THE EXTREMELY RED, YOUNG L DWARF PSO J318.5338−22.8603: A FREE-FLOATING PLANETARY-MASS ANALOG TO DIRECTLY IMAGED YOUNG GAS-GIANT PLANETS

The colors of PSO J318.5−22 are extreme, with (J − K)_MKO = 2.78 mag, (J − K)_2MASS = 2.84 mag, and (W1 − W2) = 0.76 ± 0.04 mag, all being the reddest among field L dwarfs (Figure 1 and also see Gizis et al. 2012). Such colors are thought to arise from an unusually dusty atmosphere that results from a low surface gravity (young age). The position of PSO J318.5−22 on the near-IR color–magnitude diagram is similarly extreme, being significantly fainter in J-band absolute magnitude than field L dwarfs (Figure 1). It coincides with the colors and magnitudes of the directly imaged planets around HR 8799 and 2MASS J1207−39b.

Zoom In Zoom Out Reset image size

We determine the near-IR spectral type of PSO J318.5−22 using the Allers & Liu (2013) system, which provides gravity-insensitive types consistent with optical spectral types. For late-L dwarfs, visual classification in the J and K bands and index-based classifications with the H₂OD index are applicable. For the GNIRS spectrum of PSO J318.5−22, we visually assign a J-band type of L9 ± 1 and a K-band type of L6 ± 1. The H₂OD index corresponds to L6.0 ± 0.8. The weighted mean of these three determinations leads to a final type of L7 ± 1. Spectral typing of our low-resolution SpeX spectrum gives the same classification (Table 1).

PSO J318.5−22 shows a triangular H-band continuum, which is considered a hallmark of youth (e.g., Lucas et al. 2001). However, Allers & Liu (2013) caution that very red L-dwarfs having no signatures of youth (low gravity) can display a triangular H-band shape (e.g., 2MASS J2148+40 in Figure 2). At moderate resolution, there are other indicators of youth for late-L dwarfs. Our GNIRS spectrum displays a weak 1.20 μm FeH band as well as weak Na i (1.14 μm) and K i (1.17 and 1.25 μm) lines, which indicate a low gravity. Using the gravity-sensitive indices of Allers & Liu (2013), we classify PSO J318.5−22 as vl-g, which Allers & Liu suggest correspond to ages of ∼10–30 Myr based on the (small) sample of young late-M/early-L dwarfs with good age constraints. Altogether, PSO J318.5−22 visually appears most similar to the red L dwarfs WISE J0047+68 (Gizis et al. 2012) and 2MASS J2244+20 (McLean et al. 2003), in accord with the similar spectral types and gravity classifications for these three objects.

Zoom In Zoom Out Reset image size

Allers & Liu (2013) note that objects of the same age and spectral type (temperature) can display different spectral signatures of youth, based on two young L dwarfs in the AB Dor moving group. Our new discovery affirms this idea. The spectra of 2MASS J1207−39b and PSO J318.5−22 are quite different (Figure 2), despite their similar colors and absolute magnitudes and the fact they may be coeval (Section 3.2). PSO J318.5−22 shows a negative continuum slope from 2.12–2.28 μm, whereas 2MASS J1207−39b has a positive slope. The H-band continuum of PSO J318.5−22 displays a "shoulder" at 1.58 μm, whereas 2MASS J1207−39b has a very peaked continuum. Overall, the spectrum of 2MASS J1207−39b appears most similar to that of 2MASS J0355+11 (both objects have near-IR types of L3 vl-g), despite their large differences in ages, colors, and absolute magnitudes. Altogether, these comparisons hint that determining relative ages and temperatures from NIR spectra may be unexpectedly complex.

To assess the physical parameters, we fit the low-resolution near-IR spectra of PSO J318.5−22 and 2MASS J0355+11 with the Ames/DUSTY model atmospheres (Allard et al. 2001) and the BT-Settl model atmospheres (Allard et al. 2011) with two different assumed solar abundances (Asplund et al. 2009 [AGSS] and Caffau et al. 2011 [CIFIST]). Since both objects have parallaxes, the scaling factors (R²/d²) from the fits provide an estimate of the objects' radii. Using χ² minimization, all three sets of models indicates that PSO J318.5−22 ($\mbox{T}_{{\rm eff}_{,{\rm atm}}}=1400-1600\ {\rm K}$) is 100–200 K cooler than 2MASS J0355+11 ($\mbox{T}_{{\rm eff}_{,{\rm atm}}}=1600-1700\ {\rm K}$). BT-Settl/AGSS models give the best fits, with the resulting temperatures being comparable to previous fitting of field L dwarfs but with both young objects having lower surface gravities (log (g) = 4.0–4.5 dex) than field objects (Cushing et al. 2008; Stephens et al. 2009; Testi 2009, though the number of fitted field objects is small). Most strikingly, the fitted radii are implausibly small (≈0.8 R_Jup), indicating discord between the data and models even though the quality of fit seems adequate by eye. A similar radius mismatch occurs in model atmosphere fitting of young dusty planets (e.g., Bowler et al. 2010; Barman et al. 2011a; cf. Marley et al. 2012).

There is no radial velocity (RV) for PSO J318.5−22, but we can place modest constraints on its space motion. Published RVs (Blake et al. 2010; Seifahrt et al. 2010; Rice et al. 2010; Shkolnik et al. 2012) find that low-gravity late-M and L dwarfs reside in the range of [−20, +25] km s⁻¹, with the smaller RV range compared to field objects being expected given the young ages. Figure 3 plots the spatial (XYZ) and kinematic (UVW) location of PSO J318.5−22 compared to the young moving groups (YMGs) from Torres et al. (2008).⁹ Its XYZ position is coincident with the β Pic, Tuc-Hor, and AB Dor moving groups. However, the velocity information makes the Tuc-Hor and AB Dor groups implausible, given the large offset from PSO J318.5−22. Overall, the match with the β Pic members is excellent, i.e., the velocity offset between the group and PSO J318.5−22 would be only 3 km s⁻¹ if PSO J318.5−22 has an RV of −6 km s⁻¹.

Zoom In Zoom Out Reset image size

We also computed the kinematic distance (d_kin) and angle θ from Schlieder et al. (2010). d_kin is an object's distance assuming that its tangential velocity (as derived from its observed proper motion) is the same as the mean tangential velocity of a YMG. Including the uncertainties in the β Pic group's UVW, PSO J318.5−22 has $d_{{\rm kin}}=20.8^{+1.1}_{-1.0}$ pc, in good agreement with its parallactic distance. θ gives the angle between the object's proper motion and that of the YMG at the object's sky position. PSO J318.5−22 has $\theta =8^{+4}_{-5}$, whereas the β Pic members from Torres et al. (2008) mostly have θ < 15 degs. In other words, for PSO J318.5−22 the amplitude and direction of its tangential velocity given its sky location agrees well with the space motion of known β Pic members.

Finally, we computed the YMG membership probability using the online calculator of Malo et al. (2013). Given the position and parallax of PSO J318.5−22, their Bayesian method reports a membership probability of 99.9% in the β Pic YMG and 0.01% in the field. (Removing the parallax as an input, their method gives 99.6% and a predicted distance of 22.5 ± 1.6 pc, in good agreement with our parallactic distance.) The absolute value of this method's probabilities is imperfect, since their input model for the solar neighborhood does not account for the relative numbers of objects in different YMGs and the field. However, the high probability for PSO J318.5−22 is comparable to the values computed for most of the known β Pic members (and exceeds the rest of them). The Malo et al. method also does not consider the fact that PSO J318.5−22 is spectroscopically young, which would boost the probability that it is a YMG member rather than a field object (since most field objects have old ages).

PSO J318.5−22, 2MASS J0355+11, and 2MASS J1207−39b are all moving group members and thus play an important role as "age benchmarks" (e.g., Pinfield et al. 2006; Liu et al. 2008) for very red L dwarfs. With a measured L_bol and an age estimate, we use the Saumon & Marley (2008) evolutionary models with f_sed = 2 to compute the remaining physical properties. (Results from the other commonly available evolutionary models are similar, comparable to the uncertainties.)

To derive L_bol, we combined our 0.9–2.4 μm low-resolution spectra flux-calibrated using the JHK photometry, the WISE W1 and W2 photometry, and the best-fitting atmospheric model (Section 3.1) for the flux at bluer and redder wavelengths. We integrated the fluxes and used the parallaxes to determine L_bol, with the uncertainties in the input data (spectra, photometry, and parallax) handled in a Monte Carlo fashion. We find $\log (\mbox{L}_{\rm bol}/\mbox{L}_{\odot })=-4.42\pm 0.06$ dex and −4.23 ± 0.11 dex for PSO J318.5−22 and 2MASS J0355+11, respectively, very similar to field mid/late-L dwarfs (Golimowski et al. 2004; Cushing et al. 2006). Young, very red L dwarfs are sometimes (and confusingly) described as "underluminous" compared to field objects, but we find that the luminosities of young and old field objects are comparable.¹⁰ Rather, these objects have fainter absolute magnitudes in the bluer near-IR bands (e.g., J band) but comparable L_bol's. They are therefore better described as being "very red."

For PSO J318.5−22, using an age of 12$^{+8}_{-4}$ Myr (distributed uniformly) from Zuckerman et al. (2001), we find a mass of $6.5^{+1.3}_{-1.0}$ M_Jup, temperature of 1160$^{+30}_{-40}$ K, and log (g) of 3.86$^{+0.10}_{-0.08}$ dex (Figure 4). Adopting a larger age range of 10–100 Myr (distributed uniformly), the mass becomes 12 ± 3 M_Jup, with T_eff and log (g) increasing only slightly (Table 1). Such cool temperatures correspond to a field ≈T5.5 dwarf (Golimowski et al. 2004) and yet PSO J318.5−22 shows no sign of CH₄ in its spectrum, similar to the situation for the planets around HR 8799 and 2MASS J1207−39. (e.g., Marois et al. 2008; Bowler et al. 2010; Patience et al. 2010; Barman et al. 2011a). The evolutionary-model radii are almost twice as large as those from our model atmosphere fits, as has previously been noted in analysis of directly imaged planets (Section 3.1).

Zoom In Zoom Out Reset image size

Our same calculations for 2MASS J0355+11 give a mass of $24^{+3}_{-6}$ M_Jup, temperature of 1420$^{+80}_{-130}$ K, and log (g) of $4.58^{+0.07}_{-0.17}$ dex, assuming an age of 125 ± 25 Myr (Barenfeld et al. 2013) based on its membership in the AB Dor moving group (Faherty et al. 2013; Liu et al. 2013). This object is more massive and hotter than PSO J318.5−22, as expected given its brighter absolute magnitudes and older age. Faherty et al. (2013) describe 2MASS J0355+11 as having a strong resemblance to a giant exoplanet. While it does have similar H and K-band spectra to 2MASS J1207−39b (but not in J band nor the overall SED shape; Figure 2), 2MASS J0355+11 is ≈2 mag brighter in J band (Figure 1) and ≈5× more massive than 2MASS J1207−39b. Likewise, the spectrum of HR 8799b has been shown to resemble some field dwarfs (e.g., 2MASS J2148+40 and 2MASS J2244+20 by Bowler et al. 2010 and 2MASS J2139+02 by Barman et al. 2011a), yet HR 8799b is much fainter and lower mass. This strengthens our finding that near-IR spectral morphology can not reliably determine the underlying physical properties of very red young objects.

The temperatures for these two objects are substantially lower then those measured in the same fashion (using L_bol, age, and evolutionary models) for field L dwarfs of comparable spectral types. Golimowski et al. (2004) find T_eff of 1700–1950 K for spectral types L3–L5 (corresponding to the near-IR and optical types for 2MASS J0355+11) and 1500 K for L7 (the near-IR type for PSO J318.5−22). Thus, the temperatures of these young, very red L dwarfs are ≈400 K cooler than comparable field objects. Bowler et al. (2013) show that young L-type companions tend to have cooler temperatures than (old) field objects of the same type. Our results show that this offset also occurs for young free-floating objects.