Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.arcetri.astro.it/~lt/preprints/ldwarfs_amici/ld_amici.pdf Äàòà èçìåíåíèÿ: Tue Sep 11 16:56:54 2007 Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 08:23:21 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: m 5

The Astrophysical Journal, 552:L147­L150, 2001 May 10
2001. The American Astronomical Society. All rights reser ved. Printed in U.S.A.

NICS-TNG LOW-RESOLUTION 0.85­2.45 MICRON SPECTRA OF L DWARFS: A NEAR-INFRARED SPECTRAL CLASSIFICATION SCHEME FOR FAINT DWARFS L. Testi,1 F. D'Antona,2 F. Ghinassi,3 J. Licandro,3 A. Magazzu 3,4 R. Maiolino,1 `, F. Mannucci,5 A. Marconi,1 N. Nagar,1 A. Natta,1 and E. Oliva1,3
Received 2001 March 19; accepted 2001 April 4; published 2001 May 1

ABSTRACT We present complete near-infrared (0.85­2.45 mm), low-resolution (100) spectra of a sample of 26 disk L dwarfs with reliable optical spectral type classification. The obser vations have been obtained with the Near Infrared Camera and Spectrograph at the Telescopio Nazionale Galileo using a prism-based optical element (the Amici device) that provides a complete spectrum of the source on the detector. Our obser vations show that lowresolution near-infrared spectroscopy can be used to determine the spectral classification of L dwarfs in a fast but accurate way. We present a librar y of spectra that can be used as templates for spectral classification of faint dwarfs. We also discuss a set of near-infrared spectral indices well correlated with the optical spectral types that can be used to classify accurately L dwarfs earlier than L6. Subject headings: infrared: stars -- stars: atmospheres -- stars: fundamental parameters -- stars: low-mass, brown dwarfs
1. INTRODUCTION

The latest years have witnessed the discover y of numerous brown dwarfs close to the Sun, in nearby clusters and associations, and in binaries. The strategy of the optical and nearinfrared imaging sur veys (Two Micron All-Sky Sur vey [2MASS]; Kirkpatrick et al. 1999, 2000; the Sloan Digital Sky Sur vey; Fan et al. 2000; Deep Near-Infrared Sur vey [DENIS]; Delfosse et al. 1997; Tinney et al. 1998) has been so successful that two new spectral classes (L and T) have been added to the previous types, to help to classify ver y cool stellar objects. For L dwarfs, in spite of the remaining uncertainties in model atmospheres for such cool objects (e.g., Leggett et al. 2001), it has been possible to derive a detailed spectral classification system in nine subclasses from the systematic changes obser ved in selected spectral features (Kirkpatrick et al. 1999; Marti et ´n al. 1999). This spectral classification has been developed in the red part of the optical spectrum: the beginning of the L type is set by the weakening of the TiO and VO bands, while the appearance of the CH4 bands signals the transition to the T type. However, the optical spectral confirmation and classification of a candidate DENIS or 2MASS L dwarf requires up to 1 hr of integration time with a low-resolution (1000) optical spectrograph at a large (10 m­class) telescope, depending on the spectral type and magnitude of the candidate. This prevents the applicability of the optical classification to deeper sur veys. Given that L dwarfs emit most of their radiation in the nearinfrared bands from 1 to 2.5 mm, the advantage of longer wavelengths is obvious. At present, near-infrared spectra are available for a handful of objects (14), and ver y recently Reid et al. (2001) attempted to establish a near-infrared classification
1 Osser vatorio Astrofisico di Arcetri, Largo Enrico Fermi 5, I-50125 Firenze, Italy; ltesti@arcetri.astro.it. 2 Osser vatorio Astronomico di Roma, via Frascati 33, I-00044, Roma, Italy. 3 Centro Galileo Galilei and Telescopio Nazionale Galileo, P.O. Box 565, E-38700, Santa Cruz de La Palma, Spain. 4 Osser vatorio Astrofisico di Catania, via S. Sofia 78, I-95123 Catania, Italy. 5 Centro per l'Astronomia Infrarossa e lo Studio del Mezzo Interstellare­Consiglio Nazionale delle Ricerche, Largo Enrico Fermi 5, I-50125 Firenze, Italy.

scheme from full (0.9­2.5 mm) United Kingdom Infrared Telescope spectra with resolution 500­1000; each spectrum required integration times between 1 and 4 hr, depending on the spectral type of the star. With a 4 m­class telescope the time demand is comparable to (or higher than) that required for the optical classification and thus prohibitive for large sur veys. It is clear that intermediate- and high-resolution spectroscopy, while necessar y for investigating photospheric properties of selected objects, is not suitable for candidate confirmation and classification of large, deep sur veys. In this Letter we present low-resolution (100) near-infrared spectra of a sample of 26 L dwarfs with reliable optical spectral classification from Kirkpatrick et al. (2000). The spectra have been obtained with a prism-based optical element (the Amici device), which provided a complete near-infrared spectrum of each star in less than 15 minutes on source at the Italian Telescopio Nazionale Galileo (TNG), a 3.56 m telescope. We describe the obser vations in § 2 and show in § 3 that lowresolution near-infrared spectroscopy can be used to determine the spectral classification of L dwarfs in a fast but accurate way. The potential of such an obser ving mode is discussed in § 4, which concludes the Letter.
2. OBSERVATIONS AND RESULTS

The obser vational data were collected at the 3.56 m TNG with the Near Infrared Camera and Spectrograph (NICS), a cr yogenic focal reducer designed as a near-infrared commonuser instrument for that telescope. The instrument is equipped with a Rockwell 10242 HgCdTe Astronomical Wide Area Infrared Imager array detector. Among the many imaging and spectroscopic obser ving modes (Baffa et al. 2001), NICS offers a unique, high-throughput, ver y low resolution mode with an approximately constant resolving power of 50, when the 1 wide slit is used. In this mode a prism-based optical element, the Amici device, is used to obtain on the detector a complete 0.85­2.45 mm long-slit spectrum of the astronomical source (Oliva 2001). The 26 L dwarfs in our sample cover in an approximately uniform way the optically defined spectral types ranging from L0 to L8. All the selected sources are brighter than K s L147


L148

NEAR-INFRARED SPECTRAL CLASSIFICATION OF L DWARFS

Vol. 552

Fig. 1.--0.85­2.45 mm low-resolution near-infrared spectra for all the L dwarfs in our sample. All spectra have been normalized by the average flux between 1.235 and 1.305 mm, and a constant shift has been added to each to separate them vertically. Each spectrum is labeled with the 2MASS name (the 2MASSJ prefix has been omitted) and the optical spectral type from Kirkpatrick et al. (2000). All spectra are available in electronic form upon request from the authors.

14.4, with three exceptions with K s p 14.5­14.8. The sources were obser ved during the commissioning of NICS in several obser ving runs from 2000 December to 2001 Februar y. We used the 0 . 5 wide slit, and the resulting spectra have an effective resolution of 100 across the entire spectral range. Integration times on source varied from 4 to 15 minutes depending on the source brightness. Wavelength calibration was performed using an argon lamp and the deep telluric absorption features. The telluric absorption was then removed by dividing each of the object spectra by an A0 reference star spectrum obser ved at similar air mass, and normalized using a synthetic A0 star spectrum smoothed to the appropriate resolution. Four of the targets, which are also among the fainter in our sample, were obser ved in unfavorable weather conditions, resulting in a poor

compensation of the deep atmospheric features and noisier spectra. The accuracy of the spectral shapes was checked by computing the expected 2MASS colors from our spectra. When normalized at H band, our synthesized and the 2MASS fluxes at J and Ks differ by less than 2 j in all but three cases, where one of the two bands is more discrepant. The final spectra are shown in Figure 1. The objects are shown from top to bottom and from left to right in order of increasing optical spectral type from L0 to L8. The spectra have been normalized by the average flux in the 1.235­1.305 mm region; a constant offset has been added to each one to avoid overlap. The spectra show the same general features described in Leggett et al. (2001) and Reid et al. (2001). In our low-resolution spectra the atomic lines of Na i and K i and the FeH lines in


No. 2, 2001

TESTI ET AL.

L149

Fig. 2.--Correlations between optical spectral types and the near-infrared spectral indices. The top three panels show the H2OA, H2OB, and K1 indices calculated for the stars in our sample; the dotted lines show the linear fits of Reid et al. (2001). The bottom six panels show the new indices defined in Table 1; the dashed lines depict the linear fits described in the text. The four sources with poor telluric correction spectra are shown as crosses.

the J band are not resolved, although their blended absorption features are clearly seen in the early-type dwarfs. The spectra are dominated by the H2O features at 0.95, 1.15, 1.40, 1.85, and 2.4 mm. TiO, near 0.85 mm, FeH, at 1.0 mm, and CO, longward of 2.3 mm, are visible in some of the spectra, depending on spectral type and signal-to-noise ratio.
3. THE NEAR-INFRARED CLASSIFICATION SCHEME

Despite their low resolution, the spectra of Figure 1 allow us to identify a set of spectral indices that can be used to define a near-infrared spectral classification scheme that is well correlated with the widely used optical classification scheme of Kirkpatrick et al. (1999) and Martin et al. (1999). A first attempt ´ in this direction has already been taken by Reid et al. (2001). The main conclusion of their study is that, while the J-band atomic lines are only weakly correlated with the optical spectral types, it is possible to define indices based on the H2O wings that are well correlated with the optical types (at least up to L6). For all the stars in our sample, we computed the three indices that Reid et al. (2001) found to be the most correlated with the optical spectral type: K1 (see also Tokunaga & Kobayashi 1999), H2OA, and H2OB, all related to the strength or slope of the water absorption features. In the top panels of Figure 2 we show the data points from our sample compared

with the fits reported by Reid et al. (2001); our spectra are generally consistent with their fits. Note that, as in Reid et al. (2001), the K1 index can be used only for types earlier than L5; moreover, our data indicate saturation at late spectral types also for H2OB, while H2OA shows a ver y large scatter. It is possible that this behavior of the H2OA and H2OB indices may be caused by the lower resolution of our spectra. We also computed six additional indices that are best suited for low-resolution, complete spectra. The new indices are defined in Table 1 in a similar way as the Tokunaga & Kobayashi (1999) indices. Two of the new indices (sHJ and sKJ) are based on the slope of the continuum and can be reliably defined using our spectra because the entire spectral range is obser ved simultaneously in the same atmospheric conditions, without the need of a problematic intercalibration of various spectral segments. All the other indices measure the slope of the water line wings. They have been defined so as to avoid as much as possible the spectral regions affected by the worse telluric absorption. To illustrate this point, in Figure 3 we show the relative system efficiency (including atmosphere), two of the spectra of Figure 1, representative of the extreme classes (L0.5 and L8), and the spectral regions used to define the various indices shaded in gray. In Figure 2 the values of all six indices are plotted against the optical spectral type of each star. The sH2OJ index is a mea-


L150

NEAR-INFRARED SPECTRAL CLASSIFICATION OF L DWARFS

Vol. 552

TABLE 1 Spectral Index Definitions and Fits to Optical Spectral Types Index sHJ ......... sKJ ......... sH2OJ ....... sH2O sH2O
H1

Definition AF1.265 1.305S AF1.60 1.70S 0.5(AF1.265 1.305S AF1.60 1.70S) AF1.265 1.305S AF2.12 2.16S 0.5(AF1.265 1.305S AF2.12 2.16S) AF1.265 1.305S AF1.09 1.13S 0.5(AF1.265 1.305S AF1.09 1.13S) AF1.60 1.70S AF1.45 1.48S 0.5(AF1.60 1.70S AF1.45 1.48S) AF1.60 1.70S AF1.77 1.81S 0.5(AF1.60 1.70S AF1.77 1.81S) AF2.12 2.16S AF1.96 1.99S 0.5(AF2.12 2.16S AF1.96 1.99S)

Fit to Optical Spectral Type Sp p Sp p Sp p Sp p Sp p Sp p 1.87sHJ 1.20sKJ 1.54sH2OJ 1.27sH2OH1 2.11sH2OH2 2.36sH2OK 1.67 2.01 0.98 0.76 0.29 0.60 Fig. 3.--Top: Relative efficiency of the system, including atmosphere. Bottom: Spectra of I0746425 200032 (L0.5) and W0310599 164816 (L8). Shaded in gray are the regions used to define the new indices of Table 1; dark gray regions correspond to the regions where the indices sample the water wings. All indices avoid the worse telluric absorption regions.

...... ......

H2

sH2OK .......

Note.--AFli ljS is the average flux in the range li­lj.

sure of the strength of the water absorption feature at 1.1 mm, and, although it shows a ver y nice correlation with the optical spectral type in our data, it should be used with care as it may be seriously affected by a poor correction of the telluric absorption. With only few exceptions, all stars with good spectra show a tight correlation between the newly defined spectral indices and the optical spectral type. In Table 1 we also show the linear relation between the optical spectral types and the index values; the spectral types are coded as follows: L0 { 1.0, L8 { 1.8, with 0.1 step per subclass. We did not attempt to fit again the indices of Reid et al. (2001). For the spectral range L0­L6 the linear fits offer a classification accurate to approximately one-half a subclass; at later spectral types most of the water indices saturate and the classification based on the fits is not as accurate and a direct comparison with the spectral librar y of Figure 1 is preferable. It is interesting to note that the good correlation of our "narrowband" continuum indices (sJH and sJK) with the spectral type disappears when the broadband colors are used (Kirkpatrick et al. 2000). This can be understood in terms of the competing effects of the reddening of the continuum at later spectral types and the increasing absorption from water features; the latter mostly affects the H and Ks broad bands.
4. CONCLUSIONS

magnitudes of potential candidates make other techniques prohibitive, even at large telescopes. As an example, we estimate that the Ver y Large Telescope next-generation, low-resolution near-infrared spectrograph will allow one to measure in 1 hr the 0.9­1.7 mm spectrum of faint (J 1 21) dwarfs and classify them using our sHJ, sH2OJ, and sH2OH1 indices. Finally, we want to stress the advantage of using a device that produces the complete near-infrared spectrum in one shot, without the need for intercalibration of various spectral segments obtained in var ying atmospheric conditions. This makes it possible to measure spectral indices that use the shape of the continuum, rather than the lines. These continuum indices (sHJ and sKJ) turn out to be ver y sensitive and reliable tools for the spectral classification of cool objects. Our results suggest that a rough but ver y efficient spectral classification could be obtained by narrowband imaging through filters corresponding to the flux ranges used in the definitions of sHJ and sKJ in Table 1. This Letter is based on obser vations made with the Italian TNG operated on the island of La Palma by the Centro Galileo Galilei of the CNAA (Consorzio Nazionale per l'Astronomia e l'Astrofisica) at the Spanish Obser vatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias. Support ´ from ASI grants ARS-99-15 and 1/R/27/00 to the Osser vatorio di Arcetri is gratefully acknowledged. It is a pleasure to thank the Arcetri and TNG technical staff and the TNG operators for their assistance during the commissioning of NICS.

We have presented a librar y of complete 0.85­2.45 mm lowresolution (100) spectra of 26 disk L dwarfs. This kind of spectral librar y and the spectral indices we have defined provide a unique tool for the identification and spectral classification of L dwarfs from large, deep sur veys, where the number and

REFERENCES Baffa, C., et al. 2001, A&A, submitted Delfosse, X., et al. 1997, A&A, 327, L25 Fan, X., et al. 2000, AJ, 119, 928 Kirkpatrick, J. D., et al. 1999, ApJ, 519, 802 ------. 2000, AJ, 120, 447 Leggett, S. K., Allard, F., Geballe, T. R., Hauschildt, P. H., & Schweitzer, A. 2001, ApJ, 548, 908 Martin, E. L., Delfosse, X., Basri, G., Goldman, B., For veille, Th., & Zapaterio ´ Osorio, M. R. 1999, AJ, 118, 2466 Oliva, E. 2001, Mem. Soc. Astron. Italiana, in press (astro-ph/9909108) Reid, I. N., Burgasser, A. J., Cruz, K. L., Kirkpatrick, J. D., & Gizis, J. E. 2001, AJ, 121, 1710 Tinney, C. G., Delfosse, X., For veille, T., & Allard, F. 1998, A&A, 338, 1066 Tokunaga, A., & Kobayashi, N. 1999, AJ, 117, 1010