Astrophysical The Journal, 658:557 Y 568, 2007 March 20
# 2007. The American Astronomical Society. All rights reserved. Printed in U.S.A.
A
DISCOVERY OF A HIGH PROPER MOTION L DWARF BINARY: 2MASS J152002244422419AB
Adam J. Burgasser,1,2 Dagny L. Looper,2,3 J. Davy Kirkpatrick,4 and Michael C. Liu3
Received 2006 September 6; accepted 2006 November 21
ABSTRACT We report the discovery of the wide L1.5+L4.5 binary 2MASS J152002244422419AB, identied during spectroscopic follow-up of high proper motion sources selected from the Two Micron All Sky Survey. This source was independently identied by Kendall et al. in the SuperCOSMOS Sky Survey. Resolved JHK photometry and lowresolution near-infrared spectroscopy demonstrate that this system is composed of two well-separated (1:174 00 0:016 00 ) L dwarfs. Component classications are derived using both spectral ratios and comparison to the nearinfrared spectra of previously classied eld L dwarfs. Physical association for the pair is deduced from the large common proper motion of the components ( 0:73 00 0:03 00 yr1) and their similar spectrophotometric distances (19 2 pc). The projected separation of the binary, 22 2 AU, is consistent with maximum separation/total systemmass trends for very low mass binaries. The 2MASS J15204422 system exhibits both large tangential (66 7 km s1) and radial velocities (70 18 km s1), and its motion in the local standard of rest suggests that it is an old member of the Galactic disk population. This system joins a growing list of well-separated (>0.500 ), very low mass binaries, and is an excellent target for resolved optical spectroscopy to constrain its age, as well as trace activity/rotation trends near the hydrogen-burning limit. Subject headingg: binaries: visual stars: individual (2MASS J152002244422419) s stars: low-mass, brown dwarfs Online material: color gures
1. INTRODUCTION Multiple stellar systems are important laboratories for a wide range of astrophysical phenomena, from the formation and evolution of planetary systems to the distribution of dark matter in the Galaxy. Stellar multiples drive novae outbursts in cataclysmic variable systems, are useful distance ladders for star clusters in the local Group, and provide one of the few direct means of measuring stellar mass. The production of multiples is inherent to the star formation process itself, and the aggregate properties of multiple systems in a given population can provide insight into the genesis of that population. Multiples are of particular importance in the study of very low mass (VLM; M 0:1 M) stars and brown dwarfs. Several of the rst brown dwarfs identied were found as companions to nearby stars ( Becklin & Zuckerman 1988; Nakajima et al. 1995; Rebolo et al. 1998), and >75 binaries composed entirely of VLM dwarfs are now known ( Burgasser et al. 2006b, and references therein).5 The salient properties of these systemstheir small separations, high mass ratios, and low frequencyhave been used as empirical tests of brown dwarf formation theories ( Reid et al. 2001b; Burgasser et al. 2003, 2006b; Close et al. 2003; Bouy et al. 2004; Siegler et al. 2005). Astrometric monitoring of resolved doubles has provided direct measures of mass for a few VLM binaries (Lane et al. 2001; Bouy et al. 2004; Brandner et al. 2004;
1 Massachusetts Institute of Technology, Kavli Institute for Astrophysics and Space Research, Cambridge, MA; [e-mail]. 2 Visiting Astronomer at the Infrared Telescope Facility, which is operated by the University of Hawaii under Cooperative Agreement NCC 5-538 with the National Aeronautics and Space Administration, Ofce of Space Science, Planetary Astronomy Program. 3 Institute for Astronomy, University of Hawaii, Honolulu, HI. 4 Infrared Processing and Analysis Center, California Institute of Technology, Pasadena, CA. 5 A current list is maintained by N. Siegler at http://paperclip.as.arizona.edu/ $ nsiegler/VLM _ binaries.
Zapatero Osorio et al. 2004), while the recently identied eclipsing binary 2MASS J0535218054608AB has enabled the rst radius measurement in the substellar regime (Stassun et al. 2006). Coupled with independent age determinations, these observations also constrain brown dwarf evolutionary theories (Zapatero Osorio et al. 2004; Stassun et al. 2006). Multiplicity provides explanations for other observed phenomena, such as peculiar features in individual brown dwarf spectra (Cruz et al. 2004; McGovern 2005; Burgasser et al. 2005b) and unusual photometric trends across the transition between L-type and T-type brown dwarfs ( Burgasser et al. 2006; Liu et al. 2006 ). Despite their fecundity, only a limited number of VLM binaries have resolved spectroscopic measurements.6 Such observations provide detailed information on the components of these systems, and can clarify empirical trends as a function of spectral type, temperature, or mass by eliminating the variables of age and composition (assuming coeval formation). One regime in which such conditions can be exploited is the transition between the M and L dwarf spectral classes (Kirkpatrick 2005). Occurring at an effective temperature TeA % 2300 K (Golimowski et al. 2004; Vrba et al. 2004), this transition is characterized by the formation of photospheric condensates that dominate the near-infrared spectra of L dwarfs (Tsuji et al. 1996; Jones & Tsuji 1997; Burrows & Sharp 1999; Allard et al. 2001; Ackerman & Marley 2001), a rapid decline in the frequency and strength of H emission, related to the presence of a hot chromosphere (Gizis et al. 2000; Mohanty et al. 2002; West et al. 2004), and an increase in mean rotational velocities, possibly related to a change in angular momentum loss mechanisms or age effects (Mohanty & Basri 2003).
6 Resolved spectroscopy of VLM binaries is reported in Goto et al. (2002), Potter et al. (2002), Bouy et al. (2004), Chauvin et al. (2004), Luhman (2004), McCaughrean et al. (2004), Billeres et al. (2005), Burgasser & McElwain (2006), Close et al. (2007), Jayawardhana & Ivanov (2006), Kendall et al. (2007), McElwain & Burgasser (2006), Martn et al. (2006), and Mohanty et al. (2006).
557
558
BURGASSER ET AL.
Vol. 658
Fig. 1.Field images of 2MASS J152002244422419 from SERC IN (top left), ESO R (top middle), and AAO R (top right) photographic plates, and from 2MASS (bottom). All images are scaled to the same resolution and oriented with north up and east to the left. Photographic plate images are 50 on a side. Inset boxes 20 00 ; 20 00 in size indicate the position of the source after correcting for its motion, and are expanded in the lower left corner of each image.
In addition, late M and early L spectral types encompass the transition between hydrogen-burning stars and brown dwarfs, depending on the age of the population (Chabrier et al. 2000; Burrows et al. 2001). Our group is currently conducting a near-infrared proper motion survey using multi-epoch data from the Two Micron All Sky Survey (2MASS; Skrutskie et al. 2006). The goal of this program is to identify late-type, nearby, and/or high-velocity sources that may have been missed by existing photographic plate proper motion surveys or near-infrared searches based solely on colorselection criteria. Details on the survey will be presented in a forthcoming publication (J. D. Kirkpatrick et al., in preparation). Here we report the identication of a well-resolved L dwarf binary system, 2MASS J152002244422419 (hereafter 2MASS J15204422). This source has been independently identied by Kendall et al. (2007) by combining 2MASS and SuperCOSMOS Sky Survey (SSS; Hambly et al. 2001a, 2001b, 2001c) catalog data. Our initial identication of this source is described in x 2. Its recognition as a binary in follow-up imaging observations and measurements of separation, position angle, and ux ratios are described in x 3. In x 4 we present near-infrared spectroscopic observations obtained with the SpeX spectrograph (Rayner et al. 2003) mounted on the 3 m NASA Infrared Telescope Facility (IRTF). These include resolved low-resolution spectra that identify the components as L dwarfs, and composite moderate-resolution data
that reveal detailed spectral features similar to previously studied L dwarfs. Analysis of our observations is provided in x 5, including spectral classication of the components, kinematics of the system, and additional binary parameters. We discuss 2MASS J15204422 in the context of other well-resolved binaries in x 6, and motivate future spectroscopic investigations to apply the binary Li test. Results are summarized in x 7. 2. IDENTIFICATION OF 2MASS J15204422 2MASS J15204422 was identied as a relatively bright (J 13:23 0:03) and red (J Ks 1:33 0:04) unresolved source, imaged twice by 2MASS on 1999 May 19 and 2001 February 10 (UT). The difference in the astrometry of this source between the two imaging epochs implies a proper motion 0:72 00 0:12 00 yr1 at a position angle 235 . As shown in Figure 1, counterparts of 2MASS J15204422 can be seen in SERC IN (epoch 1978 May 27 UT) and ESO R (epoch 1981 April 2 UT) photographic plates, offset by 16.600 and 15.200 , respectively, at the positions expected from the 2MASS proper motion. A marginal source is also present in a 1992 July 30 (UT) AAO R-band photographic plate. SSS measurements of 2MASS J15204422 give IN 17:04, RESO 19:41, and 0:74 00 0:17 00 yr1. To obtain a more rened measure of its proper motion, we combined the SSS and 2MASS astrometry of 2MASS J15204422
No. 1, 2007
TABLE 1 Astrometry for 2MASS J152002244422419AB Epoch (UT) 1978 1981 1999 2001 May 27.............. Apr 2................. May 19.............. Feb 10............... 15 15 15 15
2MASS J15204422AB
559
R.A.a 20 20 20 20 03.46 03.34 02.24 02.14
Decl.a 44 44 44 44 22 22 22 22 34.1 35.6 41.9 42.7
Source SERC IN (SSS) ESO R (SSS) 2MASS 2MASS
Note.Units of right ascension are hours, minutes, and seconds, and units of declination are degrees, arcminutes, and arcseconds. a Equinox J2000.0 coordinates.
spanning 22.7 yr, as listed in Table 1. Linear ts to the right ascension and declination over this period yield cos 0:634 00 0:013 00 yr1 and 0:367 00 0:012 00 yr1, for a combined motion of 0:73 00 0:03 00 yr1 at 239:9 1:0 . Note that these values do not take into consideration parallactic motion. The high proper motion of this optically faint red source made it a high-priority target for follow-up observations. 3. IMAGING OBSERVATIONS 3.1. Data Acquisition and Reduction 2MASS J15204422 was imaged again in the near-infrared on 2006 April 8 ( UT) using SpeX on IRTF. Conditions during this night were poor, with patchy cirrus and clouds. Seeing was 0.800 Y0.900 at the J band. Acquisition images with the instruments guiding camera immediately revealed two distinct sources at the position of 2MASS J15204422, separated by roughly 100 . After rechecking telescope focus, we obtained a series of 20 s exposures
of the pair in the J, H, and K lters,7 interspersed with images of the nearby, bright point source 2MASS J151959604421523 (J 13:20 0:03, J Ks 0:72 0:04) as a point-spread function (PSF) calibrator. Four dithered exposures were obtained for each source and lter, for a total of 80 s integration each. We also obtained twilight and bias exposures in all three lters on 2006 April 11 (UT) during the same observing campaign for pixel response calibration. The imaging data were cleaned using a pixel mask constructed from the bias and twilight images, then pair-wise subtracted and divided by a normalized at-eld frame constructed from the median-combined, bias-subtracted twilight observations (appropriate to the respective lter). The calibrated images were centered and co-added to produce a mosaic for each lter/pointing combination. Figure 2 displays 6 00 ; 6 00 subsections of these mosaics for the target and PSF star. Two sources are clearly resolved at the position of 2MASS J15204422, lying along a northwest-southeast axis. The northernmost source is markedly brighter in all three bands, particularly at J and H. 3.2. PSF Fitting Relative astrometry and photometry of the 2MASS J1520 4422 pair were determined using the same iterative PSF tting algorithm described in Burgasser & McElwain (2006). For each lter, we t centered 6 00 ; 6 00 subsections of each pair-wise subtracted image of 2MASS J15204422 to equivalent subsections for all four PSF images, a total of 16 ts per lter. The tting algorithm rst determined an initial estimate of the peak position
7 JHK lters on SpeX are based on the Mauna Kea Observatory near-infrared ( MKO-NIR) lter set (Simons & Tokunaga 2002; Tokunaga et al. 2002).
Fig. 2.SpeX images of 2MASS J152002244422419AB (top) and the PSF calibrator 2MASS J151959604421523 (bottom) at J (left), H (middle), and K (right). Images are 600 on a side. The image scale is indicated in the bottom left image, and the orientation of all six images (6:4 0:3 ) is indicated in the bottom right image.
560
BURGASSER ET AL.
TABLE 2 Properties of the 2MASS J152002244422419AB System Parameter Value 15h20m02.24s 44 220 41.900 0.7300 0.0300 yr1 239.9 1.0 L1.5 + L4.5 19 2 pc 1.17400 0.01600 22 2 AU 152.9 0.7 66 7 km s1 70 18 km s1 (50; 30; 20) km s1 19.41 mag 17.04 mag 13.23 0.03 mag 12.36 0.03 mag 11.89 0.03 mag 1.15 0.08 mag 0.97 0.06 mag 0.95 0.03 mag 0.14Y 0.16 M $400 yr Reference 1 1 1, 2, 3 1, 2, 3 3 3 3 3 3 3 3 3 2 2 1 1 1 3 3 3 3, 4 3, 4
Vol. 658
a ............................................... a ............................................... ................................................ ................................................. NIR SpT.................................... d est ............................................. ................................................ ................................................ Vtan............................................. Vrad............................................. (U ; V ; W ) .................................. RESOb ......................................... IN b ............................................. J b .............................................. H b ............................................. Ks b ............................................. J ............................................. H ............................................ K ............................................ Mtotc ........................................... Periodc .......................................
a b
Because relative uxes were measured using MKO-NIR lters, we rst derived the lter translation factors M M 2MASS M MKO for each component, where M 2MASS M MKO M ; B M ; A , as described in McElwain & Burgasser (2006). These factors were calculated by convolving lter response curves (lter transmission function ; optical response ; telluric absorption at an air mass of 1) for 2MASS (Cohen et al. 2003; see Cutri et al. 2003, x IV.4.a) and MKO-NIR lters (S. Leggett 2004, private communication) with the individual component spectra (x 4.1) and a spectrum of the A0 V star Vega ( Hayes 1985; Mountain et al.1985). These calculations yield J ; B J ; A 0:018, H; B H; A 0:001, and K; B K; A 0:001. The lter translation factors are therefore negligible, likely due to the similarity of the component spectra in the regions sampled by the JHK lters, and were disregarded. Component magnitudes are listed in Table 3. 3.3. Common Proper Motion The high proper motion of 2MASS J15204422 and the brightness of its components allow us to immediately deduce common proper motion for this pair. Both components are bright enough to be detected individually by 2MASS. Since the earliest 2MASS image of this eld was taken 6.89 yr prior to the SpeX observations, at least one of these sources would have moved a full 5.0500 from the SpeX position, easily resolvable by 2MASS. Since no secondary source is seen in these earlier images (Fig. 1), implying a limiting separation of $1.500 (Burgasser et al. 2005a), we conclude that the two sources are comoving to within $0.200 yr1 of motion and to within $15 Y20 in position angle. These limits are sufciently stringent that common proper motion can be condently claimed, and we refer to this pair hereafter as 2MASS J15204422AB. 4. SPECTROSCOPIC OBSERVATIONS 4.1. Prism Spectroscopy Resolved, low-resolution near-infrared spectral data for the 2MASS J15204422 components were obtained on 2006 April 8 (UT), also using SpeX. We used the 0.300 slit to better separate ux from each component, yielding 0.75Y2.5 m spectra with resolution k/k % 250 and dispersion across the chip of 20Y 30 8 pixel1. To obtain separate spectra for each component, we set the image rotator to 300 so that one source lay in the slit while the other source was positioned orthogonal to the slit axis and used for guiding (Fig. 3). This orientation was not aligned with the parallactic angle ($15 ), and as the observations were made at fairly large air mass (2.42Y2.60), differential refraction (DR) effects are expected (Filippenko 1982, see below). Six and eight exposures of 90 s each were obtained for the A and B components, respectively, both in an ABBA dither pattern. The A0 V star HD 133820 was observed immediately afterward, at a similar air mass (2.66) and with the slit aligned to the parallactic angle, for ux calibration.
Equinox J2000.0 coordinates at epoch 1999 May 19 from 2MASS. Photometry for combined (unresolved) system. c Assuming an age of 1Y10 Gyr. References. (1) 2MASS (Skrutskie et al. 2006); (2) SSS ( Hambly et al. 2001a, 2001b, 2001c); (3) This paper; (4) Burrows et al. 1997.
and uxes of the two components, then recursively matched model images (constructed from the PSF frames) to the data. The primary pixel coordinates, secondary pixel coordinates, primary ux, and secondary ux in the model image were adjusted iteratively in steps of 0.1 pixels and 0.01 fraction ux until residuals were minimized, of order 1%Y2% of the source peak ux. The mean of the ux ratios,8 separations (), and position angles () for each lter are listed in Table 2. Uncertainties include scatter in the PSF t measurements, and values take into account the pixel scale (0:120 00 0:002 00 pixel1) and rotator position (6:4 0:3 ; J. Rayner 2005, private communication) of SpeX during the observations. As in Burgasser & McElwain (2006), we found no systematic offsets in our tting results based on PSF simulations. To determine component magnitudes, we combined our relative ux measurements with composite photometry from 2MASS.
8 Note that the imaging observations were taken in nonphotometric conditions; however, we assume that the relative atmospheric transmission over the small separation of the 2MASS J15204422 pair was constant and that any differential atmospheric absorption has minimal effect on the relative ux within each individual lter.
TABLE 3 Properties of the 2MASS J152002244422419AB Components 2MASS Teffa ( K) 2200 1740 J (mag) 13.55 0.04 14.70 0.07 H (mag) 12.73 0.03 13.70 0.05 Ks (mag) 12.27 0.03 13.22 0.04 1 Gyr (M) 0.075 0.064 Estimated Massb 5 Gyr (M) 0.082 0.077 10 Gyr (M) 0.082 0.078
Component 2MASS J15204422A................ 2MASS J15204422B ................
a b
NIR SpT L1.5 L4.5
Teff estimate from Vrba et al. (2004). Based on the solar metallicity models of Burrows et al. (1997 ).
No. 1, 2007
2MASS J15204422AB
561
Fig. 3.Left: Stacked K-band images of 2MASS J15204422 during acquisition of the fainter component. The brighter component was used for guiding, while the secondary was offset slightly from the slit center to minimize contamination. Right: Model of the guider image used to compute differential refraction ( DR) and contamination effects. Both images are 7.200 on a side, and orientation on the sky is indicated by the arrow.
Internal at-eld and Ar arc lamps were then observed for pixel response and wavelength calibration. Data were reduced using the SpeXtool package, version 3.3 (Cushing et al. 2004), using standard settings. First, the images were corrected for linearity, pair-wise subtracted, and divided by the corresponding median-combined at-eld image. Spectra were optimally extracted using the default settings for aperture and background source regions, and wavelength calibration was determined from arc lamp and sky emission lines. Multiple spectral observations for each source were then median-combined after scaling the individual spectra to match the observation with the highest signal-to-noise ratio. Telluric and instrumental response corrections for the science data were determined using the method outlined in Vacca et al. (2003), with line shape kernels derived from the arc lines. Adjustments were made to the telluric spectra to compensate for differing H i line strengths in the observed A0 V spectrum and pseudovelocity shifts. Final calibration was made by multiplying the observed target spectrum by the telluric correction spectrum, which includes instrumental response correction through the ratio of the observed A0 V spectrum to a scaled, shifted, and deconvolved Kurucz9 model spectrum of Vega. Corrections for DR effects in the spectrum of the fainter component, in terms of both slit losses and contamination from the bright primary, were determined by simulating the light throughput for both sources as a function of wavelength. We rst created a model of stacked K-band guiding images (acquired at one dither position during the spectral observations) using two symmetric Gaussian surfaces and a box-car slit prole. The widths of the Gaussian proles were determined by one-dimensional Gaussian ts of image slices through the brighter component parallel to the slit axis. Note that these values are slightly larger ($100 ) than the measured seeing, due to telescope guiding jitter. We adjusted the uxes and pixel offsets (perpendicular to the slit) of the two components in our model using an iterative algorithm similar to that used for the PSF tting, minimizing residuals with respect to the K-band guider image. Figure 3 displays the best-t model; note that the secondary was actually centered slightly off the slit. We then calculated DR pixel shifts as a function of wavelength using the formalism of Roe (2002; see also Schubert & Walterscheid 2000), assuming standard atmospheric pressure and temperature, a relative humidity of 50%, and a true zenith distance of 55.5 . The calculated shifts move both components closer to the slit center at wavelengths shorter than 2.2 m, by as much as 1.5 pixels at 1.2 m. Hence, slit losses are greatest at the longest wavelengths,
9
Fig. 4.SpeX prism spectrum of 2MASS J15204422B before (dashdotted line) and after (solid line) correction of differential refraction ( DR) and contamination effects. Both spectra are normalized at 1.28 m. [See the electronic edition of the Journal for a color version of this gure.]
See http://kurucz.harvard.edu/stars.html.
while contamination from the primary is greatest at the shortest wavelengths. Both effects were calculated using our model images by shifting the centers of the Gaussian proles in accordance with the DR shifts ( keeping the seeing and relative ux ratios constant) and calculating the total light throughput for both components in the slit. We found slit losses to be roughly 20% at K relative to J, while light contamination ranged from 3% at K to >10% for k < 1:2 m. The spectrum of 2MASS J15204422B was corrected for slit losses using that components model-derived relative throughput values. Contamination from the primary was removed by subtracting a spectrum of 2MASS J15204422A scaled by the wavelength-dependent contribution of its light through the slit relative to the secondary from the DR model. Figure 4 shows the spectrum of 2MASS J15204422B before and after these effects have been accounted for. The corrected spectrum is signicantly redder, as both slit losses and light contamination suppress K-band light from this component. Similar simulations were used to calculate DR slit losses for the spectrum of the primary, assuming it to be centered on the slit at the K band. These were smaller (<10% slit loss for k > 1 m) due to the large size of the seeing disk relative to the slit width. Light contamination from the secondary was found to be negligible. The reduced prism spectra (including DR and contamination corrections) of the two 2MASS J15204422 components are shown in Figure 5. These spectra are qualitatively similar to JH spectra presented in Kendall et al. (2007). Both components have spectral energy distributions consistent with early-type L dwarfs, including red optical and near-infrared spectral slopes, deep H2O absorption bands at 1.4 and 1.9 m, strong CO absorption at 2.3 m, FeH and CrH bands at 0.86, 0.99, 1.2, and 1.6 m, a notable absence of TiO and VO features shortward of 1 m, and unresolved Na i and K i atomic lines at the J band (Reid et al. 2001a; Testi et al. 2001; McLean et al. 2003; Nakajima et al. 2004). The 0.99 m FeH band in 2MASS J15204422B is unusually strong, although this may be due to uncorrected DR effects. The stronger
562
BURGASSER ET AL.
Vol. 658
Fig. 6.Cross-dispersed spectrum of the composite 2MASS J15204422 system. Data are normalized at 1.28 m. Major molecular and atomic absorption features are indicated, as well as regions of strong telluric absorption (). The inset shows a close-up view of the 1.15Y1.35 m region, hosting several K i and Fe i lines and FeH and H2O molecular bands.
Fig. 5.Near-infrared SpeX spectra of 2MASS J15204422A (top) and B (bottom). Data are normalized at 1.28 m, with 2MASS J15204422A offset by a constant for clarity (dotted lines). Major molecular ( FeH, CrH, H2O, and CO) and atomic ( Na i and K i) absorption features are labeled, as well as regions of strong telluric absorption ().
H2O and FeH molecular bands of 2MASS J15204422B, and its redder near-infrared color, indicate that it is of later type than 2MASS J15204422A. 4.2. Cross-dispersed Spectroscopy Higher resolution cross-dispersed ( XD) spectra for the composite system (i.e., both sources simultaneously in the slit) were obtained on 2006 April 11 (UT). Conditions on this night were somewhat improved, with light cirrus and moderate seeing (0.800 at the J band). The 0.500 slit was used for a spectral resolution k/ k % 1200 and dispersion across the chip of 2.7Y5.3 8 pixel1. The slit was aligned to the parallactic angle, and 8 exposures (4 of 250 s, 4 of 300 s) dithered in an ABBA pattern were obtained for a total integration time of 2200 s. HD 133820 was again observed immediately after 2MASS J15204422 and at a similar air mass (2.75), followed by calibration lamps. As with the prism data, XD data were reduced using the SpeXtool package, following similar procedures but using a line shape kernel derived from the 1.005 m H i Pa line in the A0 V calibrator spectra. After telluric and ux calibration, the ve orders spanning 0.82Y 2.43 m were scaled and combined using the prism spectrum of 2MASS J15204422A as a relative ux template. The reduced XD spectrum for 2MASS J15204422 is shown in Figure 6. While these data have lower overall signal-to-noise ratio than the prism data (particularly shortward of 0.98 m and in the telluric bands), many of the small wiggles observed in this spectrum are real, arising largely from H2O and FeH transitions and numerous atomic metal lines. The strongest atomic features are the Na i and K i doublet lines at the J band, which are shown in detail in the inset of Figure 6. In addition, there are several lines from Fe i and FeH and broader FeH bands in this spectral region. CO band heads at 2.299, 2.328, 2.357, and 2.388 m are clearly discerned, but the 2.206/2.209 m Na i doublet is notably absent. The absence of this feature provides further evidence that 2MASS
J15204422 is composed of L dwarfs (McLean et al. 2003; Cushing et al. 2005). Similarly, the absence of the 1.314 m Al i line is consistent with an L spectral type, while the presence of the 1.189 m Fe i line suggests a composite spectral type of L3 or earlier (McLean et al. 2003). Equivalent widths (EW) of the 1.169, 1.178, and 1.253 m K i lines and 1.189 Fe i line were measured using Gaussian ts to the line proles and linear ts to the nearby pseudocontinuum (the 1.244 m K i was not measured, as its blue wing is blended with an FeH band). Values of 6 2, 8:3 1:8, 8:0 1:6, and 0:5 0:2 8, respectively, were derived and are comparable to other early-type L dwarfs (Cushing et al. 2005). 5. ANALYSIS 5.1. Spectral Classication and Estimated Distance The resolved prism spectra of the 2MASS J15204422 components allow us to derive individual spectral types. However, while a robust classication scheme for L dwarfs exists at optical wavelengths (Kirkpatrick et al. 1999, 2000), there is as yet no well-dened scheme in the near-infrared, i.e., following the tenets of the MK Process (Morgan et al. 1943; Corbally et al. 1994). We therefore determined subtypes for these sources using a variety of spectral indices from the literature that are applicable to lowresolution near-infrared spectral data. Reid et al. (2001a) have dened several indices measuring the strengths of H2O and CO bands, as well as color ratios, for late-type M and L dwarf spectra. They provide linear spectral type calibrations anchored to optical classications for their H2O-A and H2O-B indices (sampling the 1.1 and 1.4 m H2O bands) and the K1 index of Tokunaga & Kobayashi (1999) over spectral types M8 to L6. Testi et al. (2001) dened six indices, sampling H2O bands and color ratios for spectral data of comparable resolution to our SpeX prism observations, and provided linear calibrations for these indices over types L0 to L8. Burgasser et al. (2002) also dened H2O band indices and color ratios, as well as CH4 band indices, in their near-infrared classication scheme for T dwarfs; three of the H2O indices show good correlation with M5YL7 optical spectral type, and one CH4 index correlates with L3YT3 spectral types. Finally, Geballe et al. (2002) utilized several near-infrared indices to classify late-type M, L, and T dwarfs, two of which ( H2O 1.5 m and CH 4 2.2 m) are
No. 1, 2007
2MASS J15204422AB
TABLE 4 Near-infrared Spectral Classification Indices 2MASS J15204422A Index Value SpT 2MASS J15204422B Value SpT 2MASS J15204422AB ( XD) Value SpT
563
Reid et al. (2001a) H2O-A .................................... H2O-B .................................... K1a ......................................... 0.686 0.749 0.178 L1.5 L2 L1 0.597 0.639 0.297 L4 L5 L3.5 0.656 0.718 0.219 L2.5 L3 L2
Testi et al. (2001) sH2O J ..................................... sH2O H1 ................................... sH2O H2 ................................... sH2O K .................................... 0.165 0.327 0.443 0.197 L2.5 L2 L2 L0.5 0.259 0.510 0.538 0.332 L4 L4 L4 L4 0.229 0.343 0.517 0.255 L3.5 L2 L4 L2
Burgasser et al. (2002) H2O-A .................................... H2O-B .................................... H2O-C .................................... CH 4-C .................................... 0.878 0.857 0.868 0.991 L1.5 L1.5 L1 ... 0.806 0.790 0.782 0.954 L5 L5 L6.5 L4.5 0.835 0.836 0.861 0.990 L3.5 L2.5 L1.5 ...
Geballe et al. (2002) H2O 1.5 m ........................... CH 4 2.2 m ........................... Mean NIR SpT ......................
a b
1.333 0.973 L1.5 0.6
L1.5 . . .b
1.592 1.013
L5.5 L5
1.407 0.975 L2.5 0.8
L2 . . .b
L4.5 0.8
Index dened in Tokunaga & Kobayashi (1999). Index /SpT relation not dened for SpT < L3 ( Burgasser et al. 2002; Geballe et 2002 al. ).
applicable in the L dwarf regime (for spectral types L0YL9 and L3YL9, respectively). Unlike the other near-infrared schemes, Geballe et al. (2002) tie spectral types to their indices with predetermined numerical ranges set by measurements for a large sample of VLM dwarf spectra. Table 4 lists the values and associated spectral types of these indices for each component spectrum prism of 2MASS J1520 4422, as well as the composite XD spectrum. Note that we have not included color indices from Testi et al. (2001) and Burgasser et al. (2002), due to possible residual DR effects. There is fairly consistent agreement in the derived spectral types between these schemes, likely due to the similarity in features used to dene the indices (primarily H2O bands). Mean subtypes for all indices are L1.5 and L4.5 for 2MASS J15204422A and B, with scatter of 0.6 and 0.8 subtypes, respectively. This is consistent with the types derived by Kendall et al. (2007 ), L2 and L4, within the stated uncertainties. The XD spectrum is classied L2.5 0.8, as expected for contributions from both components to the combined light spectrum. To conrm these classications, we compared the prism spectra to an array of equivalent data for previously classied eld L dwarfs with similar optical and/or near-infrared subtypes, as shown in Figure 7. The spectrum of 2MASS J15204422A is nearly identical to that of 2MASS J205754090252302, classied L1.5 from both optical (Cruz et al. 2003) and near-infrared ( Kendall et al. 2004) data, and appears to be intermediate between those of optically classied L1 (2MASS J14392836+1929149; Kirkpatrick et al. 1999 spectral standard) and L2 (SSSPM J08291309; Scholz & Meusinger 2002) comparison sources. Similarly, the spectrum of 2MASS J15204422B shows remarkable agreement with that of the optically classied L4 2MASS J11040127+1959217 (Cruz et al. 2003), and less agreement with those of optically classied
L3 (SDSS J202820.32+005226.5; Hawley et al. 2002) and L5 dwarfs (GJ 1001BC; Goldman et al. 1999). These comparisons would appear to verify the index-based classications from Table 4, and indicate that the DR and light contamination corrections applied to the spectrum of 2MASS J15204422B were accurate. The agreement of the near-infrared spectra of 2MASS J1520 4422A and B with those of optically classied sources suggests that the derived subtypes could be adopted as their optical classications. However, a number of studies (Stephens 2003; Knapp et al. 2004; Kirkpatrick 2005) have found that optical and nearinfrared spectral types cannot be assumed to be identical for midand late-type L dwarfs, due to the complexity of their atmospheres. Figure 8 illustrates that this is an issue even among early-type L dwarfs, comparing SpeX prism data for three optically classied L2 dwarfsSSSPM J08291309, Kelu 1 (Ruiz et al. 1997), and SIPS J09212104 (Deacon et al. 2005)to that of 2MASS J15204422A. Despite their equivalent optical classications, these three sources have very different near-infrared spectra. Both SSSPM J08291309 and Kelu 1 appear to have redder 1Y2.5 m spectra than 2MASS J15204422A, while SIPS J09212104 is bluer. SSSPM J08291309 has weaker 1.4 and 1.7 m H2O absorption, while these same bands are markedly deeper in the spectrum of SIPS J09212104. The spectral indices used above yield mean near-infrared subtypes of L1, L2.5, and L4 for SSSPM J08291309, Kelu 1, and SIPS J09212104, respectively (see also Knapp et al. [2004] and Lodieu et al. [2005] regarding the near-infrared spectrum of Kelu 1). Such classication disagreements have been attributed to a variety of causes, including variations in photospheric dust content (Stephens 2001, 2003), unresolved multiplicity ( Burgasser et al. 2005b; Liu & Leggett 2005), and gravity effects ( McGovern et al. 2004; Kirkpatrick
564
BURGASSER ET AL.
Vol. 658
Fig. 7.Comparison of the near-infrared spectra of 2MASS J15204422A and B (solid lines in left and right panels, respectively) to equivalent data (dash-dotted lines) for the spectral comparison stars. In the left panel we compare 2MASS J15204422A to 2MASS J14392836+1929149 ( L1 optical classication), 2MASS J20575409 0252302 ( L1.5 optical and near-infrared classication), and SSSPM J08291309 ( L2 optical classication). In the right panel we compare 2MASS J15204422B to SDSS J202820.32+005226.5 ( L3 optical classication), 2MASS J11040127+1959217 ( L4 optical classication), and GJ 1001BC ( L5 optical classication, L4.5 near-infrared classication). All spectra have been normalized at 1.28 m and offset by constants (dotted lines). [See the electronic edition of the Journal for a color version of this gure.]
et al. 2006). These complexities emphasize the need for a robust, independent, and multidimensional near-infrared classication scheme for L dwarfs. Assuming that the derived near-infrared spectral types for 2MASS J15204422A and B are at least suitable proxies for their optical classications, spectrophotometric distance estimates for each component were computed using measured photometry and absolute 2MASS magnitude/spectral type relations from Dahn et al. (2002), Cruz et al. (2003), Tinney et al. (2003), and Vrba et al. (2004). These yield mean distances of 19:1 0:8 and 19:1 0:7 pc for 2MASS J15204422A and B, respectively, assuming no uncertainty in the spectral types. The agreement in the estimated distances for both components and their common proper motion are consistent with these two sources being gravitationally bound. We adopt a mean distance of dest 19 2 pc, which includes a 0.5 subtype uncertainty in the classications. 5.2. Radial Velocity and Kinematics The large proper motion of 2MASS J15204422 and its estimated distance imply a large tangential motion, Vtan 4:74dest 66 7 km s1. This is on the high end of the L dwarf Vtan distribution of Vrba et al. (2004), in which only 3/33 98 % of the 3 sources examined have Vtan > 60 km s1. Similarly, only 136 % 3 of late-type M and L dwarfs in the sample of Gizis et al. (2000) have Vtan > 60 km s1. These studies suggest that 2MASS J15204422 may be an unusually high velocity system. To determine its three-dimensional space motion, we used the XD spectrum to measure a systemic radial velocity (Vrad ). Despite the coarse velocity resolution of these data (V % 250 km s1), a relatively accurate determination of Vrad can be obtained through cross-correlation techniques. We compared the spectrum of
2MASS J15204422 to equivalent SpeX XD data from Cushing et al. (2005) for four early/mid-type L dwarfs with measured radial velocities (Bailer-Jones 2004): 2MASS J11463449+ 2230527 ( L3; 21 2 km s1), 2MASS J14392836+1929149 (L1; 26:3 0:5 km s1), 2MASS J150747691627386 (L5; 39:3 1:5 km s1), and 2MASS J222443810158521 (L4.5; 37 3 km s1). The wavelength scale for the 2MASS J1520 4422 data was shifted by velocities ranging over 300 to 300 km s1 in steps of 1 km s1. Then, for each shifted spectrum, the cross-correlation10 was computed against each of the comparison spectra in three wave bands with strong features: the 1.16Y1.26 m region, hosting numerous K i, Fe i, and FeH absorptions (see inset of Fig. 6), the 1.32Y1.34 m region, spanning the 1.33 m H2O band head, and the 2.290 Y2.295 m region, spanning the 2.292 m CO band head. After correcting for the radial velocities of the comparison sources, we derived a mean Vrad 60 12 km s1 for 2MASS J15204422, where the uncertainty is the scatter of the 12 cross-correlation measurements. As a secondary check, we measured the central wavelengths of the well-resolved 1.169, 1.177, 1.244, and 1.253 m K i lines, using Gaussian ts to the line cores and vacuum wavelengths as listed in the Kurucz Atomic Line Database11 (Kurucz & Bell 1995). These
Rk We computed C(V ) / k12 fkT (V ) fkT ( fkC fkC ) dk, where fkT (V ) is the spectrum of 2MASS J15204422 shifted to velocity V, fkC is the spectrum of the comparison source, and fk is the mean ux over the spectral band spanning k1 to k2 . The radial velocity of 2MASS J15204422 was determined as Vrad Vmax VC , where Vmax is the velocity that maximizes C(V ) and VC is the known radial velocity of the comparison source. 11 Obtained through the online database search form created by C. Heise and maintained by P. Smith; see http://cfa-www.harvard.edu /amdata /ampdata / kurucz23/sekur.html.
10
No. 1, 2007
2MASS J15204422AB
565
Fig. 8.Comparison of the near-infrared spectra of 2MASS J15204422A (solid lines) to equivalent data (dash-dotted lines) for three optically classied L2 dwarfs: SSSPM J08291309, Kelu 1, and SIPS J09212104. All spectra have been normalized at 1.28 m and offset by constants (dotted lines). [See the electronic edition of the Journal for a color version of this gure.]
burning minimum mass, M % 0:075 M (Chabrier et al. 2000; Burrows et al. 2001). However, because brown dwarfs cool over time, their Teff and spectral types are a function of both mass and age, which are difcult to disentangle for individual eld objects. In Table 3, we list estimated masses for 2MASS J15204422A and B for ages of 1, 5, and 10 Gyr, assuming Teff of 2200 and 1740 K, respectively (appropriate for L1.5 and L4.5 dwarfs; Vrba et al. 2004), and the evolutionary models of Burrows et al. (1997). The kinematics of this system suggest an age older than 1 Gyr, implying masses of k0.075 and k0.064 M for the two components, respectively. Hence, it is likely that 2MASS J15204422A is a low-mass star, while 2MASS J15204422B is either a lowmass star or a massive brown dwarf. Our spectrophotometric distance estimate of 2MASS J1520 4422 implies a projected separation of 22 2 AU, which is relatively wide for a VLM binary. Indeed, Burgasser et al. (2006b) have found that 93% of known VLM binaries have < 20 AU. However, wide VLM binaries such as 2MASS J15204422 do not violate the maximum separation/total system-mass trend of Burgasser et al. (2003), max 1400M 2 AU, which appears to tot be appropriate for nearly all known binaries with Mtot P 0:3 M. These are to be distinguished from a few ultrawide (separations >200 AU ) and/or weakly-bound (Vesc < 3:8 km s1; Close et al. 2003) VLM binaries now known (Chauvin et al. 2004, 2005; Luhman 2004; Billeres et al. 2005; Close et al. 2007; Jayawardhana & Ivanov 2006), whose origin and stability remain controversial ( Mugrauer & Neuhauser 2005) and a challenge for VLM dwarf formation models (Reipurth & Clarke 2001; Bate et al. 2003). Nevertheless, 2MASS J15204422 is sufciently wide that its estimated orbital period ($400 yr) is far too long to be useful for dynamical mass measurements in the near future. 6. DISCUSSION 2MASS J15204422 joins a growing list of VLM binaries that are sufciently well separated to allow resolved photometric and spectroscopic studies from moderate-sized ground-based telescopes (such as IRTF) without the need for adaptive optics corrections. Table 5 lists the 16 known VLM binaries with angular separations greater than 0.500 that fall into this category; all are resolvable from the ground at optical and near-infrared wavelengths during conditions of excellent seeing (indeed, most have been discovered and/or observed entirely with ground-based facilities). These systems are also among the most interesting VLM sources currently under investigation. Epsilon Indi Bab (Scholz et al. 2003; McCaughrean et al. 2004) is composed of the two brown dwarfs nearest to the Sun that are currently known, followed closely by the secondary of SCR 1845-6537AB (Hambly et al. 2004; Biller et al. 2006). The secondary of the 2MASS J12073347 3932540AB system, a member of the nearby $8 Myr TW Hydrae association (Gizis 2002), has an estimated mass in the planetarymass regime ($5 Jupiter masses; Chauvin et al. 2004), and both components may harbor circumstellar accretion disks (Mohanty et al. 2003, 2006; Gizis et al. 2005). Oph 11AB (2MASS J162225212405139; Allers et al. 2006) is an even younger and wider ($240 AU ) VLM binary whose components have masses nearly in the planetary-mass regime (Close et al. 2007; Jayawardhana & Ivanov 2006; Allers et al. 2007). 2MASS J233101610406193AB, Indi Bab, and Gliese 337CD are all members of higher order multiples, widely separated from more massive primaries (Gizis et al. 2000; Scholz et al. 2003; Burgasser et al. 2005a). DENIS J020529.0115925AB may itself be a triplet of brown dwarfs (Bouy et al. 2005).
four lines yield a mean Vrad 79 24 km s1, where the uncertainty includes systematic errors as determined from similar measurements for the four L dwarf comparison spectra used above. These two measures are consistent, and we adopt a weighted average of Vrad 70 18 km s1. Note that corrections for Earths motion in the solar frame of reference have not been included in this value, as these corrections are signicantly smaller than our estimated uncertainty. The total space velocity of 2MASS J15204422 is quite high, nearly 100 km s1 relative to the Sun. Combining position, proper motion, and radial velocity measurements with our estimates for the distance of 2MASS J15204422 yields local standard-of-rest (LSR) velocity components of (U ; V ; W ) % (50; 30; 20) km s1, where we have assumed (U ; V ; W ) (10; 5; 7) km s1 (Dehnen & Binney 1998). These values lie just outside of the 1 velocity dispersion sphere of local disk M dwarfs [(U ; V ; W ) % (40; 28; 19) km s1, centered at (13; 23; 7) km s1; Hawley et al. 1996], suggesting old disk kinematics, but not necessarily membership in the thick disk or halo populations. The absence of any distinct low-metallicity features in the near-infrared spectra of the 2MASS J15204422 components that characterize L subdwarfs (Burgasser 2004) also indicates that this system is likely an old member of the Galactic disk population. 5.3. Additional Properties of the 2MASS J15204422 Binary As early/mid-type L dwarfs, the components of 2MASS J1520 4422 are likely to have masses close to or below the hydrogen-
566
BURGASSER ET AL.
TABLE 5 VLM Binaries Wider than 0.500 (arcsec)
bc
Vol. 658
Name DENIS J020529.0115925AB ......................... 2MASS J042918423123568AB ....................... Gliese 337CD....................................................... 2MASS J233101610406193AB ....................... 2MASS J09153413+0422045AB ........................ Indi Bab ............................................................ 2MASS J120733473932540AB ....................... 2MASS J170723430558249AB ....................... DENIS J220002.0303832.9AB ........................ 2MASS J152002244422419 AB...................... SCR 1845-6537AB.............................................. 2MASS J110119267732383ABc ...................... 2MASS J162336092402209ABe ...................... 2MASS J162225212405139ABf ...................... DENIS J055146.0443412.2AB ........................
a b
d ( pc) 12 $11 21 $26 $15 3.6 $53 $15 $35 $19 3.9 $170 $125 $125 $100
Ra (mag) ... 16.2 ... 18.9 ... 20.8 19.3 18.9 19.2 19.4 16.3 19.4 17.4 19.4 ...
Ja (mag) 14.59 10.89 15.51 12.94 14.55 12.29 13.00 12.05 13.44 13.23 9.54 13.10 11.53 14.47 15.79
J (mag) 0.0 1.2 0.3 2.8 0.1 0.9 7.0 1.3 0.3 1.2 3.7 0.9 0.7 0.8 0.6
SpT L5 + L7 M7.5 + [ L1]d L8 + [ T:]d M8 + [ L7]d L7 + [ L7]d T1 + T6 M8.5 + L: M9 + L3 M9 + L0 L1.5 + L4.5 M8.5 + [ T5.5]d M7 + M8 [ M5:]d + [ M5:]d M9 + M9.5 M8.5 + L0
References 1, 2, 3, 4 5, 6, 7 8, 9 4, 10, 11, 12 7 13, 14 15, 16, 17, 18, 19 7, 15, 20 21, 22 23, 24 25, 26, 27, 28 29 30, 31 30, 31, 32, 33 34
0.51 0.53 0.53 0.57 0.73 0.73 0.78 1.01 1.09 1.17 1.17 1.44 1.70 1.94 2.20
R and J photometry for combined system from SSS and 2MASS, respectively; see also associated references. Possible triple system ( Bouy et al. 2005 ). c Resolved optical spectroscopy has been obtained for this system. d Spectral type based on resolved photometry; resolved spectroscopy has not yet been reported for this source. e aka Oph 16 (Allers et al. 2006 ). f aka Oph 11 (Allers et al. 2006 ). References. (1) Delfosse et al. 1997; (2) Koerner et al. 1999; (3) Leggett et al. 2001; (4) Bouy et al. 2003; (5) Cruz et al. 2003; (6) Siegler et al. 2005; (7) Reid et al. 2006; (8) Wilson et al. 2001; (9) Burgasser et al. 2005a; (10) Gizis et al. 2000; (11) Gizis et al. 2003; (12) Close et al. 2003; (13) Scholz et al. 2003; (14) McCaughrean et al. 2004; (15) Gizis 2002; (16) Chauvin et al. 2004; (17) Chauvin et al. 2005; (18) Mamajek 2005; (19) Mohanty et al. 2006; (20) McElwain & Burgasser 2006; (21) Kendall et al. 2004; (22) Burgasser & McElwain 2006; (23) This paper; (24) Kendall et al. 2007; (25) Hambly et al. 2004; (26) Henry et al. 2004; (27) Biller et al. 2006; (28) Henry et al. 2006; (29) Luhman 2004; (30) Allers et al. 2006; (31) Close et al. 2007; (32) Jayawardhana & Ivanov 2006; (33) Allers et al. 2007; (34) Billeres et al. 2005.
While resolved near-infrared spectroscopy exists for over half of the systems listed in Table 5, only three12 (DENIS J020529.0 115925AB, Martn et al. 2006; 2MASS J110119267732383AB, Luhman 2004; and Oph 11AB, Jayawardhana & Ivanov 2006) have resolved optical spectroscopy reported to date. Such observations are of particular importance for M dwarf/L dwarf binaries due to the presence of the 6563 8 H line, as discussed above, andthe 67088 Li i line. The latter is a powerful indicator of mass and age for M dwarf /L dwarfeld binaries, as recently pointed out by Liu & Leggett (2005). This species is depleted in the atmospheres of brown dwarfs and low-mass stars more massive than $0.065 M, due to fusion reactions in their cores (Rebolo et al. 1992). Hence, the detection of this line in the spectrum of a brown dwarf sets an upper limit for its mass and a corresponding upper limit on its age for a given Teff and assumed evolutionary model. For a binary system composed of two coeval brown dwarfs, Li i absorption may be present in one, both, or neither component, and any of these three cases can set different constraints on the age of the system. In the case of 2MASS J15204422AB, if the 6708 8 Li i line is present in the spectra of both components, then the evolutionary models of Burrows et al. (1997) predict a system age of P600 Myr. If Li i is absent in both component spectra, then the system is k1 Gyr old, consistent with its kinematics. The most interesting case is if Li i is present only in the spectrum of the secondary, as this would provide a tight constraint on the age of the system ($0.6Y1 Gyr). While this binary Li test has been indirectly used for sources with composite optical spectroscopy ( Burgasser et al. 2005b; Liu & Leggett 2005; McElwain & Burgasser 2006), the wide separation of 2MASS J15204422
12 Martn et al. (2006) report resolved optical spectroscopy for 9 VLM bina ries based on Hubble Space Telescope observations, 8 of which have separations less than 0.500 . Note that these data had insufcient resolution and signal-to-noise ratio to measure Li i EW 5 8.
makes it an excellen...