subtraction were estimated by offsetting the rotation axis over a 5 × 5 pixel grid centered on the original centroid, subtracting, and measuring aperture photometry on the secondary. The final photometric values for each band and epoch were taken as the means and standard deviations of these 25 measurements.

Zoom In Zoom Out Reset image size

Results are listed in Table 2. Separations in right ascension and declination are consistent between both epochs, and (in conjunction with the OH-Suppressing InfraRed Integral field Spectrograph (OSIRIS) observations described below) yield a mean separation of 336 ± 6 mas at position angle of 1464 ± 05, measured from primary to secondary. The mean relative magnitudes are also statistically consistent between epochs (to within 2σ), and indicate that the companion is both considerably fainter (ΔJ = 3.03 ± 0.03, ΔK_s = 5.09 ± 0.10) and bluer in the near-infrared. To our knowledge, this is the second-most-extreme near-infrared flux ratio measured for an ultracool dwarf binary to date.¹¹ We examine the physical association of the companion in Section 4.3; hereafter, we refer to the two sources as 2MASS J1315−2649A and B.

Table 2. NIRC2 and OSIRIS Astrometry and Photometry

Parameter	Value
NIRC2 epoch 2010 Mar 24 (UT)
Δαcos δ ('')	185 ± 6
Δδ ('')	−277 ± 3
ρ ('')	333 ± 7
θ ($\deg$)	143.1 ± 1.1
ΔJ (mag)	3.12 ± 0.05
ΔH (mag)	4.29 ± 0.14
ΔKs (mag)	4.91 ± 0.18
NIRC2 epoch 2010 May 13 (UT)
Δαcos δ ('')	187 ± 2
Δδ ('')	−286 ± 7
ρ ('')	342 ± 7
θ ($\deg$)	143.9 ± 0.5
ΔJ (mag)	3.00 ± 0.03
ΔH (mag)	4.59 ± 0.07
ΔKs (mag)	5.17 ± 0.12
OSIRIS epoch 2010 May 19 (UT)
Δαcos δ ('')	180 ± 5
Δδ ('')	−286 ± 6
ρ ('')	338 ± 4
θ ($\deg$)	147.8 ± 1.2

Note. Angular separation (ρ) and position angle (θ) measured from the brighter primary to the fainter secondary.

Download table as:  ASCIITypeset image

Resolved H-band spectroscopy of 2MASS J1315−2649AB was obtained using Keck II OSIRIS (Larkin et al. 2006) and LGSAO in mostly clear conditions on 2010 May 19 (UT). We used the 35 mas scale camera and H_bb filter, providing 1.47–1.80 μm spectroscopy at an average resolution of 3800 and dispersion of 2.1 Å pixel⁻¹ over a 056 × 224 field of view. The instrument rotator was set at a position angle of −45° to accommodate both components in the rectangular field of view. Six exposures of 600 s each were obtained at an average airmass of 1.52 using a linear dither pattern with steps of 04 along the long axis, and tip-tilt correction for LGSAO operation was again provided by USNO-B1.0 0631-0348160. These observations were followed by a 600 s exposure of a blank sky frame. We also obtained three dithered 20 s exposures of the A0 V star HD 107120 (V = 9.90) in natural guide star (NGS) AO mode at an airmass of 1.56 for telluric absorption correction and flux calibration.

Data were reduced with the OSIRIS data reduction pipeline (Krabbe et al. 2004), version 2.3. We first subtracted the 600 s sky frame from each of the 2MASS J1315−2649 images, and a median-combined dark frame from the calibrator images. We then used the pipeline to adjust bias levels, remove detector artifacts and cosmic rays, extract and wavelength-calibrate the position-dependent spectra (using the most current rectification files as of 2011 February), assemble three-dimensional data cubes and correct for dispersion. Spectra for the 2MASS J1315−2649 primary and HD 107120 were extracted directly from the data cube by aperture photometry in each image plane, using a 3 pixel (105 mas) aperture and 10–20 pixel (350–700 mas) sky annulus. For the faint companion, light contamination from the primary was a concern, so we first performed a partial subtraction of the primary's radial brightness profile. We sampled the profile over two position angle ranges 20°–40° away from the separation axis, generating a mean profile as a function of wavelength and separation. We then subtracted this profile ±25° about the separation axis. Figure 6 displays mosaics of both the original and subtracted images, illustrating the reduced background achieved around the companion. We measured aperture photometry for this component in each of the subtracted image planes, using a more restricted 1.5 pixel (52 mas) aperture and 3–5 pixel (105–175 mas) sky annulus. The individual spectra for all three sources were scaled and combined using the xcombspec routine in SpeXtool (Cushing et al. 2004). Flux calibration and telluric correction of the 2MASS J1315−2649AB spectra were performed using the xtellcor_general routine in SpeXtool, assuming a 20 nm Gaussian kernel for the A0 V H i lines (Vacca et al. 2003).

Zoom In Zoom Out Reset image size

Figure 7 displays the reduced spectra of the two components of 2MASS J1315−2649. The spectrum of 2MASS J1315−2649A has an exceptionally high signal-to-noise ratio (S/N ≈ 200), and is similar to that of the combined-light SpeX spectrum, with strong H₂O absorption wings shortward of 1.55 μm and longward of 1.7 μm, and weak FeH absorption in the 1.57–1.64 μm region, all indicative of a mid-type L dwarf. The notch feature, however, is no longer present. The spectrum of 2MASS J1315−2649B (S/N ≈ 25 at 1.6 μm) is unambiguously that of a late-type T dwarf, with strong CH₄ absorption at 1.6 μm.

Zoom In Zoom Out Reset image size

Component spectral types were determined by comparing the resolved OSIRIS spectra to the SpeX Prism Spectral Libraries templates, restricting the template sample to optically classified L dwarfs and near-infrared-classified T dwarfs. Following a χ²-fitting procedure similar to that described above over the 1.5–1.8 μm region, we identified the L5 2MASS J2137+0808 (Reid et al. 2008) and the T7 2MASS J07271824+1710012 (Burgasser et al. 2002a) as the best-fitting templates to 2MASS J1315−2649A and B, respectively (Figure 7). An F-test weighted average of all templates indicates mean classifications of L3.5 ± 2.5 and T7 ± 0.6 for the components. The large uncertainty for the former is largely due to the broad diversity of near-infrared spectra exhibited by L dwarfs at a given optical spectral type, arising from variations in surface gravity, metallicity, and cloud properties (Allers et al. 2007; Burgasser et al. 2008b; Looper et al. 2008b; Kirkpatrick et al. 2010). We therefore adopt the optical L5 classification for this component and the near-infrared T7 classification for the secondary.

Component brightnesses on the MKO¹² system were determined by converting the combined-light 2MASS JHK_s and relative NIRC2 JHK_s magnitudes to MKO JHK. These conversions were computed directly from the SpeX prism spectra of 2MASS J1315−2649AB (2MASS → MKO) and the best-fit templates in Figure 7 (NIRC2 → MKO) using the appropriate filter profiles and a Kurucz model spectrum of Vega (see Cushing et al. 2005). The resulting combined light and component magnitudes are listed in Tables 3 and 4, respectively. Both components appear to have relatively normal near-infrared colors for their spectral types (Leggett et al. 2010).

Component distances were computed using the MKO JHK absolute magnitude/spectral type relations of Liu et al. (2006); we considered both the "bright" and "faint" relations. Propagating uncertainties in component spectral types and photometry, and scatter in the relations, through Monte Carlo simulation yielded consistent distances for each component in all three bands and both relations, with mean values of 18 ± 4 pc and 35 ± 9 pc for 2MASS J1315−2649A and B, respectively. These distances are formally consistent with each other, differing at the 1.8σ level; the latter has a larger uncertainty due to the larger photometric error for this component (Table 4). The error-weighted mean distance of 20 ± 3 pc matches the 21.7 pc estimate of Riaz & Gizis (2007). At this distance, the projected separation of the components is 6.6 ± 0.9 AU, right at the peak of the separation distribution of resolved very-low-mass field binaries (Allen 2007).

Component luminosities were computed from the individual MKO JHK magnitudes using BC relations as a function of spectral type, as quantified in Liu et al. (2010). Apparent bolometric magnitudes were converted to absolute bolometric magnitudes by adopting a common distance of 20 ± 3 pc, and luminosities calculated assuming M_{bol, ☉} = 4.74. Luminosities computed in each of the JHK bands were again mutually consistent, resulting in log₁₀L_bol/L_☉ = −4.19 ± 0.16 and −5.86 ± 0.16 for 2MASS J1315−2649A and B, respectively. The luminosity for 2MASS J1315−2649A is similar to other L4.5–L5.5 field dwarfs as compiled by Golimowski et al. (2004) and Vrba et al. (2004), while 2MASS J1315−2649B is somewhat underluminous for its spectral type.

The similar distances and relatively small projected separation of 2MASS J1315−2649A and B indicate that these sources are co-spatial; we also find that their space motions are aligned. The three epochs of NIRC2 and OSIRIS relative astrometry are consistent with each other in both right ascension and declination, and despite the short period between the observations the high proper motion of 2MASS J1315−2649 (μ_αcos δ = −682 ± 13 mas yr⁻¹, μ_δ = −282 ± 14 mas yr⁻¹; Faherty et al. 2009) allows us to rule out either component as a (non-moving) background star at the 9σ level. Furthermore, the magnitudes and position angles of the component proper motions are identical to within ±47 mas yr⁻¹ (4.4 km s⁻¹ at 20 pc) and ±4°. The radial motions of the two components are also equivalent. Cross-correlation of the OSIRIS spectra with zero-velocity spectral model templates from Allard et al. (2011; see Section 4.4) yields identical velocities to within the uncertainties (ΔV_rad = 0 ± 9 km s⁻¹).

With a projected tangential velocity of V_tan = 70 ± 18 km s⁻¹, and adopting the radial velocity from the combined-light optical spectrum above, we find UVW space velocities in the local standard of rest¹³ (LSR) of U = −38 ± 8 km s⁻¹, V = −41 ± 9 km s⁻¹, and W = −13 ± 5 km s⁻¹. These values lie outside the 1σ velocity spheroid for the "cold" population of nearby L dwarfs reported by Schmidt et al. (2010), but are within the velocity spheroid of local thick disk stars (e.g., Soubiran et al. 2003). Both the kinematics and spectral properties of 2MASS J1315−2649 are therefore consistent with an older dwarf system in the Galactic disk population.

Assuming that 2MASS J1315−2649AB comprises a coeval system, insight into the physical properties of its components can be obtained by joint comparison to atmospheric and evolutionary models. The spectral and kinematics analyses above suggest that 2MASS J1315−2649AB is an older system; a more quantitative constraint comes from the absence of Li i absorption in the combined-light optical spectrum, which is dominated by the L5 primary. The absence of this line sets a minimum mass of ∼0.06 M_☉ for this component (Rebolo et al. 1992; Chabrier et al. 1996). Combined with its luminosity, the evolutionary models of Burrows et al. (1997), Baraffe et al. (2003), and Saumon & Marley (2008) indicate a minimum age for the primary ranging from 0.8 to 1.0 Gyr. This implies a minimum secondary mass of 0.013 M_☉, around the deuterium-burning minimum mass limit (Chabrier & Baraffe 2000; Spiegel et al. 2011).

Figure 8 displays the regions in T_eff/log g space occupied by the components as constrained by their luminosities, the minimum mass of 2MASS J1315−2649A, and the evolutionary models listed above. Table 4 details specific physical properties for select ages based on the evolutionary models of Baraffe et al. (2003). With no empirical upper limit on the age of this system, the mass of 2MASS J1315−2649A could be above the hydrogen-burning mass limit (M ≳ 0.072 M_☉ for τ ≳ 2 Gyr), while the secondary must be substellar at any age. The estimated mass ratio of the system, q ≡ M₂/M₁, is one of the smallest inferred for a very low mass binary: ∼0.3 (∼0.6) for an age of 1 Gyr (10 Gyr). This makes 2MASS J1315−2649AB a unique system, given that ∼90% of resolved brown dwarf binaries identified to date have q > 0.6 (Allen 2007; Burgasser et al. 2007). Indeed, if low-mass binaries prefer to be in higher mass ratio systems, these values further support the hypothesis that 2MASS J1315−2649 is quite old. T_eff constraints on the components—1570–1930 K for the primary and 570–790 K for the secondary (1σ ranges)—are again consistent with comparably classified field dwarfs (Golimowski et al. 2004; Cushing et al. 2008), with 2MASS J1315−2649B being somewhat on the cool side for its spectral type. The lower mass limit for 2MASS J1315−2649A tightly constrains its surface gravity to log g = 5.22–5.46 cm s⁻², while 2MASS J1315−2649B has a broader constraint of 4.46–5.35 cm s⁻².

Zoom In Zoom Out Reset image size

An independent assessment of the component atmospheric parameters was made by fitting the OSIRIS spectra to the BT-Settl models of Allard et al. (2011). These models are based on the PHOENIX code (Hauschildt et al. 1999), and reflect an update to the original Settl models of Allard et al. (2003) with a microturbulence velocity field determined from two-dimensional hydrodynamic models (Freytag et al. 2010) and updated solar abundances from Asplund et al. (2009). We followed the same fitting procedure described in Burgasser et al. (2010c; see also Cushing et al. 2008; Bowler et al. 2009), using a set of solar-metallicity models sampling T_eff = 600–2500 K (100 K steps) and log g = 4.0–5.5 cm s⁻² (0.5 cm s⁻² steps). Model surface fluxes (in f_λ units) and the OSIRIS spectra were smoothed to a common resolution of λ/Δλ = 3500 using a Gaussian kernel, and interpolated onto a common wavelength grid. The data were then scaled to the appropriate H-band apparent magnitude. Data and models were compared over the 1.5–1.75 μm region using a χ² statistic, with the degrees of freedom equal to the number of resolution elements sampled. An optimal scaling factor C ≡ (R/d)² was computed for each fit to minimize χ², where R is the radius of the brown dwarf and d is its distance from the Sun (Bowler et al. 2009). We further constrained our model selection by requiring that the model-inferred distance be within 2σ of the estimated spectrophotometric distance of the system (13–26 pc). Means and uncertainties in the atmospheric parameters were determined using the F-PDF as a weighting factor, as above; we also propagated sampling uncertainties of 50 K and 0.25 dex for T_eff and log g, respectively. Note that these uncertainties quantify experimental errors; they do not account for systematic errors that arise from the fidelity of the model fits, as discussed below.

Figure 9 displays the best-fitting models for the OSIRIS spectra. For wavelengths shortward of 1.55 μm, the models provide reasonably good fits to the forest of H₂O lines present in both component spectra. However, at longer wavelengths we see deviations in the primary arising from missing FeH opacity and a premature downturn in fluxes longward of 1.65 μm. The latter is symptomatic of overly blue spectral energy distributions across the near-infrared range exhibited by the models at these temperatures. There are also deviations in model fits to the secondary near the 1.58 μm peak and within the 1.6–1.75 μm CH₄ absorption system. Note that, quantitatively, the fit to the secondary is better (χ²_r = 1.52) than that to the primary (χ²_r = 38.4), but this mainly stems from the high S/N data for the latter; neither fit reproduces the observed spectrum with great fidelity. Nevertheless, visual inspection confirms that these are the best fits among the model sample, and the inferred mean parameters—T_eff = 1760 ± 70 K and log g ≳ 5.2 cm s⁻² for the primary and T_eff = 790 ± 70 K and log g = 5.2 ± 0.4 cm s⁻² for the secondary—are roughly in line with estimates from the evolutionary model parameters above.

Zoom In Zoom Out Reset image size

With independent determinations of luminosity and T_eff for both components, we can now examine whether the evolutionary and atmospheric models are consistent with each other assuming the system is coeval; this is the so-called coevality test (e.g., Liu et al. 2010). Prior studies of brown dwarf binaries have produced mixed results with respect to this test, with a few low-temperature systems showing evidence of ∼50–100 K offsets (both high and low) between evolutionary and atmospheric models (Liu et al. 2008, 2010; Dupuy et al. 2009a, 2009b; Konopacky et al. 2010). However, spectroscopic T_eff's have generally been taken from estimates of comparably classified field dwarfs, rather than spectroscopic fits to the binary components themselves.¹⁴ Such temperature estimation by proxy could result in systematic biases. The single exception is the T1+T6 binary Indi BC, for which resolved optical and near-infrared spectroscopy have enabled determinations of individual component luminosities and T_eff's (Kasper et al. 2009; King et al. 2010), and comparison of these components on the H-R diagram indicates that they are coeval with each other and with their stellar primary (Liu et al. 2010).

For 2MASS J1315−2649AB, we find good agreement between atmospheric and evolutionary models for the primary but not for the secondary. Figure 8 compares the T_eff and log g constraints from our spectral models fits to those from the evolutionary model comparisons. The atmospheric T_eff and log g constraints for the primary overlap reasonably well for all four evolutionary model sets, although there is less agreement for the cloudy models of Saumon & Marley (2008). Moreover, these values are consistent with the ∼0.8–1.0 Gyr minimum age of the system based on the absence of Li i in the optical spectrum. For the secondary, overlap in T_eff and log g regions is not as good, with essentially no overlap for the Saumon & Marley (2008) cloudy models. This discrepancy can also be seen in Figure 10, which compares the H-R diagrams for all four evolutionary models to the luminosities and spectral model fit T_eff's for the 2MASS J1315−2649 components. Both sit at the lower envelope of the evolutionary tracks, consistent with older ages; however, 2MASS J1315−2649B falls off the Saumon & Marley (2008) cloudy tracks entirely.

Zoom In Zoom Out Reset image size

The sense of the deviations between the model comparisons of 2MASS J1315−2649B is that the spectral model fit T_eff's are systematically higher than the evolutionary model T_eff's. As we do not have any other independent constraints on the system (e.g., age, metallicity, or component masses), we cannot determine a priori whether this mismatch is specifically attributable to errors in the spectral or evolutionary models. However, based on the fits shown in Figure 9, we suspect the former given the poor match between the BT-Settl models and spectral data around the 1.6 μm CH₄ band. This feature causes persistent problems in spectral model fits due to incomplete CH₄ opacities at T dwarf temperatures (Saumon et al. 2006; Freedman et al. 2008). A decrease of just 100–200 K in the derived secondary T_eff would bring both components in precise alignment with evolutionary models in both T_eff/log g and H-R diagrams, a shift previously suggested in prior low-mass binary analyses (although not necessarily in the same direction; Konopacky et al. 2010; Liu et al. 2010). It is also notable that spectral model fits for 2MASS J1315−2649A are even worse than those for 2MASS J1315−2649B. It may be that the alignment of evolutionary and spectral model parameters for this component is merely an example of "chance agreement." In any case, without more accurate fitting of L and T dwarf spectral data, such coevality tests are fundamentally inconclusive about the underlying accuracies of model-derived parameters.

We emphasize that the discrepancies noted here are only at the 1σ level, and should be verified through more precise constraints on the component luminosities and T_eff's. These can best be accomplished through a parallax distance measurement of the system, as the distance uncertainty dominates the luminosity uncertainty. In addition, resolved spectroscopy spanning the near-infrared (and possibly optical) range would provide a more robust test of the atmosphere models, and allow us to test different sets of models. Unfortunately, mass measurements for this widely separated system are probably not feasible in the near future. A probability analysis of the possible orbits for this system¹⁵ predicts likely periods of 45–60 yr (15–95 yr at 90% confidence). This rules out a "fast" astrometric orbit determination, and the relative orbital radial velocities (≲0.5 km s⁻¹) are comparable to current systematic uncertainties for isolated late-type dwarfs (e.g., Blake et al. 2010). Despite these challenges, the 2MASS J1315−2649 system is an important benchmark for empirical tests of atmospheric and evolutionary models given its proximity, well-separated components, component types, and relatively old age.