MEETING THE COOL NEIGHBORS. X. ULTRACOOL DWARFS FROM THE 2MASS ALL-SKY DATA RELEASE

The primary criteria used to define the 2MU2 and 2MUA samples are tied to the NIR photometric properties and Galactic location. Drawing from the experience gained in compiling follow-up observations of the 2MU2 sample, we have modified these selection criteria in certain important respects.

• 1.  
  We raised the Galactic latitude criterion from |b|>10° to |b|>15°. Regions near the Galactic Plane suffer from two major problems for our type of survey: high source density, leading to incompleteness and photometric inaccuracies due to image crowding, and extensive reddening due to interstellar dust. To minimize the effects on the 2MU2 sample, we excluded all 2MASS IDR2 tiles that are centered at Galactic latitudes |b| < 10°. This had a relatively small impact on areal coverage, since the 2MASS IDR2 release covered predominantly high galactic latitudes. Nonetheless, the 2MU2 ultracool candidates included a significant number of reddened sources. With the higher latitude limit adopted in the present analysis, the majority of those sources are eliminated a priori. The |b|>15° requirement reduces coverage to ∼70% of the celestial sphere.
• 2.  
  We increased the blue (J − K_S) limit at bright magnitudes from (J − K_S)>1.00 to (J − K_S) ⩾ 1.06. The 2MU2 candidate list includes several hundred sources with J > 12 and (J − K_S) ⩽ 1.05 (Figure 4 in Paper V), almost all of which have proven to be M6/M6.5 dwarfs at distances of 30–50 pc. Eliminating the M6 dwarfs comes at a price: with the redder color limit, the 2MUA sample includes few M7 and only a subset of nearby M8 dwarfs. Based on the spectral-type/color distributions derived by Gizis et al. (2000), we expect approximately 50% of M8 dwarfs to meet the current color limits.
• 3.  
  We considered only sources with magnitudes J > 9. Five hundred and eighty-eight of the 2MU2 candidates are brighter than this limit, but only four of these sources proved to be mid- or late-type M dwarfs.

Other selection criteria outlined in Paper V, based on location in the JHK_S (including the "giant star" criteria defined in Equation (4) of that paper), on location in the J/(R − J) diagrams, and on the 2MASS photometric confusion/solar system flags (cc_flg = 000, mp_fig = 00), were retained unaltered.

The upper section of Table 1 (corresponding to Table 1 in Paper V) breaks down the steps used to construct the 2MUA candidate list. As with the 2MU2 sample, the initial catalog of 2.14 million 2MASS sources with (J − K_S) ⩾ 1.06 and latitude |b|>15° is reduced to manageable proportions, primarily through cuts in the (J, (J − K_S)) and (J − H)/(H − K_S) planes, and the elimination of sources in highly-crowded fields near the |b| = 15° cutoff or near known star-forming regions. As discussed in Paper V, rough positions and dimensions for highly-reddened regions were taken from Dame et al. (1987) and Dutra & Bica (2002), and enlarged as necessary where visual inspection revealed high source densities around the edges of the excised regions. Sources eliminated based on this criterion are designated as "clouds/crowded" in Table 1.

The 2MASS All Sky Point Source Catalogue includes the same 48% of the sky covered by the 2MASS IDR2. Our prior analyses, described in Papers V and IX, resulted in the identification of 1672 ultracool candidates within those regions (Paper V, Table 1), and follow-up observations have confirmed 369 to be ultracool dwarfs. Eliminating the 2MU2 sources from the all-sky sample gives a total of 1018 candidate ultracool dwarfs in the 2MUA sample. Figure 1 plots the (α, δ) and (l, b) distributions of the 2MU2 and 2MUA candidates. Combined, the two data sets cover approximately 26,500 deg² or ∼65% of the sky.

Zoom In Zoom Out Reset image size

2MASS and Digitized Sky Survey (DSS) images of 1018 ultracool candidates in the 2MUA sample were inspected individually, and the results from those inspections are given in Table 1. Our inspection showed that almost half the sample proved to be artifacts, mainly diffraction spikes and blends with background nebulosity. (The 2MASS catalogue includes a number of flags to identify sources of dubious image quality, and almost all of these sources are indeed flagged as suspect.) A further 26 sources were eliminated since they lie near small star-forming regions ("clouds" in Table 1), and a similar number were disqualified because they lie within 15° of the Galactic equator (the phase I cuts are based on the Galactic latitude of the center of the 2MASS tile, not the positions of the individual objects in each tile). Thirty-one objects have optical/IR colors (based on DSS data) that are obviously inconsistent with ultracool dwarfs, and, finally, eight objects are artifacts associated with bright (H-D catalogue) stars. Removing those sources reduces the candidate list to 467 sources.

Over 70 sources in our candidate list have been observed in the course of other surveys for ultracool dwarfs. Fifty-four stars (and brown dwarfs) have extant optical spectroscopy of sufficient signal-to-noise (S/N) and resolution to allow unambiguous classification as late-type dwarfs; in a few cases, we have supplemented the literature data with our own observations. The relevant characteristics of these objects are given in Table 2. Most were identified from follow-up observations of extremely red sources from the 2MASS survey (K00; Gizis et al. 2000; Gizis 2002), the DENIS survey (Martín et al. 1999; Phan-Bao et al. 2001), and the SDSS survey (Fan et al. 2000; Hawley et al. 2002). Twenty-one dwarfs listed in Table 2 have trigonometric or spectroscopic parallaxes that indicate distances within 20 pc of the Sun.

We have obtained intermediate-resolution optical spectroscopy of 376 sources from the 2MUA sample. The overwhelming majority of the observations, covering some 355 sources, were obtained in the course of several observing runs between 2003 March and 2004 February. We used the Ritchey-Chrétien (RC) spectrograph on the Kitt Keak National Observatory 2.1 m telescope in 2003 March and October; the MARS spectrograph on the KPNO 4 m telescope in 2003 July and 2004 February; the RC spectrograph on the 1.5 m telescope at Cerro-Tololo Interamerica Observatory in 2003 May and November; and the RC spectrograph on the CTIO 4 m telescope in 2003 April, 2004 August, and 2006 January. In each case, the spectra cover the wavelength range 6300–10000 Å at a resolution of ∼7 Å.

Twenty-one fainter candidates were observed using the Gemini telescopes. The Gemini Multi-Object Spectrometer (GMOS; Hook et al. 2004) was used on Gemini north (GN) and Gemini south (GS) during queue observations taken between 2004 August and 2005 November (Program IDs: GN-2004B-Q-10, GS-2004B-Q-30, GN-2005B-Q-20, GS-2005B-Q-21). The observations were made using the RG610_G0307 filter and R400_G5305 disperser on GN, while the RG610_G0331 filter and R400_G5325 disperser were used on GS. In both cases, the data cover the wavelength range 6000–10000 Å. Two consecutive observations, with the central wavelength offset, were taken of each target to provide a complete wavelength coverage. On both telescopes, the nod and shuffle mode was used with a 075-wide slit to provide good sky subtraction and a resolution of 5.5 Å (4 pixels).

The spectroscopic data acquired from the KPNO and CTIO telescopes were bias-subtracted and flat-fielded using the IRAF CCDRED package, and the spectra extracted, wavelength- and flux-calibrated using standard techniques. The wavelength calibration is based on a single HeNeAr arc, usually taken at the start of the night. Each night we also observed one of the following flux standards: BD+26 2606, BD+17 4708, HD 19445 (from Oke & Gunn 1983); Feige 56, Feige 110, or Hiltner 600 (from Hamuy et al. 1994).

The Gemini GMOS package was used to reduce the data from GN and GS.⁹ Nod and shuffle dark frames were subtracted and the data were flat-fielded using the gsreduce task and the sky lines were subtracted using gnsskysub. Flux calibration was provided through observations of the flux standards G191B2B, LTT 1020 and EG 21 (Massey et al. 1988; Massey & Gronwall 1990; Hamuy et al. 1994). All spectra were extracted using gsextract and the flux calibration applied with calibrate. As discussed in Cruz et al. (2007), the slope of the spectra from GN-2004B-Q-10 is systematically too steep longward of 8700 Å. None of the spectra have been corrected for telluric absorption.

The 2MASS sources targeted in these observations are listed in Tables 3–6, where the coordinates and near-infrared photometry are from the 2MASS All Sky Point Source Catalogue, and the results deduced from the observations. Table 3 lists data for 44 sources that we identify as ultracool dwarfs likely to lie within 20 pc of the Sun; Table 4 presents data for 228 ultracool dwarfs at larger distances; Table 5 lists 83 spectroscopically confirmed K and M giants; and Table 6 catalogs data for 22 carbon stars. Combining literature data and our own observations, we have optical spectra for 430 of the 467 ultracool candidates. Twenty-eight of the remaining 37 sources have been observed spectroscopically at near-infrared wavelengths, and nine sources have no follow-up observations. Twenty-five of the 28 sources with observations have spectra consistent with ultracool dwarfs lying more than 20 pc from the Sun. Those infrared observations will be discussed in detail in a future paper in this series.

We have applied the methods described in Papers V and IX to determine spectral types, and hence absolute magnitude and distance estimates, for the dwarfs in the 2MUA sample. As discussed in those papers, although molecular band strengths are well correlated with luminosity for early- and mid-type M dwarfs, there are ambiguities for later-type dwarfs. We therefore determine spectral types for the latter dwarfs from the overall spectral energy distribution from 6000 to 10000 Å, using side-by-side comparison with spectral standards. The uncertainties are generally ±0.5 subtypes for well-exposed spectra, rising to ±1 − 2 subtypes for low S/N data. In cases where we have multiple observations of a particular candidate, we have used the highest S/N spectrum to estimate the spectral type.

Absolute magnitudes of late-type dwarfs (>M6) are derived directly from the spectral types using the calibration given in Paper V. For earlier-type dwarfs, we derive distances using the TiO5, CaH2, and CaOH band strengths and the relations listed in Paper III. In both cases, the calibrations are tied to the 2MASS J passband. The results are collected in Tables 3 and 4. Table 3 presents observations of 44 ultracool dwarfs (spectral types M7 and later) with formal distances less than 20 pc. Table 4 lists data for a further 228 dwarfs that lie beyond our distance limit, including 84 L dwarfs and 132 ultracool M dwarfs. Many late-type M dwarfs and a handful of L dwarfs exhibit Hα emission. Schmidt et al (2007) present a thorough analysis of chromospheric activity in these low-mass dwarfs and also discuss the proper motions and correlations between activity and kinematics. Detailed discussions of individual objects of interest are given in Section 4.

As in our previous spectroscopy of 2MU2 ultracool candidates, a number of late-type dwarfs exhibit anomalously strong VO absorption and/or weaker K i and Na i atomic absorption. This is generally interpreted as evidence for surface gravities that are lower than the typical values for field dwarfs of similar spectral types (Kirkpatrick et al. 2006). Twenty-seven candidate low-gravity dwarfs are identified in Tables 2–4. We have assigned these objects spectral types and, where necessary, spectroscopic parallaxes using conventional criteria; however, if the systems prove to be young, low-gravity dwarfs, it is likely that both spectral types and distances will require revision. Consequently, the spectral types for these objects are listed in parentheses in Tables 2–4. The full characteristics of these candidates' low-gravity dwarfs will be discussed in more detail by K. L. Cruz et al. (2009, in preparation).

Finally, late-type first giant branch and asymptotic giant branch stars can have near-infrared colors that meet our selection criteria, and our follow-up spectroscopic observations have identified a number of such stars. Tables 5 and 6 list data for 83 K and M giants and 22 carbon stars, respectively. A number of carbon-rich dwarfs have been discovered through follow-up observations of 2MASS ultracool candidates (e.g., Lowrance et al. 2003). However, all of the stars listed in Table 6 have classical carbon giant spectra, and none show evidence for significant proper motion. Consequently, it is likely that all are giants rather than nearby dwarfs.

Spectral classification is a comparative technique, where the overall appearance of a program object is matched against a set of reference calibrators. It is therefore important to have well-defined standard objects. This is particularly the case for ultracool dwarfs, where the spectral type appears to be the empirical parameter that is linked most closely to physical characteristics, such as luminosity and temperature.

Ideally, spectral standards should be bright objects that are accessible to even moderate-aperture telescopes. The primary L-dwarf spectral standards are specified by Kirkpatrick et al. in their definition of spectral class L (Table 6 of K99). At that juncture, only ∼25 L dwarfs were known, and, with only a limited parent sample, the later-type standards are relatively faint. Moreover, several of the brightest standards have proven to be close binaries. This is not unexpected, given that this initial set was drawn from a magnitude-limited sample.

Spectroscopic observations have now been obtained for more than 500 L dwarfs, including some that are significantly brighter (in apparent magnitude) than the primary standards in the initial sample. In particular, the present survey, which concentrates on the nearest (and therefore the brightest) L dwarfs, provides an excellent resource for supplementing the reference set of primary standards. All of these observations are cataloged in the online L-dwarf database maintained at http://DwarfArchives.org.

We have selected supplemental spectral standards based on three criteria: apparent brightness, the absence of a known close companion, and spectral morphology. We have not given consideration to the declination of the source (i.e., accessibility from northern and southern ground-based observatories). All bright (J ≲ 14) objects of each subclass that are currently not known to be binary were considered initially. The candidate standards were matched against the original standards through overplotting the spectra, and by comparing the four spectral indices (CrH-a, Rb-b/TiO-b, Cs-a/Vo-b, and color-d) used for spectral typing in K99. Indices for the original standards and new candidates were measured using the same script; our measurements reproduce the values reported in K99 for the original standards. Table 7 catalogs both the original standards and the new objects that best match the original classification scheme, both quantitatively (via spectral indices) and qualitatively (via overplotting). Figure 2 shows how the spectral indices measured for the new standards compare with the primary sequence; and Figure 3 directly compares the far-red optical spectra of the primary and supplementary standards.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We have not been able to identify any completely acceptable L6- or L7-type supplemental standards from the present observational data set. In order to confidently choose a spectral standard, a spectrum of fairly high S/N is required. These late-type L dwarfs are of low luminosity and few objects in our library have spectra of sufficient quality to enable a reliable comparison with the original standard. (Note that we are fortunate that the new L8 standard, with a distance (just) less than 5 pc, is one of the closest brown dwarfs known.) However, we identify 2MASS J15150083+4847416 and 2MASS J09083803+5032088 as potential L6 and L7 standards, respectively. Higher S/N data than our current observations are required before those dwarfs can formally be confirmed as secondary standards

All of the new standards except LSR J0602+3910 have been imaged with NICMOS as part of our search for low-mass companions; none is resolved as a binary (Reid et al. 2006a, 2007). We do not have a spectrum for the new L4 standard 2MASS 0500+0330 that extends far enough into the red for a color-d index to be measured; however, in all other respects, the object meets the criteria that define a spectral standard.

2MASS J 21392676+0220226 is a faint source (J = 15.26) with red near-infrared colors ((J − H) = 1.10, (H − K_S) = 0.58). The prime aim of the present survey is the identification of late-type M dwarfs and L dwarfs, and these colors are broadly consistent with a mid-type L dwarf at a distance of 25–30 pc. However, the optical spectrum is smooth and largely featureless, with the exception of absorption by Cs I at 8521 and 8963 Å, and H₂O at 9300 Å (Figure 4). Moreover, Burgasser et al. (2006) have obtained low-resolution, near-infrared spectra that indicate the presence of methane absorption. This shows that 2M2139+0220 is a nearby early-type T dwarf, with a spectral type ≈ T1.5. On that basis, we estimate a distance of ∼15 pc. Further near-infrared spectroscopy of this dwarf will be particularly interesting.

Zoom In Zoom Out Reset image size

It is now well established that the presence of lithium absorption in ultracool dwarfs indicates that those objects have substellar masses (Rebolo et al. 1992). The critical temperature for lithium burning is ∼2 × 10⁶K, or ∼10⁶K cooler than the critical temperature for hydrogen burning. Low-mass dwarfs are fully convective; thus, the presence of detectable lithium in the photosphere indicates that the core temperature has never reached the critical value for hydrogen fusion. Theoretical models (Chabrier & Baraffe 1997) predict that lithium remains undepleted in brown dwarfs with masses below 0.055  M_☉, while lithium is subject to partial depletion in dwarfs with masses in the range $0.055 < {M \over M_\odot } < 0.075$, with the rate of depletion scaling with increasing mass.

We have examined our optical spectra, and identified lithium absorption in eight L dwarfs in the present sample. The measured equivalent widths for those sources are given in Table 8. We also list new observations of a number of L dwarfs from the 2MU2 sample. This represents a very low detection rate for the current sample, which is likely to be explained by the spectral resolution of our observations, coupled with the moderate S/N of our spectra of many late-type L dwarfs. It is notable that all of the lithium dwarfs listed in Table 8 were observed using GMOS on Gemini. Higher resolution and higher-S/N data are likely to reveal Li 6708 Å in a number of other dwarfs in both the present sample and 2MU2 samples.

Turning to the data listed in Table 8, in most cases the lithium lines are moderately strong, with an equivalent width of 3–4 Å. This may indicate that lithium is partly depleted in those systems, suggesting a mass close to 0.065 M_☉. There are, however, a handful of dwarfs with much stronger lithium absorption, such as 2M0310−2756, 2M0652+4710, and 2M2317−4838. We also note that several dwarfs listed in this table have spectral signatures consistent with low surface gravity (the spectral types for those dwarfs are enclosed in parentheses). The presence of lithium clearly adds further weight to the hypothesis that these are young, low-mass brown dwarfs.

Our optical spectra also allow us to probe chromospheric activity through measurements of Hα emission. The overall statistics for activity among ultracool dwarfs are discussed by Schmidt et al. (2007), and we consider the 20 pc L dwarfs in Section 5. Here, we draw attention to two particularly active dwarfs in the present sample: 2M0407+1546 and 2M1022+5825. The latter dwarf, which is discussed by Schmidt et al. (2007), is an L1 dwarf that shows substantial (order of magnitude) variations in the Hα line strength on a timescale of 1–2 days. The L3.5 2M0407+1546 has only one optical observation, with GN, but that observation shows Hα emission with an equivalent width of ∼60 Å. This makes 2M0407+1546 one of the latest-type dwarfs to show substantial chromospheric activity. Further observations may shed light on why this particular dwarf has maintained such a high level of activity at this juncture in its spectral evolution.