Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/~golim/papers/liu2006.pdf
Äàòà èçìåíåíèÿ: Thu Sep 14 00:27:48 2006
Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 07:12:56 2012
Êîäèðîâêà:

Ïîèñêîâûå ñëîâà: ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï
The Astrophysical Journal, 647:1393Y1404, 2006 August 20
# 2006. The American Astronomical Society. All rights reserved. Printed in U.S.A.

SDSS J1534+1615AB: A NOVEL T DWARF BINARY FOUND WITH KECK LASER GUIDE STAR ADAPTIVE OPTICS AND THE POTENTIAL ROLE OF BINARITY IN THE L/T TRANSITION1
Michael C. Liu,2,3 S. K. Leggett,4 David A. Golimowski,5 Kuenley Chiu,5 Xiaohui Fan, T. R. Geballe,7 Donald P. Schneider,8 and J. Brinkmann9
Received 2005 November 10; accepted 2006 April 28
6

ABSTRACT We have resolved the newly discovered T dwarf SDSS J153417.05+161546.1 into a 0B11 binary using the Keck sodium laser guide star adaptive optics system. With an integrated-light near-IR spectral type of T3:5 ô 0:5, this binary provides a new benchmark for studying the distinctive J-band brightening previously noted among early and mid-T dwarfs, using two brown dwarfs with different spectral types but having a common metallicity and age and very similar surface gravities. We estimate spectral types of T1:5 ô 0:5 and T5:5 ô 0:5 for the two components based on their near-IR colors, consistent with modeling the integrated-light spectrum as the blend of two components. The observed near-IR flux ratios of SDSS J1534+1615 are unique compared to all previously known substellar binaries: the component that is fainter at H and K 0 is brighter at J. This inversion of the near-IR fluxes is a manifestation of the J-band brightening within this individual binary system. Therefore, SDSS J1534+1615 demonstrates that the brightening can be intrinsic to ultracool photospheres (e.g., arising from cloud disruption and /or rapid increase in cloud sedimentation) and does not necessarily result from physical variations among the observed ensemble of T dwarfs (e.g., a range in masses, ages, and /or metallicities). We suggest that the apparently large amplitude of the J-band brightening may be due to a high incidence of unresolved binaries and that the true amplitude of the brightening phenomenon could be more modest. This scenario would imply that truly single objects in these spectral subclasses are relatively rare, in agreement with the small effective temperature range inferred for the L/T transition. Subject headingg binaries: close -- binaries: general -- infrared: stars -- stars: atmospheres -- s: stars: low-mass, brown dwarfs -- techniques: high angular resolution

1. INTRODUCTION For many decades, color-magnitude diagrams (CMDs) have provided fundamental insights into the physics of stellar evolution and stellar atmospheres. In an analogous fashion, the CMDs of brown dwarfs are beginning to shed light on the physical properties of these low-mass objects, as an increasing number of distances have been determined from trigonometric parallaxes for field brown dwarfs ( Dahn et al. 2002; Tinney et al. 2003; Thorstensen & Kirkpatrick 2003; Vrba et al. 2004) and from brown dwarf companions around stars with known distances (e.g., Becklin & Zuckerman 1988; Nakajima et al. 1995; Rebolo et al. 1998; Burgasser et al. 2000; Kirkpatrick et al. 2001; Liu et al. 2002; Potter et al. 2002; Freed et al. 2003; Metchev & Hillenbrand 2004). However, interpretation of brown dwarf CMDs is considerably less developed than for the case of stellar CMDs. Brown
1 The data presented herein were obtained at the W. M. Keck Observatory, which is operated as a scientific partnership among the California Institute of Technology, the University of California, and the National Aeronautics and Space Administration. The Observatory was made possible by the generous financial support of the W. M. Keck Foundation. 2 Institute for Astronomy, University of Hawaii, 2680 Woodlawn Drive, Honolulu, HI 96822; mliu@ifa.hawaii.edu. 3 Alfred P. Sloan Research Fellow. 4 United Kingdom Infrared Telescope, Joint Astronomy Centre, 660 North A`ohoku Place, Hilo, HI 96720. 5 Department of Physics and Astronomy, Johns Hopkins University, 3400 North Charles Street, Baltimore, MD 21218-2686. 6 Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ 85721. 7 Gemini Observatory, 670 North A`ohoku Place, Hilo, HI 96720. 8 Department of Astronomy and Astrophysics, Pennsylvania State University, University Park, PA 16802. 9 Apache Point Observatory, 2001 Apache Point Road, P.O. Box 59, Sunspot, NM 88349.

dwarfs continually cool as they age, unlike stars, which occupy a fixed position on the CMD during the main-sequence phase. Therefore, there are degeneracies in determining the masses and ages of brown dwarfs from the CMD. Substellar photospheres have a rich appearance due to the presence of dust and molecules (e.g., metal hydrides, water vapor, and methane), and spectroscopic methods to determine the associated physical properties, such as effective temperature and surface gravity, are still under development (e.g., Kirkpatrick 2005). The optical and infrared CMDs of brown dwarfs show considerable scatter, which could be due to differences in metallicity, surface gravity (i.e., mass), rotation, dust properties, and/or unrecognized binarity among the observational ensemble (e.g., Leggett et al. 2002; Knapp et al. 2004). Study of ultracool binary systems is a promising means to address these issues, since binaries provide systems with common ages and metallicities. About two dozen L dwarf binaries are now known from high-resolution imaging of over 100 L dwarfs, mostly by the Hubble Space Telescope (HST; Koerner et al. 1999; Martin et al. 1999; Reid et al. 2001; Gizis et al. 2003; Bouy et al. ´ 2003; Golimowski et al. 2004b). In contrast, only five ultracool binaries that include at least one T dwarf have been identified: 2MASS J1225þ2739AB ( T6+T8; Burgasser et al. 2003b), 2MASS J1534þ2952AB ( T5.5+T5.5; Burgasser et al. 2003b), Ind Bab ( T1+T6; McCaughrean et al. 2004), SDSS J0423þ 0414AB ( L6+T2; Burgasser et al. 2005b), and 2MASS J2252þ 1730AB ( L6+T2; Reid et al. 2006).10 Understanding the transition
10

Some other objects are suspected of being L+T binaries, i.e., systems composed of an L dwarf and a T dwarf, based on their resolved optical data (2MASS J0850+1057 [ Reid et al. 2001] and 2MASS J1728+3948 [Gizis et al. 2003]), near-IR flux ratios (Gl 337CD; Burgasser et al. 2005a), or integrated-light spectra (2MASS J0518þ2828 [Cruz et al. 2004] and 2MASS J0920+3517 [Reid et al. 2001; Nakajima et al. 2001; Burgasser et al. 2006]).

1393


1394

LIU ET AL.

Vol. 647

between objects of spectral type L, marked by very red IR colors and metal hydride absorption, and those of spectral type T, marked by blue IR colors and methane absorption, has been an active area of inquiry (e.g., Leggett et al. 2005). An expanded sample of L-type and T-type binaries should shed light on this issue. We are conducting an imaging survey of nearby brown dwarfs using the new Keck laser guide star adaptive optics system ( LGS AO). LGS AO enables near diffraction-limited imaging over most of the sky with a 10 m telescope. Thus, LGS AO is a powerful new capability for finding and characterizing substellar binaries. We have previously used Keck LGS AO to discover that the nearby L dwarf Kelu-1 is a binary system ( Liu & Leggett 2005), thereby explaining the unusual properties of Kelu-1 (high luminosity, very red color, high inferred effective temperature, and low lithium absorption line strength) compared to other early L dwarfs. In this paper, we present the discovery of the binarity of SDSS J153417.05+161546.1 (hereafter SDSS J1534+1615), a recently discovered T dwarf with an integrated-light near-IR spectral type of T3:5 ô 0:5 (Chiu et al. 2006), and consider the implications for understanding the L / T transition. 2. OBSERVATIONS We observed SDSS J1534+1615 on 2005 May 1 UT using the sodium LGS AO system ( Bouchez et al. 2004; Wizinowich et al. 2004) of the 10 m Keck II telescope on Mauna Kea, Hawaii. We used the facility IR camera NIRC2 and the J (1.25 m), H (1.64 m), and K 0 (2.12 m) filters from the Mauna Kea Observatories ( MKO) filter consortium (Simons & Tokunaga 2002; Tokunaga et al. 2002). We used NIRC2's narrow camera, which produces a 0B00994 pixelþ1 scale and a 10B2 field of view. Conditions were photometric with excellent seeing conditions during the night, better than 0B6 in the optical as reported by the neighboring Keck I telescope. The total setup time for the telescope to slew to SDSS J1534+1615 and for the LGS AO system to be fully operational was 18 minutes. SDSS J1534+1615 was observed over an air mass range of 1.14Y1.24. The LGS brightness, as measured by the flux incident on the AO wave front sensor, was equivalent to a V % 10:3 mag star. The LGS provided the wave front reference source for AO correction, with the exception of tip-tilt motion. Tip-tilt aberrations and quasi-static changes in the image of the LGS as seen by the wave front sensor were sensed using the R ¼ 15:9 mag star 1062-0236346 from the USNO-B1.0 catalog ( Monet et al. 2003), located 1400 away from SDSS J1534+1615. The overall image quality was somewhat worse compared to other data obtained on the same night ( Liu & Leggett 2005) due to higher winds during the SDSS J1534+1615 observations. The LGS AOYcorrected images have full widths at half-maximum ( FWHMs) of 0B08, 0B06, and 0B07 at JHK 0 , respectively, with rms variations of about 5% Y10%. The corresponding Strehl ratios were 0.02, 0.06, and 0.20 with rms variations of about 5%Y20%. The FWHM and Strehl measurements were derived from the result of the binary fitting process described below. We obtained a series of five images in each filter, dithering the telescope by a few arcseconds between each image. The total onsource integration time per filter was 7.5 minutes. The sodium laser beam was steered with each dither such that the LGS remained on SDSS J1534+1615 for all exposures. This object was easily resolved into a binary system in all our data. The images were reduced in a standard fashion. We constructed flat fields from the differences of images of the telescope dome interior with and without lamp illumination. Then we created a master sky frame from the median average of the bias-subtracted, flat-fielded im-

ages and subtracted it from the individual images. Images were registered and stacked to form a final mosaic ( Fig. 1). The known small geometric distortions in NIRC2 have a negligible effect on our analysis. To measure the flux ratios and relative positions of the two components, we adopted two methods. (1) The first method used an analytic model of the point-spread function ( PSF ) as the sum of two elliptical Gaussian components, a narrow component for the PSF core and a broad component for the PSF halo. (2) The second method empirically determined the PSF using the Starfinder software package ( Diolaiti et al. 2000), which was designed for analysis of blended AO images. For the individual images obtained with each filter, we fitted for the flux ratio, separation, and position angle of the binary. The averages of the results were adopted as the final measurements and the standard deviations as the errors. The results from the two different methods were generally consistent. In order to gauge the accuracy of these two measurement methods, we created artificial binary stars from data of single stars that had comparable image quality (similar tip-tilt star properties, Strehl ratios, and FWHM ) to the SDSS J1534+1615 data. Our two methods were applied to the artificial binaries to assess the fitting results and derived errors. Based on this analysis, we adopted the Starfinder measurements for the H- and K 0 -band data and the Gaussian fitting results for the J-band data. These choices were not unexpected: the higher Strehl, long-wavelength data had a more structured PSF that was better treated by Starfinder, while the much smoother, low-Strehl J-band images were modeled better by the two-Gaussian PSF. Table 1 presents the resulting flux ratios and astrometry results. Table 2 reports the IR apparent magnitudes and colors of the individual components, as derived from our flux ratio measurements and the integrated IR magnitudes reported by Chiu et al. (2006). The absolute magnitudes for SDSS J1534+1615 are unknown, as a trigonometric parallax is not available for this system. In addition, in principle variability could affect the photometry determined for the individual components, since the integrated-light photometry was obtained at a different epoch than the Keck LGS AO measurements of the flux ratios. IR monitoring of a limited number of T dwarfs does find low-level variability in some objects, although the timescales and persistence are not well constrained ( Enoch et al. 2003; Koen et al. 2004, 2005). SDSS J1534+1615 has been observed at four epochs by the Sloan Digital Sky Survey (SDSS) over a span of 11 months and shows no z-band (0.9 m) variability at the $0.1 mag level. Our NIRC2 J- and H-band imaging used the same MKO filters as Chiu et al. (2006), and hence computing magnitudes for the individual components from our measured flux ratios is straightforward. However, our NIRC2 data were obtained with the MKO K 0 -band (2.12 m) filter in order to minimize the thermal background from the sky + AO system, whereas the Chiu et al. photometry used the MKO K-band (2.20 m) filter. Liu & Leggett (2005) reported a polynomial fit for the (K 0 þ K )MKO color term as a function of spectral type for L and T dwarfs using synthetic photometry based on the data from Knapp et al. (2004). However, for the range of spectral types relevant to SDSS J1534+ 1615 (see x 3), the color term depends strongly on the spectral type inferred from the J þ H color. Instead, we use the synthetic photometry to find a direct relation for the color term as a function of the J þ H color: (K 0 þ K )
MKO

¼ þ0:100 × 0:143(J þ H )

MKO

;

Ï


No. 2, 2006

NOVEL T DWARF BINARY FROM LASER GUIDE STAR AO

1395

Fig. 1.--JHK 0 -band imaging of SDSS J1534+1615AB from Keck LGS AO. North is up and east is to the left. Each image is 0B75 on a side. The binary separation is 0B110 ô 0B005. Contours are drawn from 90%, 45%, 22.5%, 11.2%, 5.6%, and 2.8% of the peak value in each bandpass. The source to the southeast (left) is component A (brighter at H and K 0 , fainter at J, spectral type T1:5 ô 0:5), and the source to the northwest (right) is component B ( fainter at H and K 0 , brighter at J, spectral type T5:5 ô 0:5). The images are slightly east-west elongated due to telescope windshake.

TABLE 1 SDSS J153 4+16 15AB Bi nary Pr o pe rt ie s Property Separation....................................................... Position angle................................................. à J .................................................................. à H ................................................................. à K 0 ................................................................ Proper motion: ............................................................... ................................................................ Mass ratio MB /MA .......................................... Luminosity ratio log (LA /LB ) ......................... Surface gravity ratio log ( gA /gB ) ................... Estimated distance ......................................... Total mass: 0.5 Gyr ....................................................... 1.0 Gyr ....................................................... 5.0 Gyr ....................................................... Estimated orbital period................................. Note.--All photometry on the MKO system. Measurement 110 287N6 þ0.17 0.70 1.07 þ142 þ57 0.80 0.3 0.10 ô ô ô ô ô 5 mas 1N3 0.04 mag 0.04 mag 0.05 mag
þ1 þ1

valid for L and T dwarfs with þ0:5 < (J þ H ) < 1:0. The resulting (K 0 þ K ) terms are 0.02 and þ0.13 mag for components A and B, respectively. The rms scatter about the fit is 0.03 mag , which we add in quadrature when determining the K-band photometry errors of the individual components. Figure 1 shows the unique behavior of the near-IR flux ratios for SDSS J1534+1615AB compared to all other known substellar binaries: namely, the identification of the brighter component depends on the observing wavelength. We designate the eastern object as the ``A'' component, based on the fact that its inferred
TABLE 2 Resolv ed Pr operties of SDSS J1534+1615 AB Property J (mag) ................................. H (mag) ................................ K (mag) ................................ J þ H (mag) ........................ H þ K (mag) ........................ J þ K (mag) ......................... Estimated spectral typea....... SDSS J1534+1615A 17.43 16.83 16.37 0.60 0.46 1.06 T1.5 ô ô ô ô ô ô ô 0.04 0.04 0.06 0.06 0.07 0.07 0.5 SDSS J1534+1615B 17.26 17.53 17.58 þ0.27 þ0.05 þ0.32 T5.5 ô ô ô ô ô ô ô 0.04 0.04 0.06 0.06 0.07 0.07 0.5

ô 14 mas yr ô 10 mas yr ô 0.05 ô 0.1 dex ô 0.05 dex %36 pc 60MJup 85MJup 145MJup 22Y33 yr

Note.--All photometry on the MKO system. a Based on the Burgasser et al. (2006) classification scheme.


1396

LIU ET AL.

Vol. 647

spectral type is earlier than the western component, as discussed in x 3. The eastern object is brighter in the H and K 0 bands, and we would also expect it to be the brighter component at optical wavelengths, based on the T dwarfs with known distances ( Tinney et al. 2003). The western, later type object is brighter only in the J-band images. This can be understood as a manifestation of the J-band brightening seen in the absolute magnitudes of T dwarfs, wherein objects of spectral types from about T2 to T5 are notably brighter at J band compared to earlier type objects ( Dahn et al. 2002; Tinney et al. 2003; Vrba et al. 2004). We discuss this further in x 4. No other sources are detected in our images. For objects at k0B3 from SDSS J1534+1615, the final mosaics reached point-source detection limits of 21.0, 21.0, and 20.7 mag at JHK 0 , respectively, out to a separation of 300 . For an age range of 0.5Y5 Gyr and a estimated distance of 36 pc (see x 3), these detection limits correspond to about 10MJupY30MJup companions around SDSS J1534+1615 based on the models of Baraffe et al. (2003). 3. RESULTS: PROPERTIES OF SDSS J1534+1615AB With only one epoch of imaging, we cannot ascertain whether the two objects in our Keck images are physically associated based solely on common proper motion. However, the combination of the small proper motion and the distinctive optical / IR colors can be explained only if the system is a binary T dwarf. No relative motion was detected between the two components during the 30 minutes elapsed time for the Keck LGS AO imaging, so component B cannot be a solar system body (which would be expected to show a relative motion of k0B5). SDSS J1534+ 1615 was detected at the J and H bands by the Two Micron All Sky Survey (2MASS) in 1999 and by SDSS in both 2004 and 2005, giving a proper motion of 0B15 yrþ1 at position angle 248 east of north. The small proper motion indicates that the optical and IR photometry in Chiu et al. (2006) represents the combined light of both objects, and thus our Keck IR flux ratio measurements can be used to derive the resolved photometry of the individual components. The resulting IR colors for the two components are entirely consistent with those of T dwarfs. Component B does have neutral JHK colors similar to hot O stars and white dwarfs, but if it were either of these objects, the integrated optical light of the system would be predominantly blue rather than very red as observed (z þ J ¼ 3:61 ô 0:10 mag; Chiu et al. 2006). Hence, we conclude that the two objects constitute a binary T dwarf. To infer spectral types, we compare the resolved JHK colors of SDSS J1534+1615AB to those of ultracool dwarfs from Knapp et al. (2004) and Chiu et al. (2006), excluding any known binaries. Our measurements and the published photometry were both obtained with the MKO filter system. For the L dwarfs, we use the published spectral types, which are based on the Geballe et al. (2002) near-IR classification scheme. For the T dwarfs, we use near-IR classifications from the Burgasser et al. (2006) scheme. ( We adopt these two classification schemes throughout this paper, which includes updating older published T dwarf types to the new Burgasser et al. scheme.) We assume that the components of SDSS J1534+1615 are themselves single, not unresolved binaries. Figure 2 shows that component A has IR colors most typical of the known T1 dwarfs, with T2 also being possible. ( The colors of component A are also similar to the few late L dwarfs with unusually blue near-IR colors in Fig. 2. However, as discussed below, modeling of the integrated-light spectrum indicates that this is unlikely.) Component B has IR colors most similar to the T5 dwarfs and some of the T6 dwarfs. We

estimate spectral types of T1:5 ô 0:5 and T5:5 ô 0:5 for the two components.11 We modeled the integrated-light spectrum of SDSS J1534+ 1615 as the blend of two template T dwarfs, in order to check the consistency with the flux ratios and spectral types inferred from our Keck LGS imaging. For templates, we chose T dwarfs from Figure 2 with complete JHK spectra and having near-IR colors similar to those of the two components of SDSS J1534+1615 (color differences of P0.1 mag). Then we scaled the template spectra to match the observed J-band flux ratio, summed the templates to form a composite spectrum, and scaled the resulting composite to the observed integrated-light spectrum. ( Note that this approach has only one free parameter in the modeling, namely, the distance to the binary.) The comparisons between the model composite spectra and the observed spectrum show good agreement ( Fig. 3).12 Likewise, although SDSS J1534+1615A has near-IR colors similar to some of the bluest late L dwarfs, it was not possible to create a well-matched composite spectrum with a blue late L dwarf as a template, and thus we discount this possibility. Overall, the spectral types estimated from modeling the integrated-light spectrum agree well with those from the resolved Keck near-IR colors. In the absence of a trigonometric parallax, the model composite spectra can provide a rough distance estimate for SDSS J1534+1615AB. The spectra were created using 2MASS J2356þ 1553 ( T5.5; Burgasser et al. 2002b) and SDSS J2124+0100 ( T5; Chiu et al. 2006) to represent component B. Vrba et al. (2004) have measured a trigonometric parallax for 2MASS J2356þ1553. For SDSS J2124+0100, we assumed an absolute J-band magnitude of 14.37 mag, based on the polynomial fit for the absolute J-band magnitude as a function of spectral type from Knapp et al. (2004). Assuming that these two mid-T templates are single objects, the inferred distance modulus for SDSS J1534+1615AB is 2.7Y2.9 mag (%36 pc). To check on the distance estimated from our spectral modeling, we applied the same procedure to the integrated-light spectrum of the binary T dwarf Ind Bab ( McCaughrean et al. 2004) and derived a distance estimate of 3.3 pc, in good agreement with the measured distance of 3:626 ô 0:009 pc ( Perryman et al. 1997). We can estimate the mass ratio of SDSS J1534+1615AB, independent of its distance. For the assumed spectral types of T1:5 ô 0:5 and T5:5 ô 0:5, the difference in bolometric corrections for the two components is 0:55 ô 0:15 mag, with BCK being larger for the T1:5 ô 0:5 component (Golimowski et al. 2004a). The K-band flux ratio for the system is 1:21 ô 0:05 mag, based on our Keck LGS images and the (K 0 þ K ) color term calculated in x 2. Hence, the difference in bolometric luminosities is 0:7 ô 0:2 mag, with component A having the higher luminosity. Burrows et al. (2001) provide an approximate massluminosity relation for coeval solar metallicity brown dwarfs, Lbol / M 2:64 . This implies a mass ratio of about 0:80 ô 0:05 for
11 We assume that the metallicity of SDSS J1534+1615 is not very different from that of the bulk of the known T dwarfs. The low proper motion argues against it being a halo object. Likewise, the one candidate low-metallicity T dwarf reported by Burgasser et al. (2003a) is distinguished by an exceptionally blue H þ K color, whereas the colors of SDSS J1534+1615 are consistent with other T dwarfs. 12 We also tried using T dwarfs with known distances as templates (as compiled in Knapp et al. 2004), directly summing the flux-calibrated spectra without imposing the J-band flux ratio. The resulting set of blended spectra could not simultaneously match the SDSS J1534+1615 integrated-light spectrum and the observed JHK flux ratios. The mismatch is not surprising given that most of the T dwarfs with measured distances do not have the same near-IR colors as SDSS J1534+1615A and B. As Fig. 2 shows, there is significant color scatter within a given T subclass.


No. 2, 2006

NOVEL T DWARF BINARY FROM LASER GUIDE STAR AO

1397

Fig. 2.--Near-IR colors of SDSS J1534+1615AB compared with nearby late L and T dwarfs from Knapp et al. (2004) and Chiu et al. (2006). The numbers indicate the near-IR spectral subclass of the objects, with half-subclasses being rounded down (e.g., T3.5 is labeled as ``3''). The late L dwarfs (classified on the Geballe et al. 2002 scheme) are plotted as bare numbers. The T dwarfs (on the Burgasser et al. 2006 scheme) are plotted as circumscribed numbers, with squares for integer subclasses (e.g., T3) and diamonds for half-subclasses (e.g., T3.5). The photometry errors are comparable to or smaller than the size of the plotting symbols. Known binaries are excluded except for SDSS J1534+1615AB, which is labeled on the plot as A, B, and AB. The inferred spectral types for components A and B of SDSS J1534+1615 are T1:5 ô 0:5 and T5:5 ô 0:5, respectively. The individual components of Ind Bab are shown for comparison, where we have converted the published photometry to the MKO system (see x 4.1) and assumed errors of 0.05 mag.

the two components of SDSS J1534+1615, which is similar to the nearly equal mass ratios found for other binary brown dwarfs (e.g., Burgasser et al. 2003b). Likewise, we can estimate the surface gravity difference for the two components, independent of the distance. Since the radii of brown dwarfs older than k300 Myr are nearly independent of the masses, the near-unity mass ratio means that the surface gravities of SDSS J1534+1615A and B are very similar. Burrows et al. (2001) present analytic fits to substellar evolutionary modþ els that show that the quantity g1:7 TeA2:8 is constant at fixed age. Adopting TeA ranges of %1300Y1400 and %1100Y1200 K for components A and B, respectively (Golimowski et al. 2004a, which assumes an age of 3 Gyr), the inferred difference in surface gravities is approximately 0.05Y0.15 dex. We can make a rough estimate of the orbital period of SDSS J1534+1615AB, assuming that the true semimajor axis is not very different from the projected separation. ( For example, Fischer & Marcy [1992] show for binaries with a random distribution of orbital elements that on average the true semimajor axis is about

1.3 times larger than the projected semimajor axis.) To estimate the total system mass, we use the 36 pc distance inferred from the composite spectra fitting and the aforementioned BCK to estimate bolometric luminosities of 2:2 ; 10þ5 and 1:2 ; 10þ5 L for the two components. Models by Burrows et al. (2001) give total masses for the system of about 60MJup ,85MJup, and 145MJup for assumed ages of 0.5, 1.0, and 5.0 Gyr, respectively. For an assumed semimajor axis of 4.0 AU, these mass estimates lead to an expected orbital period of 22Y33 yr. More generally, Torres (1999) shows that %85% of randomly oriented orbits have a true semimajor axis of 0.5Y2.0 times the projected separation, corresponding to a range of orbital periods of %8Y95 yr for SDSS J1534+1615AB. 4. DISCUSSION 4.1. The Nature of the L / T Transition The rapid changes in near-IR colors at the L / T transition are generally understood to be due to a change in the photospheric


Fig. 3.--Results from modeling the integrated-light near-IR spectrum of SDSS J1534+1615 (black lines; Chiu et al. 2006) as the sum of two T dwarfs (red lines): one early T template (blue solid lines) and one late T template (green dashed lines). The template objects are chosen to have IR colors similar to those of the resolved components of SDSS J1534+1615AB, and the observed J-band flux ratio is used in scaling and combining the template spectra. The only free parameter in the fitting is the distance to SDSS J1534+1615AB. (See x 3 for details.) The agreement between the observed spectrum and the modeled blends is good, especially for the topmost panel, meaning that the spectral types inferred for SDSS J1534+1615A and B from the near-IR colors are consistent with the integrated-light spectrum. The bottommost panel shows the match attempted using a blue late L dwarf as component A; the depth of the J-band water absorption is not well matched. The template spectra are from Chiu et al. (2006), Knapp et al. (2004), and Burgasser et al. (2002b).


NOVEL T DWARF BINARY FROM LASER GUIDE STAR AO dust content, but the very abrupt nature of this effect over a small inferred range in effective temperature remains a challenge to theoretical models. One key observational characteristic of the L / T transition is the unusual strong ($1 mag) brightening of the J-band flux among the early- and mid-T dwarfs compared to the late L dwarfs ( Dahn et al. 2002; Tinney et al. 2003; Vrba et al. 2004). This brightening is also seen at far-red (I and Z ) and other near-IR wavelengths (H and K ), although the effect is greatest at the J band. This wavelength dependence is expected, since at the J band the photospheric opacities are most sensitive to the properties of the clouds ( Burrows et al. 2006). Models can qualitatively accommodate the observed transition from the very red L dwarfs to the very blue T dwarfs ( Marley et al. 2002; Tsuji 2002; Burrows et al. 2006). However, no model thus far has successfully reproduced the observed J-band brightening along a single evolutionary track. Based on a suggestion by Ackerman & Marley (2001), Burgasser et al. (2002a) show that disruption of the condensate clouds at fixed TeA could produce an enhancement of the J-band flux, as holes in the cloud deck reveal the deeper, hotter regions of the photosphere. Knapp et al. (2004) speculate that a rapid increase in the dust sedimentation efficiency over a small range of TeA could be responsible. However, as pointed out by Burrows et al. (2006), these explanations are ad hoc, lacking a firm physical basis for the rapid transition. A very different picture is suggested by Tsuji & Nakajima (2003). They propose that the J-band brightening does not occur along the evolutionary track of a given object. Rather, in their ``unified cloudy model'' the TeA at which the L dwarf sequence merges with the T dwarf sequence depends on surface gravity: for younger or lower mass objects, the L / T transition occurs at younger ages and brighter J-band absolute magnitudes than for older or higher mass objects. Thus, in their models the J-band brightening arises from the mass and age spreads within the known ensemble of L and T dwarfs and does not occur during the evolution of a single object. Objects that comprise the J-band brightening are predicted to be younger and to have lower surface gravities than objects that do not (cf. Tinney et al. 2003). The flux crossover in the near-IR that we observe in SDSS J1534+1615AB provides a benchmark for understanding the J-band brightening, as the binary presumably constitutes a system of common age and metallicity and nearly identical surface gravity. This crossover cannot be explained by the aforementioned unified cloudy model of Tsuji & Nakajima (2003). Their model isochrones show that no J-band brightening occurs for a coeval population, in disagreement with our observations of SDSS J1534+1615AB. The fact that SDSS J1534+1615AB displays the J-band brightening between its two coeval components reveals that the phenomenon can originate from processes intrinsic to ultracool photospheres with very similar surface gravities, such as the aforementioned cloud disruption and /or rapid dust settling scenarios.13 Similarly, models by Burrows et al. (2006) indicate that variations in metallicities among the observed L and T dwarfs could be responsible for the J-band brightening (although this is not the preferred solution of Burrows et al.). SDSS J1534+1615AB demonstrates that objects of the same metallicity can exhibit the J-band brightening. A good point of comparison is the T dwarf binary Ind Bab ( McCaughrean et al. 2004). Scholz et al. (2003) determined an
13 The reappearance of KI and FeH absorption in the far-red spectra of T2YT5 dwarfs, after having weakened among the late L dwarfs ( Burgasser et al. 2002a; Nakajima et al. 2004; Cushing et al. 2005), also argues that true photospheric processes are involved in the J-band brightening.

1399

integrated-light spectral type of approximately T2.5 using the Geballe et al. (2002) classification scheme; we analyzed the published spectrum and found that the type is unchanged with the newer Burgasser et al. (2006) scheme. The individual components of this 0B7 (2.6 AU ) binary have spectral types of T1 and T6 in the Burgasser et al. scheme. The near-IR colors are consistent with Ind Ba having a spectral type comparable to that of SDSS J1534+1615A and Ind Bb being about one subclass later than SDSS J1534+1615B (Fig. 2).14 However, the flux ratios of Ind Bab do not show the crossover seen between the T1.5 and T5.5 components of SDSS J1534+1615AB. This suggests that by spectral type T6 the J-band brightening no longer occurs, consistent with the observed CMD of T dwarfs (and ignoring possible complications due to surface gravity effects, as the Ind system is somewhat younger [1Y2 Gyr] than expected for field objects). 4.2. Is the J-Band Brightening Enhanced by Unrecognized Binarity? The flux crossover seen in the near-IR for SDSS J1534+ 1615AB demonstrates that the J-band brightening is a true physical effect and not solely a manifestation of large physical variations among the L and T dwarfs with known distances (unless SDSS J1534+1615B itself is an unresolved binary). However, the integrated-light photometry of SDSS J1534+1615 is in accord with the other J-band brightening objects ( Figs. 2, 4, and 5), most of which have not yet been imaged at high angular resolution.15 This suggests that unrecognized binarity among early/ mid-T dwarfs may strongly affect the appearance of the J-band brightening in the brown dwarf CMD. Furthermore, Burgasser et al. (2005b) have recently noted that objects with spectral types of L7YT2 appear to have a higher binary frequency relative to earlier type L dwarfs and mid / late-type T dwarfs (although a complete analysis of the selection biases is still needed ). This finding covers an earlier spectral type range than the J-band brightening objects, but it does provide more evidence that binarity likely plays an important role in the L / T transition. To examine this notion further, here we consider a scenario in which truly single J-band brightening objects are in fact rare. In this situation, it is the blended light of two components with different spectral types that produces an integrated-light spectral type of %T2YT4.5. Thus, the binary frequency of these spectral subclasses is greater compared to other T dwarfs, and the apparent amplitude of the J-band brightening effect is enhanced. Vrba et al. (2004) suggested that the J-band brightening cannot be due to unresolved binarity because a significant number of objects contribute to the effect. However, their implicit assumption is that a given spectral subclass contains a mix of singles and binaries. Their objection would not be relevant in our scenario, wherein the reason that most objects are classified as T2YT4.5 is because they are binaries.

14 We recomputed photometry for the individual components of Ind Bab. The McCaughrean et al. (2004) determination assumed that the Very Large Telescope NACO and 2MASS filter systems were identical. Based on synthetic photometry, we determined the (small) color terms between the two filter systems specific for T dwarfs and then used the published measurements to recompute the colors: (J þ H )MKO ¼ f0:53; þ0:34g, (H þ K )MKO ¼ f0:18; þ0:34g, and (J þ K )MKO ¼ f0:71; þ0:69g for components A and B, respectively. The resulting colors are 0.03 mag different from the published values. 15 Likewise, the integrated-light spectrum of SDSS J1534+1615 shows no strong differences from the spectra of the other two known T3.5 objects, SDSS J1750+1759 (Geballe et al. 2002) and SDSS J1214+6316 (Chiu et al. 2006 ).


1400

LIU ET AL.

Vol. 647

Fig. 4.-- J-band absolute magnitude as a function of near-IR spectral type and J þ K color, based on data tabulated by Knapp et al. (2004) and excluding sources with S/ N < 4 parallax measurements (SDSS J1435þ0043 [T6], SDSS J0207+0000 [T4.5], and SDSS J0837þ0000 [T0.5]). Spectral types are based on the Geballe et al. (2002) scheme for the L dwarfs and the Burgasser et al. (2006) scheme for the T dwarfs. The integrated-light properties of the binary T dwarfs SDSS J1534+1615AB, Ind Bab, and SDSS J0423þ0414AB are plotted as colored open circles, and the colored filled squares show the individual components. (For the latter two objects, see x 4 for a description of their resolved measurements.) The distance to SDSS J1534+1615AB is based on modeling the integrated-light spectrum as the blend of two components (see x 3) and thus is more uncertain than the trigonometric distances for Ind Bab and SDSS J0423þ0414AB. Errors in M (J ) are comparable to the symbol size. On the left, the solid line shows a fifth-order polynomial fit for M (J ) versus near-IR spectral type, excluding known binaries. To illustrate the potential significance of binarity, the dashed line shows the same order fit, excluding both known and possible binaries. Possible binaries were selected based on their relatively high TeA compared to objects of similar spectral type in the Golimowski et al. (2004a) measurements (see x 4.2); these are SDSS J0032+1410 ( L8), 2MASS J0328+2302 ( L9.5), SDSS J1254þ0122 ( T2), SDSS J1021þ0304 ( T3), SDSS J1750+1759 ( T3.5), and 2MASS J0559þ1404 ( T4.5). Table 3 provides the coefficients of the polynomial fits.

The observed CMD of T dwarfs lends some support to the idea that binarity might enhance the amplitude of the J-band brightening. Unrecognized binarity can add up to 0.75 mag of scatter in the absolute magnitudes, in addition to scatter caused by other effects, e.g., variations in surface gravity ( Knapp et al. 2004; Burrows et al. 2006). Therefore, we would naturally expect at least this much scatter in the absolute magnitudes at a given spectral subclass, and this is largely observed (e.g., Fig. 4 in this paper; Fig. 2 of Vrba et al. 2004; Fig. 9 of Tinney et al. 2003; Fig. 8 of Knapp et al. 2004). However, the J-band brightening objects appear to have a smaller scatter in their absolute magnitudes compared to earlier and later spectral subclasses. This reduced scatter could be explained if in fact most J-band brightening objects are binaries, with truly single objects being rare. (An alternative explanation would be that the reduced scatter arises from a selection effect in the targets chosen by the parallax programs. However, it is not obvious what effect would act only on the J-band brightening objects and not the other subclasses.) A similar hint, in a slightly different form, is seen in the effective temperatures of L and T dwarfs derived by Golimowski et al. (2004a). Their results suggest that T2YT4.5 dwarfs may have slightly higher temperatures than the late L and T0YT1 dwarfs (see their Fig. 6). While this discrepancy is within the uncertainties in the age estimates (which affect the model-derived radii used in the TeA calculation), the systematically higher TeA 's are a curious trend in the data, especially under the assumption that the spectral type scale is a proxy for the TeA scale (cf. Tsuji 2005). One possibility would be that the T2YT4.5 dwarfs in Golimowski et al. have systematically younger ages than the earlier or later type objects, but again it is not obvious how such a selection bias would occur. A more natural explanation is that the elevated TeA 's result from unrecognized binarity. This is amply

demonstrated by the L3 dwarf Kelu-1 and the T0 dwarf SDSS J0423þ0414 --their effective temperatures appear to be anomalously large compared to objects of similar spectral types, but these deviations are in fact an artifact of previously unrecognized binarity ( Liu & Leggett 2005; Burgasser et al. 2005b). For the case of equal-luminosity components, TeA will be overestimated by about 20% (21/4 ); applying this correction to the Golimowski et al. T2YT4.5 dwarfs would bring their temperatures into agreement with the earlier type objects.16 In fact, if all the L and T dwarfs with anomalously high temperatures in the Golimowski et al. sample are actually unresolved binaries, we would infer that the J-band absolute magnitudes across the L / T transition are relatively constant ( Fig. 4). Similarly, the inflections in the brown dwarf CMDs at other wavelengths would also be affected. Accounting for the possible binaries, Figure 5 shows that the apparent plateau in the H-and K-band fluxes of the early/mid-T dwarfs may be lessened, leading to a more monotonic dependence on the spectral type. Table 3 provides polynomial fits for the absolute magnitude as a function of spectral type, with and without the possible binaries. At face value, a higher binary frequency among the %T2Y T4.5 objects relative to earlier and later type objects may seem physically implausible. However, the binary frequency as a function of spectral type is not the same as the frequency as a function of TeA . While the frequency as a function of TeA is expected to change smoothly, the frequency as a function of spectral type may not show the same behavior, since it also depends on how

16 By the same rationale, SDSS J0032+1410 ( L8) and 2MASS J0328+ 2302 ( L9.5) are candidates for being unresolved binaries based on their relatively high TeA 's.


No. 2, 2006

NOVEL T DWARF BINARY FROM LASER GUIDE STAR AO

1401

Fig. 5.--Same as Fig. 4, but for H and K-band absolute magnitudes as a function of near-IR spectral type and J þ K color.

the light of individual binary components blends together. In particular, the large changes in spectral morphology and the nonmonotonic behavior of the absolute magnitudes during the L / T transition may lead to interesting observational consequences when these objects form binaries.17 For the sake of illustration, if we consider an extreme example in which truly single T2YT4.5 objects do not exist at all, the SDSS J1534+1615AB ( T3.5) and Ind Bab ( T2.5) systems demonstrate that these subclasses can be populated with binaries composed of an early- and a late-type T dwarf--in this case, the T2YT4.5 dwarfs would have a 100% binary fraction, even if the earlier and later type T dwarfs have much lower binary fractions.
Since the IR absolute magnitudes around the L/T transition do not decrease monotonically as a function of spectral type, unresolved binaries in the L/T transition region with a range of mass ratios can be significantly displaced in the CMD in both absolute magnitude and integrated-light spectral type compared to the individual components. In contrast, if the magnitudeYspectral type relation were monotonic and very steep, then a population of unresolved binaries would mostly lead to a spread in the magnitudes, due to the equal-mass binaries; the unequal-mass binaries would not have much affect on the CMD, since the secondary components would not substantially contribute to the integrated-light properties.
17

A relative paucity of single objects in the J-band brightening regime can be naturally explained if the corresponding TeA range is very small, e.g., as suggested by Kirkpatrick et al. (2000) and indicated by measurements from Vrba et al. (2004) and Golimowski et al. (2004a). The small TeA range would correspond to a relatively short-lived evolutionary phase, and therefore relatively few substellar objects would be found in the corresponding spectral subclasses. However, even if truly single %T2YT4.5 objects are rare, a higher binary frequency among these objects may not be apparent in the spectral type distribution of objects found by 2MASS and SDSS. In such magnitudelimited surveys, the effective detection volume for binary stars is larger than for single stars, and thus a survey's observed spectral type distribution depends on several factors--the true binary frequency, the slope of the relationship between absolute magnitude and spectral type, and the effect of unresolved binarity on spectral types determined from integrated-light photometry and /or spectroscopy. A Monte Carlo simulation, beyond the scope of this paper, would be valuable in understanding the interplay of these effects. Given that only six T2YT4.5 dwarfs have measured distances, the discussion here is inevitably speculative--such a small sample


1402

LIU ET AL.
TABLE 3 Coefficien ts of Polynomial F its t o Absolute M agnitudes v ersus Near-IR Sp ectral Type rms (mag)

Vol. 647

Magnitude

c0

c1

c2

c3

c4

c5

Excluding Known Binaries M (JMKO ) ............... M (HMKO ).............. M (KMKO ) .............. 11.746 11.263 10.731 þ2.259Eþ1 þ4.164Eþ1 þ3.964Eþ1 3.229Eþ1 3.565Eþ1 3.150Eþ1 þ5.155Eþ2 þ5.610Eþ2 þ4.912Eþ2 2.966Eþ3 3.349Eþ3 2.994Eþ3 þ5.648Eþ5 þ6.720Eþ5 þ6.179Eþ5 0.39 0.35 0.39

Excluding Known and Possible Binaries M (JMKO ) ............... M (HMKO ).............. M (KMKO ) .............. 11.359 10.767 10.182 3.174Eþ1 2.744Eþ1 3.688Eþ1 1.102Eþ1 8.955Eþ2 2.028Eþ2 þ1.877Eþ2 þ1.552Eþ2 þ4.488Eþ3 9.169Eþ4 8.319Eþ4 2.301Eþ4 þ1.233Eþ5 þ1.315Eþ5 þ2.466Eþ6 0.37 0.30 0.34

Notes.--These are the coefficients of the fifth-order polynomial fits plotted in Figs. 4 and 5 for the IR absolute magnitudes as a function of near-IR spectral type. The fits are defined as magnitude ¼
5 X i¼0

ci ; (SpT)i ;

where the numerical spectral type is defined as SpT ¼ 1 for L1, SpT ¼ 2 for L2, SpT ¼ 10 for T0, etc. The L dwarfs are classified on the Geballe et al. (2002) scheme and the T dwarfs on the Burgasser et al. (2006) scheme. The fits are applicable from L1 to T8. Possible binaries were selected based on their relatively high TeA compared to objects of similar spectral type; see x 4.2 and the legend of Fig. 4. The last column gives the standard deviation about the fit.

hinders more detailed considerations.18 Similarly, our adopted spectral range of %T2YT4.5 for the strong J-band brightening phenomenon should be considered as approximate. The T dwarf binaries SDSS J1534+1615AB, Ind Bab, and SDSS J0423þ 0414AB all have components with estimated spectral types of T1YT2. Unless these components are themselves binary, this shows that single early T dwarfs are not uncommon. Indeed, the positions of these components in brown dwarf CMDs are quite similar ( Figs. 4 and 5) and are somewhat fainter than the T2 dwarf SDSS J1254þ0122, which thus far appears to be a single object ( Burgasser et al. 2003b). If binarity does play a significant role in the J-band brightening objects, there are some natural consequences. (1) More extensive imaging surveys with HST and LGS AO should find a greater binary frequency among these objects compared to earlier and later type objects, after accounting for selection effects inherent in current magnitude-limited samples of brown dwarfs. Of course, some systems may have projected separations below the resolution limit of imaging surveys and would only be detectable as spectroscopic binaries (e.g., Maxted & Jeffries 2005). (2) Such binaries should often be composed of an earlier type object, say late L to early T, and an object of approximately T5 or later. (3) More generally, the J-band brightening objects should have relatively lower space densities compared to other ultracool dwarfs. In this regard, a populous volume-limited sample of L and T dwarfs with trigonometric parallaxes would be valuable for studying the nature of the L / T transition. Such a sample is unlikely to be achieved with current parallax programs, which
18 The six T2YT4.5 dwarfs with known distances are SDSS J1254þ0122 ( T2), IndBab ( T2.5in integratedlight), SDSS J1021þ0304 ( T3), SDSS J1750+1759 ( T3.5), 2MASS J0559þ1404 ( T4.5), and SDSS J0207+0000 ( T4.5); see compilation in Burgasser et al. (2006). In addition, SDSS J0423þ0414 ( T0 in near-IR integrated light) has recently been resolved into a binary by Burgasser et al. (2005b); the components have estimated spectral types of L6 ô 1 and T2 ô 1 based on their near-IR fluxes and modeling of the combined-light spectrum. Resolved JHK magnitudes are not available for this system, so we estimate them for Figs. 4 and 5 by scaling template spectra of the individual components to the resolved optical fluxes.

can target only a limited number of preselected objects. The upcoming Pan-STARRS project ( Panoramic Survey Telescope and Rapid Response System; Kaiser et al. 2002) will monitor the entire sky observable from Hawaii with high astrometric precision ( planned one-dimensional rms astrometric error of 10 mas per observing epoch). The resulting data set should be a promising means to assemble a large sample of brown dwarfs with accurate distances. 5. SUMMARY We have obtained high angular resolution imaging of the T3.5 dwarf SDSS J1534+1615 using the Keck sodium LGS AO system. SDSS J1534+1615 is resolved into a 0B11 binary, and its low proper motion and distinctive optical/IR colors indicate that it is a physical pair. The resolved near-IR colors give estimated spectral types of T1:5 ô 0:5 and T5:5 ô 0:5, consistent with modeling the integrated-light spectrum as the blend of two components. This analysis also produces an estimated distance of 36 pc, based on the distances to the mid-T dwarfs used in the modeling. The inferred (distance-independent) differences in the masses and surface gravities of the two components are about 20% and 0.1 dex, respectively. SDSS J1534+1615AB serves as a new benchmark for understanding the J-band brightening seen among the early/mid-T dwarfs. The binary's near-IR flux ratios show an inversion, wherein component A is brighter at the H and K bands but component B is brighter at the J band by about 0.2 mag. Thus, this binary system embodies the J-band brightening phenomenon seen in the CMD of T dwarfs. Since the components of SDSS J1534+1615 presumably have a common age, a common metallicity, and a very similar surface gravity, our results demonstrate that the J-band brightening is intrinsic to ultracool photospheres and not necessarily a by-product of observing an ensemble of brown dwarfs with different masses, ages, and /or metallicities. Likewise, the brightening mechanism must be able to operate over a relatively small range (%200 K ) in effective temperature. The magnitudes and colors of SDSS J1534+1615AB and the binary


No. 2, 2006

NOVEL T DWARF BINARY FROM LASER GUIDE STAR AO

1403

T dwarf Ind Bab are quite similar. However, Ind Bab, with resolved spectral types of T1 and T6, does not show a crossover in its near-IR flux ratios, suggesting that the J-band brightening may end by around spectral type T5. Based on the limited observations to date, we have considered the possibility that the large amplitude of the J-band brightening among early/mid-T dwarfs is caused by a high frequency of unresolved binaries. SDSS J1534+1615AB and Ind Bab provide examples of objects that appear to reside at /near the CMD location of the J-band brightening, but once their resolved properties are taken into account, the individual components are consistent with early T's and mid / late T's. We speculate that the possibly smaller scatter seen in the infrared CMDs and the slightly higher TeA 's inferred for the %T2YT4.5 dwarfs could be hints of unresolved binarity, leading to an enhancement in the apparent amplitude of the J-band brightening. In this interpretation, truly single %T2YT4.5 objects would be relatively rare, with most objects of these spectral subclasses being composed of the blended light of an earlier and a later type component. This hypothesis is in accord with the small TeA range inferred for the late L to mid T subclasses; photospheric dust removal seems to occur very rapidly during the L / T transition. Further study of SDSS J1534+1615AB will shed light on the nature of the L / T transition and the J-band brightening. A trigonometric parallax will accurately determine the CMD positions of the combined system and individual components. Resolved spectroscopy of the two components will allow for more accurate spectral classification. In addition, given the common metallicity and similar surface gravities of the two components, such data can isolate the temperature-sensitive and dust-sensitive features in the spectra of the early/mid-T dwarfs. Given the estimated $28 yr period, follow-up high angular resolution imaging over the next several years can trace a significant fraction of the orbital motion and lead to a dynamical mass estimate. Ultimately, a much larger sample of T dwarf binaries from HST, LGS AO, and IR radial velocities will provide the means to fully explore the physical processes encapsulated in the substellar colormagnitude diagram. We gratefully acknowledge the Keck LGS AO team for their impressive efforts in bringing the LGS AO system to fruition. It is a pleasure to thank Antonin Bouchez, David LeMignant, Marcos van Dam, Randy Campbell, Robert LaFon, Gary Punawai,

Peter Wizinowich, and the Keck Observatory staff for assistance with the observations; John Rayner, Robert Jedicke, Adam Burrows, Adam Burgasser, Neill Reid, and Marcos van Dam for productive discussions; and Dagny Looper for a careful reading of the manuscript. We thank Alan Stockton for a fortuitous swap of observing nights. Our research has benefited from the 2MASS data products; NASA's Astrophysical Data System; the SIMBAD database operated at CDS, Strasbourg, France; and the M, L, and T dwarf compendium19 maintained by Chris Gelino, Davy Kirkpatrick, and Adam Burgasser ( Kirkpatrick 2003; Gelino et al. 2004). M. L. acknowledges support for this work from NSF grant AST 05-07833 and a Sloan Research Fellowship. X. F. is supported by NSF grant AST 03-07384, a Sloan Research Fellowship, and a Packard Fellowship for Science and Engineering. Funding for the SDSS and SDSS-II has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, the US Department of Energy, the National Aeronautics and Space Administration, the Japanese Monbukagakusho, the Max Planck Society, and the Higher Education Funding Council for England. The SDSS20 is managed by the Astrophysical Research Consortium for the Participating Institutions. The Participating Institutions are the American Museum of Natural History, Astrophysical Institute Potsdam, University of Basel, Cambridge University, Case Western Reserve University, University of Chicago, Drexel University, Fermilab, the Institute for Advanced Study, the Japan Participation Group, Johns Hopkins University, the Joint Institute for Nuclear Astrophysics, the Kavli Institute for Particle Astrophysics and Cosmology, the Korean Scientist Group, the Chinese Academy of Sciences ( LAMOST ), Los Alamos National Laboratory, the Max-Planck-Institute for Astronomy ( MPIA), the Max-PlanckInstitute for Astrophysics ( MPA), New Mexico State University, Ohio State University, University of Pittsburgh, University of Portsmouth, Princeton University, the United States Naval Observatory, and the University of Washington. Finally, the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of Mauna Kea has always had within the indigenous Hawaiian community. We are most fortunate to have the opportunity to conduct observations from this mountain.

19 20

Housed at http://www.dwarfarchives.org. The SDSS Web site is at http://www.sdss.org.

REFERENCES Ackerman, A. S., & Marley, M. S. 2001, ApJ, 556, 872 Burrows, A., Sudarsky, D., & Hubeny, I. 2006, ApJ, 640, 1063 Baraffe, I., Chabrier, G., Barman, T. S., Allard, F., & Hauschildt, P. H. 2003, Chiu, K., Fan, X., Leggett, S. K., Golimowski, D. A., Zheng, W., Geballe, T. R., A&A, 402, 701 Schneider, D. P., & Brinkmann, J. 2006, AJ, 131, 2722 Becklin, E. E., & Zuckerman, B. 1988, Nature, 336, 656 Cruz, K. L., Burgasser, A. J., Reid, I. N., & Liebert, J. 2004, ApJ, 604, L61 Bouchez, A. H., et al. 2004, Proc. SPIE, 5490, 321 Cushing, M. C., Rayner, J. T., & Vacca, W. D. 2005, ApJ, 623, 1115 Bouy, H., Brandner, W., Martin, E. L., Delfosse, X., Allard, F., & Basri, G. ´ Dahn, C. C., et al. 2002, AJ, 124, 1170 2003, AJ, 126, 1526 Diolaiti, E., Bendinelli, O., Bonaccini, D., Close, L., Currie, D., & Parmeggiani, G. Burgasser, A. J., Geballe, T. R., Leggett, S. K., Kirkpatrick, J. D., & Golimowski, 2000, A&AS, 147, 335 D. A. 2006, ApJ, 637, 1067 Enoch, M. L., Brown, M. E., & Burgasser, A. J. 2003, AJ, 126, 1006 Burgasser, A. J., Kirkpatrick, J. D., Liebert, J., & Burrows, A. 2003a, ApJ, 594, 510 Fischer, D. A., & Marcy, G. W. 1992, ApJ, 396, 178 Burgasser, A. J., Kirkpatrick, J. D., & Lowrance, P. J. 2005a, AJ, 129, 2849 Freed, M., Close, L. M., & Siegler, N. 2003, ApJ, 584, 453 Burgasser, A. J., Kirkpatrick, J. D., Reid, I. N., Brown, M. E., Miskey, C. L., & Geballe, T., et al. 2002, ApJ, 564, 466 Gizis, J. E. 2003b, ApJ, 586, 512 Gelino, C. R., Kirkpatrick, J. D., & Burgasser, A. J. 2004, AAS Meeting, 205, 11.13 Burgasser, A. J., Marley, M. S., Ackerman, A. S., Saumon, D., Lodders, K., Gizis, J. E., Reid, I. N., Knapp, G. R., Liebert, J., Kirkpatrick, J. D., Koerner, Dahn, C. C., Harris, H. C., & Kirkpatrick, J. D. 2002a, ApJ, 571, L151 D. W., & Burgasser, A. J. 2003, AJ, 125, 3302 Burgasser, A. J., Reid, I. N., Leggett, S. K., Kirkpatrick, J. D., Liebert, J., & Golimowski, D. A., et al. 2004a, AJ, 127, 3516 Burrows, A. 2005b, ApJ, 634, L177 ------. 2004b, AJ, 128, 1733 Burgasser, A. J., et al. 2000, ApJ, 531, L57 Kaiser, N., et al. 2002, Proc. SPIE, 4836, 154 ------. 2002b, ApJ, 564, 421 Kirkpatrick, J. D. 2003, in Proc. IAU Symp. 211, Brown Dwarfs, ed. E. Marti ´n Burrows, A., Hubbard, W. B., Lunine, J. I., & Liebert, J. 2001, Rev. Mod. (San Francisco: ASP), 189 Phys., 73, 719 ------. 2005, ARA&A, 43, 195


1404

LIU ET AL.
Nakajima, T., Oppenheimer, B. R., Kulkarni, S. R., Golimowski, D. A., Matthews, K., & Durrance, S. T. 1995, Nature, 378, 463 Nakajima, T., Tsuji, T., & Yanagisawa, K. 2001, ApJ, 561, L119 ------. 2004, ApJ, 607, 499 Perryman, M. A. C., et al. 1997, A&A, 323, L49 Potter, D., Martin, E. L., Cushing, M. C., Baudoz, P., Brandner, W., Guyon, O., ´ & Neuhauser, R. 2002, ApJ, 567, L133 ¨ Rebolo, R., Zapatero Osorio, M. R., Madruga, S., Bejar, V. J. S., Arribas, S., & Licandro, J. 1998, Science, 282, 1309 Reid, I. N., Gizis, J. E., Kirkpatrick, J. D., & Koerner, D. W. 2001, AJ, 121, 489 Reid, I. N., Lewitus, E., Burgasser, A. J., & Cruz, K. L. 2006, ApJ, 639, 1114 Scholz, R.-D., McCaughrean, M. J., Lodieu, N., & Kuhlbrodt, B. 2003, A&A, 398, L29 Simons, D. A., & Tokunaga, A. 2002, PASP, 114, 169 Thorstensen, J. R., & Kirkpatrick, J. D. 2003, PASP, 115, 1207 Tinney, C. G., Burgasser, A. J., & Kirkpatrick, J. D. 2003, AJ, 126, 975 Tokunaga, A. T., Simons, D. A., & Vacca, W. D. 2002, PASP, 114, 180 Torres, G. 1999, PASP, 111, 169 Tsuji, T. 2002, ApJ, 575, 264 ------. 2005, ApJ, 621, 1033 Tsuji, T., & Nakajima, T. 2003, ApJ, 585, L151 Vrba, F. J., et al. 2004, AJ, 127, 2948 Wizinowich, P. L., et al. 2004, Proc. SPIE, 5490, 1

Kirkpatrick, J. D., Dahn, C. C., Monet, D. G., Reid, I. N., Gizis, J. E., Liebert, J., & Burgasser, A. J. 2001, AJ, 121, 3235 Kirkpatrick, J. D., Reid, I. N., Liebert, J., Gizis, J. E., Burgasser, A. J., Monet, D. G., Dahn, C. C., Nelson, B., & Williams, R. J. 2000, AJ, 120, 447 Knapp, G. R., et al. 2004, AJ, 127, 3553 Koen, C., Matsunaga, N., & Menzies, J. 2004, MNRAS, 354, 466 ´ Koen, C., Tanabe, T., Tamura, M., & Kusakabe, N. 2005, MNRAS, 362, 727 Koerner, D. W., Kirkpatrick, J. D., McElwain, M. W., & Bonaventura, N. R. 1999, ApJ, 526, L25 Leggett, S. K., Allard, F., Burgasser, A. J., Jones, H. R. A., Marley, M. S., & Tsuji, T. 2005, in Proc. 13th Cool Stars Workshop, ed. F. Favata et al. ( ESA SP-560; Noorwijk: ESA), 143 Leggett, S. K., et al. 2002, ApJ, 564, 452 Liu, M. C., Fischer, D. A., Graham, J. R., Lloyd, J. P., Marcy, G. W., & Butler, R. P. 2002, ApJ, 571, 519 Liu, M. C., & Leggett, S. K. 2005, ApJ, 634, 616 Marley, M. S., Seager, S., Saumon, D., Lodders, K., Ackerman, A. S., Freedman, R. S., & Fan, X. 2002, ApJ, 568, 335 Martin, E. L., Brandner, W., & Basri, G. 1999, Science, 283, 1718 ´ Maxted, P. F. L., & Jeffries, R. D. 2005, MNRAS, 362, L45 McCaughrean, M. J., Close, L. M., Scholz, R.-D., Lenzen, R., Biller, B., Brandner, W., Hartung, M., & Lodieu, N. 2004, A&A, 413, 1029 Metchev, S. A., & Hillenbrand, L. A. 2004, ApJ, 617, 1330 Monet, D. G., et al. 2003, AJ, 125, 984

Note added in proof.-- A. J. Burgasser et al. (ApJ, in press [2006]) report that the T3 dwarf SDSS 1021þ0304 is a 0 B17 binary based on HST near-IR imaging, with estimated spectral types of T1 and T5. The observed flux ratios combined with modeling of the integratedlight spectrum show that this binary also possesses the J-band flux ratio inversion between its two components, as in the case for SDSS 1534+1615AB. Burgasser et al. independently reach conclusions similar to those of our paper about the significance of binarity in the L / T transition.