Resolved Near-Infrared Spectroscopy of WISE J104915.57-531906.1AB: A Flux-Reversal Binary at the L dwarf/T dwarf Transition
aa r X i v : . [ a s t r o - ph . S R ] M a r Submitted to ApJ 27 March 2013
Preprint typeset using L A TEX style emulateapj v. 5/2/11
RESOLVED NEAR-INFRARED SPECTROSCOPY OF WISE J104915.57 − Adam J. Burgasser , Scott S. Sheppard , and K. L. Luhman Submitted to ApJ 27 March 2013
ABSTRACTWe report resolved near-infrared spectroscopy and photometry of the recently identified brown dwarfbinary WISE J104915.57 − ± O and CO absorption features in thespectra of both components, with the secondary also exhibiting weak CH absorption at 1.6 µ m and2.2 µ m. Spectral indices and comparison to low-resolution spectral standards indicate componenttypes of L7.5 and T0.5, the former consistent with the optical classification of the primary. Relativephotometry reveals a flux reversal between the J - and K -bands, with the T dwarf component beingbrighter in the 0.95–1.3 µ m range. As with other L/T transition binaries, this reversal likely reflectssignificant depletion of condensate opacity across the transition, a behavior that may be enhanced inWISE J1049 − ± Subject headings: binaries: visual — stars: individual (WISE J104915.57 − INTRODUCTION
The low temperatures and luminosities of browndwarfs near the Sun are a consequence of their ageand inability to sustain core hydrogen fusion (Kumar1962; Hayashi & Nakano 1963). As these objects evolvedown the spectral sequence (through the M, L, T andY spectral classes; Kirkpatrick 2005), their spectral en-ergy distributions become increasingly complex as molec-ular compounds come to dominate photospheric opac-ity. These compounds include condensed species: min-erals and metals in late-M and L dwarf atmospheres(Lunine et al. 1989; Tsuji et al. 1996), sulfide and alkalisalts in late-T and Y dwarf atmospheres (Burgasser et al.2010b; Morley et al. 2012). Grain scattering and absorp-tion from these condensates cause significant redistribu-tion of flux, veiling of molecular bands, changes in the gaschemistry, modification of cooling rates, and even vari-ability if the condensates are not uniformly distributed inthe photosphere (Allard et al. 2001; Ackerman & Marley2001; Burrows & Sharp 1999; Saumon & Marley 2008;Radigan et al. 2012). Models of cool atmospheres pa-rameterizing “cloudiness” are now used in fits of browndwarf spectra (Stephens et al. 2009), and condensateclouds are also seen as playing a major role in shap- Center for Astrophysics and Space Science, Universityof California San Diego, La Jolla, CA 92093, USA; [email protected] Department of Terrestrial Magnetism, Carnegie Institutionof Washington, 5241 Broad Branch Rd. NW, Washington, DC20015, USA Department of Astronomy and Astrophysics, The Pennsyl-vania State University, University Park, PA 16802, USA Center for Exoplanets and Habitable Worlds, The Pennsyl-vania State University, University Park, PA 16802, USA ing the spectra of directly imaged giant exoplanets(Currie et al. 2011; Barman et al. 2011; Skemer et al.2012; Marley et al. 2012).Discerning the parameters relevant to condensate for-mation and evolution in cool atmospheres is partly hin-dered by uncertainties in the characteristics of individ-ual brown dwarfs, which for a given spectral type canspan a broad range of age, mass and composition. Re-solved brown dwarf binaries have proven to be useful inthis regard, as their common distance and genesis elimi-nates many of the uncertainties in their (relative) phys-ical and spatial characteristics. Such systems are alsoamenable to direct mass measurement (e.g., Lane et al.2001; Dupuy et al. 2009; Konopacky et al. 2010), allow-ing direct comparison to and tests of evolutionary mod-els. Binaries with L dwarf and T dwarf components havebeen particularly valuable in probing the disappearanceof mineral clouds at this spectral transition. A hand-ful are found to be “flux-reversal” pairs, in which thelater-type secondary is brighter in the 1.0–1.3 µ m regionthan its earlier-type (and overall more luminous) primary(Burgasser et al. 2006b; Liu et al. 2006; Looper et al.2008a). These binaries validate the 1 µ m brighten-ing found in near-infrared color magnitude diagrams offield L and T dwarfs (the “J-band bump”; Tinney et al.2003), and suggest that this brightening may be a gen-eral feature of brown dwarf evolution. Physical inter-pretation of this brightening remains under debate, how-ever. It may be evidence of a rapid and/or patchy disrup-tion of the condensate cloud layer (Burgasser et al. 2002;Knapp et al. 2004) or it may reflect variation amongsources due to metallicity, mass, age, rotation or otherphysical characteristics (Burrows et al. 2006). Unfortu- Burgasser et al.nately, binaries straddling the L dwarf/T dwarf tran-sition are relatively rare, and the compact separationstypical of brown dwarf pairs (1–10 AU; Burgasser et al.2007) means that they must be very near to the Sun tobe resolved.The recent discovery of the brown dwarf bi-nary WISE J104915.57 − − ± − consistent with a spectral type of L8 ± ⊙ (Rebolo et al.1992; Magazzu et al. 1993). The observation that thesecondary is fainter in the red optical (∆ i = 0.45) indi-cates that this component could be a late-L or T dwarf.In this article, we report resolved near-infrared pho-tometry and spectroscopy of the WISE J1049 − − OBSERVATIONS
Magellan/FIRE Prism Spectroscopy and Imaging
WISE J1049 − . ′′ J -band. We deployed both the cross-dispersed echelle mode ( λ/ ∆ λ ≈ λ/ ∆ λ ≈ ′′ ′′ along the 30 ′′ slit. We alsoobserved the A0 V star HD 99338 ( V = 8.26) in six 1 snodded exposures. NeAr and quartz lamp exposures, re-flected off of the Baade secondary screen, were obtainedwith the target and A0 V stellar observations for wave-length and pixel response calibration, respectively. Datawere reduced using the FIREHOSE low-dispersion reduc-tion package ( firehose ld ), which produces a 2D estimateof the sky spectrum to remove the background in each ex-posure (Kelson 2003), determines a 2D wavelength map-ping from the NeAr arc exposure, and extracts 1D sourceand A0 V spectra and variances. Modified routines from We follow the convention of Luhman (2013), labeling the south-eastern component A and the northwestern component B based ontheir relative i -band magnitudes. the SpeXtool package (Cushing et al. 2004) were used toscale and combine the individual spectra, correct for tel-luric absorption and apply a relative flux calibration fol-lowing the procedures described in Vacca et al. (2003).Signals-to-noise (S/N) for both components peaked at ≈
400 in the K -band region.We also obtained a resolved image of the binary usingFIRE’s acquisition camera, which has a 50 ′′ × ′′ field ofview (FOV), a pixel scale of 0 . ′′ J -band filter. Several 1 s exposures were obtained withthe binary on and off slit; these were median-combinedto produce an overall sky frame that was subtracted fromall images. We extracted a 9 . ′′ × . ′′ ×
49 pixels) sub-frame from a single image in which the binary is well-separated from the slit, shown in Figure 1 and analyzedbelow.
IRTF/SpeX Spectroscopy and Imaging
WISE J1049 − . ′′ . ′′ J -band). We de-ployed the 0 . ′′ λ/ ∆ λ ≈
120 spectra covering 0.7–2.5 µ m. In this case,the slit was aligned along the binary axis (position angleof 5 ◦ , 50 ◦ off parallactic) to obtain concurrent spectra,and six 90 s exposures were obtained in an ABBA ditherpattern at an airmass of 3.3–3.4. We also observed theA0 V star HD 92518 ( V = 6.87) at an airmass of 3.3 withthe slit aligned to the same position angle. Internal flatfield and Ar arc lamp exposures were obtained for pixelresponse and wavelength calibration. Data were reducedusing SpeXtool, applying standard settings. Due to thepoor seeing, we did not attempt to extract componentspectra; rather, we extracted the combined-light spec-trum of the binary using a wide aperture. Average S/Nwas roughly 400 in the J , H and K -band peaks, respec-tively. While combining the individual spectral frames,we verified that the variable cloud extinction during theobservation was grey and had minimal impact on theobserved spectral shape; however, differential color re-fraction may still be an issue, and is addressed below.We obtained images of the binary on the same nightusing the SpeX slit-viewing camera (60 ′′ × ′′ FOV, pixelscale 0 . ′′
12) in each of the MKO J , H and K filters,with the instrument oriented at a position angle of 0 ◦ .Four exposures were obtained in each filter using a two-point dither pattern with a 7 ′′ nod, with total inte-grations of 32 s, 56 s and 84 s, respectively. We in-terleaved these with observations of a nearby red star2MASS J10490107 − J = 10.75, J − K s = 1.16)for point spread function (PSF) calibration. Frames werepair-wise subtracted to remove sky background, mirror-flipped along the y -axis to reproduce sky orientation, and9 . ′′ × . ′′ ×
81 pixels) subframes were excised for anal-ysis. The frames with the best seeing in each of the filtersare shown in Figure 1. ANALYSIS
Spectral Characteristics of WISE J1049 − The reduced FIRE spectra of theWISE J1049 − Mauna Kea Observatory filter system; see Tokunaga et al.(2002) and Simons & Tokunaga (2002). ear-Infrared Spectra of WISE J1049 − J A B
JH K
Fig. 1.—
FIRE (upper left frame) and SpeX slitviewer images of the WISE J1049 − J , H and K -bands (labeled).All three images are oriented with North up and East to the left, and display 9 . ′′ × . ′′ Figure 2, scaled to their inferred absolute flux densitiesas described below. Strong absorption features typicalof late-L type brown dwarfs are present, notably deepH O bands at 1.4 and 1.9 µ m; strong CO absorption at2.3 µ m; marginally resolved Na I and K I doublets at1.14, 1.17 and 1.25 µ m; and a steep 0.8–1.1 µ m spectralslope, shaped primarily by the pressure-broadened redwing of the 0.77 µ m K I doublet (Burrows et al. 2000).WISE J1049 − at 1.6 µ m and 2.2 µ m, characteristic of anearly-type T dwarf (Burgasser et al. 2006a). Enhancedabsorption at 1.15 µ m can also be attributed in partto CH absorption. The near-infrared K I lines arestronger in the spectrum of this component, and ahint of FeH can be seen at 0.99 µ m (the Wing-Fordband). The overall near-infrared spectral energy dis-tribution of WISE J1049 − − ± ± ± − µ m spectra to low-resolutionspectral standards defined in Kirkpatrick et al. (2010),following the prescription for near-infrared classifica-tion outlined in that work. The best-match stan-dards for WISE J1049 − − µ m)02468 F λ ( - e r g / s / c m / µ m ) WISE J1049-5319A + 5WISE J1049-5319B H O H O H O COFeHK I K IK INa ICH CH CH CIA H ⊕ ⊕ Fig. 2.—
Reduced FIRE prism spectra of WISE J1049 − − × − erg s − cm − µ m − as indicated by the dotted lines. Major absorption features are labeled, aswell as regions of strong telluric absorption ( ⊕ ). µ m)0.00.20.40.60.81.01.2 N o r m a li ze d f λ WISE J1049-5319A µ m)0.00.20.40.60.81.01.2 N o r m a li ze d f λ WISE J1049-5319B
SDSS 0837-0000 (T1)
Fig. 3.—
Comparison of the FIRE prism spectra ofWISE J1049 − µ m is indicated; both sources are notablyredder than their comparison stars. SpeX data for the standardsare from Burgasser et al. (2006a) and Burgasser (2007). nificantly redder spectral energy distributions than theircorresponding standards. Finally, we compared the full0.9–2.4 µ m FIRE spectra to low-resolution templatesfrom the SpeX prism Spectral Libraries . We found bestmatches to the L8 SDSSp J085758.45+570851.4 and theL9.5 SDSSp J083008.12+482847.4 (Geballe et al. 2002)for WISE J1049 − ± ± Component Photometry: A Flux Reversal Binary
Resolved photometry by Luhman (2013) identifiedWISE J1049 − i -band, but inspection of the images in Figure 1 indicatesthat the two components “flip” in relative brightness,with WISE J1049 − J but fainterat K (WISE J1049 − H as well). To quantify the amplitude ofthis reversal, we performed PSF-fitting analyses on ourFIRE and SpeX images using a Monte Carlo MarkovChain (MCMC) technique. For the FIRE image, our ; see Burgasser et al.(2010a) for details. ear-Infrared Spectra of WISE J1049 − TABLE 1Classification Indices for WISE J104915.57 − WISE J1049 − − O-J 0.672 ± ± -J 0.870 ± · · · ± · · · O-H 0.681 ± ± -H 1.093 ± · · · ± O-K 0.716 ± ± -K 0.936 ± · · · ± · · · O-1.2 1.519 ± · · · ± O-1.5 1.751 ± ± -1.6 1.073 ± · · · ± -2.2 1.068 ± ± ± · · · ± · · · · · · L7.5 ± · · · T0.5 ± · · · Core Spectral Match · · · L8 ± · · · T1 ± · · · L7 ± · · · L9.5 ± · · · L7.5 · · ·
T0.5 · · ·
References . — (1) Burgasser et al. (2006a); (2) Geballe et al. (2002); (3)Kirkpatrick et al. (2010); (4) Based on comparison to SpeX templates and usingnear-infrared spectral types as computed in Burgasser (2007).
PSF model was a 2D ellipsoidal gaussian for which themajor and minor axes were allowed to vary separatelyin width and orientation. For the SpeX images, weused both gaussian profiles and images of the PSF starobtained in the same filter as independent checks, andfound that the PSF star provided much more robust fits.Following initial guesses for the pixel positions of bothprimary and secondary components and their integratedfluxes, our code explored the model parameter space (in-cluding gaussian PSF shape) in randomized steps drawnfrom a normal distribution, and compared model anddata at each step using the χ statistic. Sub-pixel shiftsfor the SpeX PSF model were made using a damped sincfunction based on code developed by John Spencer andMike Ressler. We found that chain lengths of 2000 (witha 200 step burn-in) were sufficient for convergence. Forthe FIRE analysis, we marginalized the distribution ofeach parameter in the single chain to determine uncer-tainties, and included a 5% systematic uncertainty toaccount for the non-gaussian PSF shape. For the SpeXanalysis, we used the mean and standard deviation of the16 binary and PSF image pairings in each filter as ouroverall measurement and uncertainty. Separations andposition angles from all four image sets were averagedafter weighting by individual uncertainties.Results from these fits are listed in Table 2. Therelative photometry confirms the observed flux rever-sal: WISE J1049 − ± J (combination of FIRE and SpeX analyses),0.02 ± H (a marginal detec-tion) and 0.29 ± K . The re-versal at J is highly significant, although not asextreme as that reported for the T1+T5 binary2MASS J14044941 − J = − ± − − µ m and 1.3 µ m,most notably at the 1.05 µ m Y - and 1.27 µ m J -band peaks, where the excessive flux reaches 40%. TABLE 2Relative Photometry forWISE J104915.57 − Parameter A B ∆ ρ ( ′′ ) · · · · · · . ′′ ± . ′′ ρ (AU) · · · · · · ± ◦ ) · · · · · · ◦ ± ◦ MKO J a ± ± − ± H ± ± − ± K ± ± ± M J ± ± · · · J − K ± ± · · · a Uncertainty-weighted average from FIRE( − ± − ± WISE J1049 − µ m H -band peak. Importantly, theseare the wavebands where condensate scattering opacityplays a prominent role in shaping brown dwarf spectra(Ackerman & Marley 2001). As shown in Figure 4, thesum of these scaled spectra are an excellent match tothe combined light SpeX spectrum of the binary, whichrequired modest reddening ( A V = 0 .
6) to reproduce thecombined light colors, likely to account for differentialcolor refraction at the high airmass of the observation. DISCUSSION
Our relative photometry allows us to infer in-dividual component magnitudes and colors forWISE J1049 − − J − K colors for L dwarfs have beenattributed to youth (Kirkpatrick et al. 2006), un-usually thick clouds (Looper et al. 2008b) or both(Currie et al. 2011; Faherty et al. 2013). The kinematics Burgasser et al. µ m)0.00.20.40.60.81.0 N o r m a li ze d f λ WISE J1049-5319AB (SpeX)WISE J1049-5319A (FIRE)WISE J1049-5319B (FIRE)Combined
Fig. 4.—
Relative flux scaling of the FIRE spectra of WISE J1049 − A V = 0.6 to account for differential color refraction). Note that the secondary is brighter than the primary between 0.95–1.3 µ m and1.55–1.70 µ m. The relative spectrophotometric magnitudes (MKO system) of the combination shown are ∆ J = − H = 0.02 and∆ K = 0.26. of WISE J1049 − ∼ < ⊙ , T eff ≈ − µ m flux, whereas reduced cloudopacity in the secondary allows light to emerge fromdeeper, hotter layers at these wavelengths, making thiscomponent relatively brighter. This argument has beenmade to explain the relative brightnesses of other flux-reversal binaries, but this is the first time resolved spec-troscopy has been reported for such a system. Cloud-induced variations in the degree of flux reversal have alsobeen previously suggested among candidate L/T transi-tion spectral binaries (Burgasser et al. 2010a).Why would there be a difference in the cloud proper-ties of these coeval brown dwarfs? Overall color trendsamong L and T dwarfs suggest that this binary may sim-ply straddle the evolutionary stage during which mineralclouds are disrupted, and the presumably lower-mass andlower-temperature secondary has begun to lose its clouds first. Note that the mass differential between the com-ponents could potentially constrain the timescale overwhich this evolutionary phase occurs (Burgasser 2007).An alternate—or possibly concurrent—hypothesis is thatthese two brown dwarfs have different cloud propertiesdue to other factors; e.g., differences in surface grav-ity, rotation rate or viewing orientation (metallicity orage differences are assumed negligible). In this case, thethermal evolution of the cloudy primary may have sloweddue to its greater condensate opacity (Saumon & Marley2008), potentially leading to a situation in which thelower-temperature secondary is the more massive com-ponent of the system.These hypotheses are potentially testable for this sys-tem, as its relatively tight separation of (3.13 ± ⊙ (appropriate for T eff ≈ . ′′ − (corresponding to RV dif-ferences of up to 1.6 km s − ). The RV velocity measure-ments are potentially feasible with current instrumenta-tion (Bean et al. 2010; Blake et al. 2010), and the rela-tively dense stellar field around WISE J1049 − − -1 0 1 2MKO J-K1615141312 M K O M J AB -1 0 1 2MKO J-K1615141312 M K O M J AB-1 0 1 2MKO J-K16151413121110 M K O M K AB -1 0 1 2MKO J-K16151413121110 M K O M K AB Fig. 5.—
Near-infrared color-magnitude diagrams (top: M J versus J − K ; bottom: M K versus J − K ) for field L and T dwarfs withreported parallax measurements having absolute magnitude and color uncertainties ≤ − − able over the course of a single year.The authors thank Hugo Rivera at Magellan andEric Volquardsen at IRTF for their assistance with theobservations. K.L. acknowledges support from grant NNX12AI47G from the NASA Astrophysics Data Anal-ysis Program. The Center for Exoplanets and HabitableWorlds is supported by the Pennsylvania State Univer-sity, the Eberly College of Science, and the PennsylvaniaSpace Grant Consortium.Facilities: Magellan: FIRE, IRTF: SpeX REFERENCESAckerman, A. S., & Marley, M. S. 2001, ApJ, 556, 872Allard, F., Hauschildt, P. H., Alexander, D. R., Tamanai, A., &Schweitzer, A. 2001, ApJ, 556, 357Baraffe, I., Chabrier, G., Barman, T. S., Allard, F., & Hauschildt,P. H. 2003, A&A, 402, 701 Barman, T. S., Macintosh, B., Konopacky, Q. M., & Marois, C.2011, ApJ, 733, 65Bean, J. L., Seifahrt, A., Hartman, H., Nilsson, H., Wiedemann,G., Reiners, A., Dreizler, S., & Henry, T. J. 2010, ApJ, 713, 410