WISEP J004701.06+680352.1: AN INTERMEDIATE SURFACE GRAVITY, DUSTY BROWN DWARF IN THE AB DOR MOVING GROUP

A moderate-resolution near-infrared spectrum covering the range of 0.8 to 2.4 μm was obtained using the IRTF SpeX spectrograph (Rayner et al. 2003) in cross-dispersed mode (hereinafter SXD) on UT Date 2012 September 26. The resolution is 1200. The data were reduced using the facility pipeline Spextool (Cushing et al. 2004). Telluric features were corrected using a nearby A star using the method of Vacca et al. (2003). Spectra with the same observing setup were used by Allers & Liu (2013a) to develop a spectroscopic classification system that is sensitive to surface gravity: the key indicators include the K i lines in the J band and the FeH bands in J and H bands. In Figure 1, we compare our spectrum to other unusually red L dwarfs 2M2244+20, 2MASS J01033203+1935361 (Kirkpatrick et al. 2000; Faherty et al. 2012), and 2M2148+40. W0047+68 is similar to 2M2244+20, but has considerably lower surface gravity than 2M2148+40. Applying the Allers & Liu (2013a) system, we classify W0047+68 as L7 INT-G (Table 1).

Zoom In Zoom Out Reset image size

Table 1. WISEP J004701.06+680352.1

Parameter	W0047+6803	Remarks
Sp. type	L7 INT-G	Infrared, Allers & Liu (2013a) system
Sp. type	L7 pec	Optical, Kirkpatrick et al. (1999) system
πabs	82 ± 3	mas
μrel	434 ± 3	mas yr−1
θ	117.0 ± 0.5	degrees, east of north
vrad	−20.0 ± 1.4	km s−1
vsin i	4.3 ± 2.2	km s−1
z	18.95 ± 0.03	AB mag, Pan-STARRS
y	18.12 ± 0.03	AB mag, Pan-STARRS
U	−8.6 ± 1.0	km s−1
V	−27.8 ± 1.3	km s−1
W	−13.6 ± 0.5	km s−1
MJ	15.17 ± 0.11	2MASS
MH	13.54 ± 0.09	2MASS
MKs	12.62 ± 0.09	2MASS
MW1	11.47 ± 0.08	WISE
MW2	10.82 ± 0.08	WISE
MW3	9.71 ± 0.10	WISE
mbol	16.30 ± 0.03	Section 2.5
L/L☉	(3.58 ± 0.29) × 10−5	Section 2.5
BCJ	0.70 ± 0.08
BCK	3.25 ± 0.04
MJ (MKO)	15.05 ± 0.11	Color term from spectrum, see Paper I
MH (MKO)	13.65 ± 0.09	See Paper I
MK (MKO)	12.56 ± 0.09	See Paper I
$M_{M^{\prime }}$	10.6	Using 2M2244, see Section 2.4
If AB Dor Association member
Teff	∼1270	K, see Section 3.3
Age	∼120	Myr
Mass	∼0.018	M☉
log g	∼4.5
If older Galactic Disk member
Teff	∼1410	K; see Section 3.3
Age	⩽1000	Myr
Mass	<0.055	M☉
log g	∼5.0

Download table as:  ASCIITypeset image

High-resolution near-infrared spectra of W0047+68 were obtained with Keck/NIRSPEC (McLean et al. 1998, 2000) on UT Date 2013 September 17 to determine W0047+68's radial velocity and projected rotational velocity (vsin i). Data were obtained using the N7 filter and 0432 × 12'' slit (3 pixel dispersion), providing 1.99–2.39 μm spectroscopy in seven orders at a measured resolution of 25,000 (Δv = 12 km s⁻¹). Two dithered exposures of 1200 s were obtained at an airmass of 1.50, followed by observations of a nearby A star. Data in the order of 33, spanning 2.293–2.319 μm, were extracted and analyzed due to the strong CO bands present in this band. We used a combination of REDSPEC¹² and custom IDL routines that perform a forward-modeling Monte Carlo Markov Chain analysis of the extracted spectrum using solar-metallicity BT-Settl models (Allard et al. 2011) and a telluric absorption spectrum from Livingston & Wallace (1991); this analysis will be described in detail in a forthcoming paper (A. J. Burgasser et al., in preparation; see Blake et al. 2010 for a similar algorithm). Figure 2 displays our extracted spectrum and the best-fit model. The individual best-fit model has v_rad = −21.5 km s⁻¹and vsin i = 2 km s⁻¹, but marginalizing over the fit parameters, including other temperatures and gravities, we find vsin i = 4.3 ± 2.2 km s⁻¹and a heliocentric radial velocity of −20.0 ±1.4 km s⁻¹, which we adopt as our best estimates for W0047+68. Our modeling requires higher resolution than is publicly available for the full BT-Settl grid; the subset of the grid that we use has T_eff > 2000 K, hotter than we believe W0047+68 can possibly be. However, these CO lines are not very sensitive to temperature. We show in Figure 2(b) a comparison of the T_eff = 1400 K, log g = 4.0 model (which we discuss later) to the hotter models used by the modeling routine at the highest resolution available. The lines are very similar, suggesting that the radial velocity and vsin i, but not the temperature or gravity, can be extracted from the Keck data.

Zoom In Zoom Out Reset image size

Optical (far-red) long-slit spectra were obtained with the MMT and Gemini-North telescopes. The MMT observations were on UT Date 2012 August 26 with the Red Channel spectrograph with grating 270, using three 600 s exposures. Conditions were non-photometric. This time was allocated by NOAO as program 2012B-0233. The wavelength coverage was 6170–9810 Å with a resolution of ∼12 Å, but we make no use of the spectrum redder than 9250 Å due to strong telluric water absorption. The Gemini-North observations (Gemini program GN-2012B-Q-105) were on UT Date 2012 September 11 with the GMOS spectrograph (Hook et al. 2004) using grating R831. Four 600 s exposures were taken during cloudy conditions. The wavelength coverage was 6340–8460 Å with a resolution of ∼2 Å, but for all plots, we smooth to a resolution of ∼10 Å due to low signal-to-noise. The spectra were processed using standard IRAF tasks.

The MMT and Gemini spectra are shown in Figure 3, along with 2M2244+20, classified as L6.5 in the optical (Kirkpatrick et al. 2008), and 2M2148+40, classified as L7 in the optical (Looper et al. 2008). The core of the K i doublet and the CrH and FeH features are shown in Figure 4 along with the original Kirkpatrick et al. (1999) L6–L8 standard star spectra. The CrH and FeH features most strongly support an L7 classification, and the K i doublet and overall appearance of the spectrum is consistent with this. It is evident, however, that the Rb i and Cs i lines are much weaker in W0047+68 than in the standard, or typical, L6–L8 dwarfs (Kirkpatrick et al. 1999, 2000), suggesting that a classification of "L7 peculiar" is appropriate. In Figure 5, we show the same comparison to the unusually red L dwarfs 2M0355+11 (Cruz et al. 2009), 2M2148+40 (Looper et al. 2008), and 2M2244+20 (Kirkpatrick et al. 2008). These spectra were classified as "L5γ," L6, and L6.5 respectively. The weak Cs i is also evident in 2M2148+40 and 2M2244+20. W0047+68's alkali lines are stronger than in the low gravity 2M0355+11, but 2M0355+11 is a warmer type. Given our near-infrared classification of INT-G, the atomic line peculiarities hint that W0047+68 might be classified as L7β in the optical, but we believe that extending the Cruz et al. (2009) system to such a late-type object would be premature without more objects. Both W0047+68 observations show lithium absorption, but no detectable Hα emission.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

W0047+68 is being observed as part of the USNO optical parallax program, which is described in Dahn et al. (2002). Based on 23 frames taken in two seasons, the preliminary absolute parallax is 82 ± 3 mas with relative proper motion μ = 434 ± 3 mas yr⁻¹ at position angle θ = 117.0 ± 05. We derive the U, V, W space velocity of W0047+68 using the observed radial velocity and astrometry and the IDL code gal_uvw.pro, except that we adopt the traditional astronomer convention that U is positive toward the Galactic center. Note that W is strongly constrained (±0.5 km s⁻¹) by the astrometry because of the position on the sky (l = 122.5, b = +05.2). We note that the absolute magnitude at the H band agrees with that of typical L7 dwarfs (Dupuy & Liu 2012); W0047+68 is fainter ("sub-luminous") at shorter wavelengths (J band) and brighter ("over-luminous") at longer wavelengths (K, WISE).

Few trigonometric parallaxes of unusually red L5 or later dwarfs exist, complicating comparisons of W0047+68 to other objects. We are able to also report a preliminary USNO optical parallax of 2M2148+40, which was classified as L6 in the optical and L6.5 peculiar in the near-infrared by its discovery paper (Looper et al. 2008) and is L6 FLD-G in the Allers & Liu (2013a) system. The observed absolute parallax is 124.07 ± 0.55 mas, with a proper motion of 901.9 ± 0.3 mas yr⁻¹ at an angle of 591 ± 01 (Table 2). The distance of 8.060 ± 0.036 pc places it just outside the "8 pc" sample. The radial velocity is unknown, but because of the position on the sky (l = 89.0, b = −10.5), this mainly affects the V component of space motion. U is strongly constrained: if we adopt v_rad = 0 ± 40 km s⁻¹, then U =−34.0 ± 0.5 km s⁻¹, and W=−5.6 ± 7 km s⁻¹. This rules out membership in any of the known very young nearby moving groups. 2M2148+49 is intrinsically fainter at the J band than ordinary field L6 dwarfs and brighter at WISE bands.

W0047+68 was observed as part of normal survey operations of the Pan-STARRS wide-field optical/NIR PS1 survey (Kaiser et al. 2010). We report AB magnitudes for the y and z bands, which are described in Tonry et al. (2012). The source was not detected in the shorter wavelength bands. WISE and Two Micron Sky Survey (2MASS) photometry were previously reported in Paper I.

We monitored W0047+68 in the J band using the FLAMINGOS camera (Elston et al. 2003) on the Kitt Peak 2.1 m telescope in 2012 October (NOAO-Program 12B-0233) on three consecutive nights for four hours per night, after the main target for the program had set. We find no evidence of variability at the level of 0.05 mag or greater.

Although unobservable with WISE, many of the directly imaged planets have been observed at L' and M'. W0047+68 has not been observed through these filters, but we may estimate the magnitudes using WISE photometry and other L dwarfs. Leggett et al. (2007) measured M' = 11.90 ± 0.03 for 2M2244+20. Scaling by the W2 magnitudes, this suggests M' = 11.04 for W0047+68 and $M_{M^{\prime }} = 10.6$.

We can obtain the bolometric magnitude of W0047+68 by integrating the spectral energy distribution, improving the estimate from Paper I using our new data. The near-infrared component is well-constrained by the Paper I spectrum normalized to agree with the 2MASS photometry. In the optical, we use the MMT spectrum. There is negligible contribution at less than 0.6 μm. We use the Spitzer Infrared Spectrograph (IRS) spectrum of 2M2244+20 (Stephens et al. 2009) renormalized by a factor of 2.23 (from the relative WISE W1 and W2 photometry) in the range of 5.15 μm–14.15 μm and a Rayleigh–Jeans tail beyond that. This spectral energy distribution is shown in Figure 6. The wavelength range between 2.55 μm and 5.15 μm is constrained by the WISE photometry, but we require higher resolution to account for the non-smooth spectrum. We consider the A-type (1200 K) and AE-type (1300–1500 K) models of Madhusudhan et al. (2011) and the preliminary BT-Settl (1300–1500 K) of Allard et al. (2013). In each case, we integrate the model spectra over the range of 2.55–4.00 μm, normalized to the WISE W1 photometry, and the range of 4.00–5.15 μm, normalized to the WISE W2 photometry. (Because the observed colors are redder than the models, we allow discontinuities.) Averaging the three best-fitting near-infrared models (A 1200 K, AE 1300 K, and BT-Settl 1400 K, all log g = 4.0) we obtain an apparent bolometric magnitude of 16.30 with an uncertainty of ±0.02 magnitudes due to the spread in models, all of which feature deep water bands. If blackbodies without absorption bands are used for this region, the resulting bolometric magnitude is up to 0.05 mag brighter, while an observed AKARI spectrum of an ordinary L8 dwarf (Sorahana & Yamamura 2012; Sorahana et al. 2013) normalized to the W1 photometry is 0.02 mag fainter; other, higher-gravity bluer models are within this range of uncertainty. There is an additional uncertainty from the uncertainty in the 2MASS and WISE photometry and other spectra, which we estimate is ±0.025 mag. The derived W0047+68 luminosity is (3.58 ± 0.29) × 10⁻⁵ L_☉. The cumulative contribution to the luminosity is also shown in Figure 6: half of the energy is emitted redward of 2.9 μm. Following the same procedure for 2M2148+40, but with the BT-Settl 1500 K model yields an apparent bolometric magnitude of 15.12 ± 0.03 and luminosity of L = (4.64 ± 0.14) × 10⁻⁵ L_☉.

Zoom In Zoom Out Reset image size

The spectra of W0047+68 include a number of age indicators. The most conservative age estimate for W0047+68 comes from the fact that we detect lithium: for the observed luminosity, this indicates that its age is less than one billion years old (Chabrier et al. 2000). The near-infrared spectroscopy, however, points to an even younger age, less than that of most lithium L-type brown dwarfs. Thompson et al. (2013) suggested that W0047+68 is a "low-gravity object and hence young," based on their analysis of a moderate-resolution spectrum similar to the one we have obtained. Our near-infrared surface gravity classification of INT-G (Section 2.1) indicates an age of ∼100–150 Myr for W0047+68, but not <50 Myr, though this system is primarily based upon warmer objects. Many of the youngest stars are found to be members of moving groups.

The three-dimensional space velocity (Figure 7) is consistent with the young AB Dor Moving Group (Zuckerman et al. 2004; Torres et al. 2008).¹³ Most ABDMG members are in the southern hemisphere, but in fact the group is quite loose and includes northern hemisphere members. Given the parallax, proper motion, radial velocity, and lithium age constraint, we find that the BANYAN II model (Gagné et al. 2013) assigns an ABDMG membership probability of 99.96% to W0047+68. As seen in Figure 7, W0047+68 lies fairly close to the group in physical space, and the peculiar velocity necessary to move W0047+68 from the nucleus of the ABDMG to its present position is <1 km s⁻¹. Furthermore, as discussed in detail in Section 3.5, mid-L dwarf ABDMG candidate members, including the companion CD−35 2722B (Wahhaj et al. 2011) which is a definite member, are unusually red in J − K color.

Zoom In Zoom Out Reset image size

Therefore, on the basis of three indicators (three-dimensional space motion, spectroscopic surface gravity, and near-infrared color), we propose that W0047+68 is a bona fide member of the ABDMG. One could ask if it is plausible that an ABDMG brown dwarf member is so close to the Sun. Torres et al. (2008) list eight ABDMG M dwarf members within 16 pc of the Sun and Kirkpatrick et al. (2012) find that the ratio of brown dwarfs to stars is 1:6, so the presence of one or two nearby ABDMG brown dwarfs is not surprising.

We can leverage the current knowledge of the ABDMG, which is based on a wide variety of stellar masses and spectral types, to understand the properties of W0047+68 and the origin of its peculiar spectrum. Although early discussions of ABDMG (Zuckerman et al. 2004) assumed an age of ∼50 million years, current evidence points to an age of ∼100–125 Myr (Luhman et al. 2005; Barenfeld et al. 2013). Indeed, Luhman et al. (2005) argue that ABDMG is "roughly co-eval" with the Pleiades and is physically related given the similar space velocities, an argument supported by Ortega et al. (2007). This age is in excellent agreement with our spectroscopic estimate of surface gravity intermediate between the youngest association and older field brown dwarfs. Of course, even the age of the rich, well-studied Pleiades is still somewhat uncertain (see Soderblom et al. 2009), but W0047+68's age constraint is much better than that of most known brown dwarfs.

The relative importance of surface gravity and composition is debated in both unusually red brown dwarfs and massive gas giant planets. Crucially, the composition of the ABDMG is also known to be near solar. Ortega et al. (2007) compiled the recent literature measurement of 15 ABDMG members and found a mean [Fe/H] = −0.02 ± 0.02. Viana Almeida et al. (2009) measured a mean [Fe/H] = −0.01] for 12 ABDMG members, with a standard deviation of 0.09, but after applying a correction for a trend in effective temperature found a mean [Fe/H] = 0.04 with a standard deviation of 0.05. They also reported that for their entire young star sample the mean values are [Si/H] = −0.11 and [Ni/H] = 0.01. Biazzo et al. (2012) reported a mean [Fe/H] = 0.10 ± 0.03 for the five stars in the ABDMG, and found that 10 elements (Na, Mg, Al, Si, Ca, Ti, Cr, Fe, Ni, and Zn) have abundance patterns that are solar and "typical" of the local Galactic thin disk. Barenfeld et al. (2013) claim a somewhat lower [Fe/H] = 0.02   ±   0.02 for the ABDMG nucleus and measure seven other elements (Na, Mg, Al, Si, Cr, Mn, and Ni) to be close to solar values. Since these [Fe/H] values are characteristic of the bulk of the local thin disk, the vast majority of field L dwarfs must have similar values. Whether the scatter in values is a matter of observational uncertainties or true scatter could be debated. Barenfeld et al. (2013) argue that they see real scatter in composition in the ABDMG "stream" members and that, therefore, many of the more distant ABDMG objects are not from a common origin. Nevertheless, even these objects are all near solar in [Fe/H] and other elements, as is the Pleiades (Soderblom et al. 2009), so we conclude that W0047+68's composition is constrained to be near-solar even if the ABDMG stream is not truly homogeneous. This constraint includes the key elements involved in cloud formation: Fe and Al (along with O) are the major elements for high-temperature condensates (i.e., in the deeper atmospheric layers of W0047+68) and Mg and Si (and O) at moderate temperatures characteristic of the upper atmospheric layers (Marley et al. 2002; Helling et al. 2008b; Visscher et al. 2010).

Using the Chabrier et al. (2000) models and the observed luminosity, W0047+68 would have log g ≈ 4.5 and M ≈ 0.018 M_☉ for age 120 Myr. Similar values are found with the Burrows et al. (1997) models. The low surface gravity and solar metallicity suggest that the thick clouds of W0047+68 are a normal feature of a young brown dwarf. In Figure 8, we show the color–magnitude diagram for brown dwarfs and planetary mass objects (Liu et al. 2013b), updated with W0047+68 and 2M2148+40. W0047+68 lies close to PSO J318-22, adding to the evidence of a low-gravity sequence that extends redward of the normal L/T brown dwarf sequence.

Zoom In Zoom Out Reset image size

In Paper I, we found that although models from different groups agreed that unusually thick clouds were needed to fit the near-infrared spectrum, the predicted effective temperatures ranged from as high as 1600 K to as low as 1100 K. With our measurement of the luminosity, we can make another estimate of the effective temperature using theoretical predictions of the radius. We find

$$\begin{equation} T_{\rm eff} = (1413\,{\rm K} \pm 28\,{\rm K) } \left({0.1\,R_\odot }\over {R}\right)^{0.5}. \end{equation} \tag{ 1 }$$

The highest temperature and smallest radius for W0047+68 can be obtained by using the oldest evolutionary models in which lithium has not been depleted. This would allow T_eff ≈ 1450 K and an age of ∼1 billion years, with log g ≈ 5. ABDMG membership implies a larger radius and lower temperature. Adopting an age of 125 Myr for ABDMG membership, Burrows et al. (1997) have M_☉ = 0.020 M_☉ and R = 0.117 R_☉; Chabrier et al. (2000) have M = 0.018 M_☉ and R = 0.124 R_☉. Correspondingly, our best estimate of W0047+68's effective temperature is T_eff ≈ 1270–1300 K with log g ≈ 4.5. Burrows et al. (2011) report that high metallicity or thick cloud models each lead to an increase in the radius over "standard" models, which would lower the temperature below 1400 K even if W0047+68 is old.

Given an effective temperature of ∼1300 K, W0047+68 shows that intermediate surface gravity brown dwarfs have a much different L/T-type transition. For older, high gravity field brown dwarfs, this effective temperature is normally associated with the early T dwarfs (Cushing et al. 2008; Stephens et al. 2009) and certainly with spectral types that have already begun the move to the blue in J − K. The early T dwarfs have methane bands in the H band and K band (Burgasser et al. 2006). No such features are seen in W0047+68, and we also did not detect the high amplitude J-band variations seen in ∼1300 K early T dwarfs and attributed to patchy clouds (Artigau et al. 2009; Radigan et al. 2012; Apai et al. 2013). (However, W0047+68's low vsin i suggests that we may be viewing it near pole-on.) We emphasize that this result depends upon the evolutionary models: W0047+68 has a luminosity typical of that of ordinary L7 dwarfs (Vrba et al. 2004; Golimowski et al. 2004), but its young age (and theoretically larger radius) requires a lower temperature. As shown by Liu et al. (2013b), the younger object PSO J318-22 is an even more extreme case, with nearly the same luminosity and spectral type as W0047+68, but a deduced temperature of $1160^{+30}_{-40}$ K.

The properties of condensate cloud layers depend on complex processes that are difficult to model completely, including nucleation, growth, and settling (Ackerman & Marley 2001; Helling et al. 2008b). Each theoretical effort to predict synthetic spectra makes different choices in parameterizing the cloud properties. These were compared by Helling et al. (2008a) in test cases aimed at modeling typical field brown dwarfs. In Paper I, we discussed how theoretical models from three different groups (UCM; Tsuji 2005; f_sed, Ackerman & Marley 2001; Marley et al. 2002; Stephens et al. 2009; very thick (A-type) and intermediate thickness (AE-type) cloud models, Madhusudhan et al. 2011) were able to fit the low-resolution W0047+68 near-infrared spectrum. In each case, models can fit the J − K color and qualitatively reproduce the spectrum by adjusting a free parameter to increase the thickness of the condensate clouds. Another set of models, DRIFT-PHOENIX (Witte et al. 2009), was used by Witte et al. (2011) to fit low-resolution near-infrared spectra of 2M2148+40 and 2M2244+20. They derived warm temperatures (T_eff = 1500 K) and very low gravities (log g = 3.0 and 3.5, respectively), but discussed the model grid limitations for such red objects and concluded that their temperatures are likely overestimated and the gravities underestimated.

We can now also compare to a fifth set of models. Allard et al. (2012, 2013) have presented a new model family called BT-Settl. The mixing and cloud properties are informed by two-dimensional radiation hydrodynamic (hereafter RHD) simulations (Freytag et al. 2010). The result is that for cool L dwarfs, low-gravity objects are predicted to be redder, with thicker clouds. We show the W0047+68 observed spectrum with the model predictions in Figure 9. While our analysis of W0047+68 favors T_eff ≈ 1300 K and log g ≈ 4.5, the BT-Settle grid requires an even lower gravity but higher temperature (T_eff = 1400–1500 K, log g = 4.0) to match the spectrum. (Note that the log g = 3.5 BT-Settl models are even redder and can also fit.)

Zoom In Zoom Out Reset image size

It has long been recognized that J − K colors are relatively poor temperature indicators for brown dwarfs, even among the typical field population because of the effects of clouds (Marley et al. 2002; Knapp et al. 2004; Stephens et al. 2009). Our results again underline this issue. The Unified Cloudy Model and BT-Settl give qualitatively good fits to W0047+68, but are warmer by 100–200 K than our estimate of 1300 K, while, on the other hand, the very thick (A-type) and f_sed = 1 models give temperatures of 100–200 K too low when matching the spectra. The plots in Figure 9 have arbitrary normalizations. An interesting check is to derive the radius from the near-infrared region alone instead of using our procedure in Section 3.3: given a model, the observed parallax, and the observed flux-calibrated spectrum, the only adjustable parameter is the radius. The BT-Settl results when fitting over the range of 1.2–2.5 μm with the log g = 4.0 models are 0.13 R_☉ (1200 K), 0.12 R_☉ (1300 K), 0.94 R_☉ (1400 K), and 0.82 R_☉ (1500 K), consistent with the discussion in Section 3.3. The 1500 K model has a qualitatively good fit but requires a radius that is considerably smaller than we expect for a brown dwarf with lithium. We believe the balance of evidence is against this possibility, though it should be noted that Sorahana et al. (2013) have used their own model fitting to AKARI data to argue that (ordinary) field late-L dwarfs are warmer and much smaller than predicted. Cushing et al. (2008) showed that fits to limited wavelength regions can almost always work, but can give much different effective temperatures than fits to the entire spectral energy distribution.

One important conclusion from our work is that model colors should not be used to derive surface gravities because they cannot yet predict cloud properties well enough. In the cases with a full grid of surface gravities (Madhusudhan et al. 2011; Allard et al. 2012), one must go to log g = 4.0 or even less to get thick enough clouds to match in the infrared. This would in turn require that W0047+68 be below the deuterium-burning limit and very young, much younger than the best estimates of the ABDMG age. For example, in the Baraffe et al. (2002) models at 20 Myr, a 0.007 M_☉ would have log g = 4.0 and match the observed luminosity with T_eff = 1200 K, while a 0.009 M_☉ would have log g = 4.1 and T_eff = 1400 K but would be 90% more luminous than observed. Manjavacas et al. (2014) have argued that the high altitude dust in BT-Settl models should be increased to better fit the warmer (M9.5–L3) young brown dwarfs; the same is evidently true for W0047+68.

The optical spectrum of W0047+68 also poses a challenge for our understanding. The spectrum is dominated by the broad wings of the K i which, in turn, are the dominant feature that can be used for spectral typing; these are evidently saturated around 1300 K even though their weakness is an indicator of low gravity in warmer objects (Cruz et al. 2009). This was recognized by Kirkpatrick et al. (2008) who argued that 2M2244+20 is young, yet found a normal L dwarf spectrum (shown in our Figures 3 and 5). They speculated that "the lower gravity (optical) spectrum will have weaker alkalis and stronger hydrides than a high-gravity L dwarf of comparable temperature." This, in turn, creates "a degeneracy at mid- and late-L-types whereby low-gravity objects are very difficult or impossible to distinguish in the optical." If our interpretation of W0047+68 is correct, this must be the case because (except for the strengths of the Cs and Rb lines) it is very close to the L7 standard. Another important factor in creating degeneracy may be additional opacity due to dust, which we know must be different in the near-infrared and may be different in the optical. Pavlenko et al. (2000) discussed the optical spectra of L dwarfs and showed that the broad K i line and the sharper Cs i and Rb i lines in the T_eff = 1200–1500 K range are sensitive to surface gravity and dust extinction; they could change the best-fit effective temperature and surface gravity by increasing or decreasing them together, and additional opacity helped the quality of fits. Lodders (1999), Lodders & Fegley (2002), and Marley et al. (2002) discuss the sensitivity of the K i chemistry to sedimentation and cloud formation. Thus, it may be possible to find models where the "L7" optical spectra are relatively insensitive to gravity.

Synthetic optical spectra are available for the BT-Settl and Madhusudhan et al. (2011) model families, and have not been tuned to match observations. In Figure 10, we compare the predicted and observed far-red spectra using the Madhusudhan et al. 2011 very-thick (A) and intermediate thickness (AE) models normalized to agree in the H band (as in Figure 6 of Paper I). In both cases, the absorption by the broad K i wings is underestimated and, as a result, the continuum is too high. Since the optical and near-infrared spectra of typical L dwarfs are fit well by the "E type" thin cloud models and log g = 5.0 (Burrows et al. 2006), one would expect them to also fit the optical spectrum of W0047+68. Such a fit is also shown in Figure 10. The optical BT-Settl models are shown in Figure 11. There are also disagreements with the observations, with the pseudo-continuum lying too low, but the T_eff = 1300 K, log g = 4.0 model is reasonably similar in overall morphology and slope, though it lacks the hydrides; CrH and FeH are better matched with the log g = 4.5 models. The change in shape of the core of the K i resonant doublet in the log g = 4.0 models from 1500 K to 1300 K is similar to the differences between 2M0355+11 and W0047+68 (Figure 5), and suggests that the basic picture in which the core of the K i feature is saturated for late L dwarfs even for low gravities may be correct. In both the BT-Settl and A/AE models, the log g = 4.0 models have no CrH or FeH features: this prediction could be tested in the β Pic Moving Group planetary mass object PSO J318-22.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

There are now a number of L-type¹⁴ brown dwarfs proposed as ABDMG members. The strongest members, CD−35 2722B (Wahhaj et al. 2011), 2M0355+11 (Faherty et al. 2013b; Liu et al. 2013a), and W0047+68, show a surprising spread in spectral properties. 1RXS J2351+3127 B (Bowler et al. 2012) and 2MASS 01222439 B (Bowler et al. 2013) still need kinematic confirmation. The similarity of 2M2244+20 and W0047+68 supports ABDMG membership for 2M2244+20; using the W0047+68 absolute magnitudes for 2M2244+20 implies a distance consistent with ABDMG tangential motion (Gagné et al. 2014). The signal-to-noise on the 2M2244+20 spectrum is too low for a reliable measurement of lithium. In Figure 12, we plot the intermediate-resolution near-infrared spectra of 1RXS J2351+3127 B, CD−35 2722B, 2MASS 01222439 B, 2M0355+11 (Liu et al. 2013a), and W0047+68.

Zoom In Zoom Out Reset image size

Allers & Liu (2013a) have discussed the discrepancies between 2M0355+11 (L3 VL-G) and CD−35 2722B, which is L3 INT-G and a companion to one of the nucleus AB Dor members. Most notably, 2M0355+11 is more strongly peaked in the H band. Despite the very similar J − K_s colors of W0047+68 and 2M0355+11, the surface gravity sensitive features are very different, and W0047+68 appears to be cooler. 2M0355+11 is also a magnitude brighter in the color–magnitude diagram (Figure 8). The Pleiades has a similar age and space motion, and infrared spectroscopy shows early L dwarfs in the Pleiades are unusually red (Bihain et al. 2010). Allers & Liu (2013b) report that these spectra can be classified as VL-G or INT-G but show a spread in properties. Unfortunately, no late-type L dwarfs in the Pleiades are yet known, but our adopted age of ABDMG implies that W0047+68-like members should be found in deep Pleiades proper motion searches. Together, the ABDMG moving group brown dwarfs form an empirical intermediate surface gravity (log g = 4.5) sequence.

One puzzle is that 2M0355+11 has two very obvious indicators of low surface gravity—the very peculiar optical spectrum (L5γ), a very sharply peaked H band—while W0047+68 has more subtle features. The inconsistency of the H-band shape as a gravity indicator is discussed in detail by Allers & Liu (2013a). With regard to this pair of objects, three possibilities merit discussion. One possibility is that both 2M0355+11 and W0047+68 are normal ABDMG members whose differences are simply a consequence of temperature. Liu et al. (2013b) have found that the luminosity of 2M0355+11 is log L/L_☉ = −4.23 ± 0.11, i.e., ∼60% more luminous than W0047+68. This suggests that it is ∼170 K warmer than W0047+68, consistent with the two spectral types' earlier classification in the optical. The optical observed differences in the K i core (Figure 5) appear qualitatively similarly to the BT-Settl offsets between the 1500 K/1400 K models and the 1300 K models for log g = 4.0; the BT-Settl low-gravity models are most "peaky" at 1500 K, but the H band becomes more rounded for lower temperatures. The A/AE models are also not "peaky" at the H band for 1300 K. A less likely possibility is both 2M0355+11 and W0047+68 are ABDMG members, but 2M0355+11 is actually a lower mass object very close to the deuterium-burning limit, resulting in a higher luminosity and temperature than W0047+68, but a lower surface gravity. This kind of complex behavior near the deuterium-burning limit at ages up to ∼120 Myr appear in all published model tracks and have recently been discussed again theoretically by Spiegel et al. (2011) and observationally by Allers et al. (2010) and Bowler et al. (2013). Whether or not this suggestion matches in detail depends on the age of ABDMG and the exact behavior of model tracks near the deuterium-burning limit. The third possibility is that either, or both, are not ABDMG members. Of course, if 2M0355+11 has a lower gravity then it is a younger interloper, perhaps ∼30–50 Myr old. This may not be as unlikely as it seems. Pre-Hipparcos work identified a "Local Association" (also known as the "Pleiades Super-Cluster" or "Pleiades Moving Group") including many young stars in the age range of ∼50–150 Myr (Eggen 1992; Jeffries 1995), with a dispersion of ±10 km s⁻¹. The ABDMG is part of this broader association. Montes et al. (2001) identified many young stars with Local Association motions, and found kinematic "substructure" as well as a fairly large spread in velocities. Makarov (2003) confirmed the Local Association streaming motion in an X-ray selected sample. Antoja et al. (2008; see their Figure 13) find many young (<100 Myr) stars concentrated in this broad region of U – V space. More controversially, López-Santiago et al. (2006) argued that the AB Dor group could be divided into a ∼30–50 Myr-old subgroup and a ∼80–120 subgroup. (There are also, of course, old stars with similar motion so kinematics alone cannot prove youth. In a non-age-selected Hipparcos sample, Bovy et al. 2009 find that ∼10% of stars can be associated with a Pleiades moving group.)