2MASS Discoveries of Cool, Dust‐enshrouded Carbon Stars out of the Galactic Plane1

James Liebert  

Steward Observatory, University of Arizona, Tucson, AZ 85721;

Roc M. Cutri, Brant Nelson, J. Davy Kirkpatrick, and John E. Gizis  

Infrared Processing and Analysis Center, California Institute of Technology, Pasadena, CA 91125; , , ,

I. N. Reid  

Department of Physics and Astronomy, University of Pennsylvania, Philadelphia, PA 19104‐6396;

ABSTRACT

Several very cool carbon stars have been found in or beyond the Galactic halo in the Two Micron All Sky Survey (2MASS). All appear to be N (C‐N) type asymptotic branch giants, much redder than most carbon stars found at high Galactic latitude at optical wavelengths. Three are heavily dust enshrouded, radiate primarily in the thermal infrared at temperatures of order 1000 K, and are IRAS sources. Low‐resolution red spectrophotometry with the Keck II and Palomar 5 m telescopes shows red to extremely red continua with CN absorption bands, similar to IRAS SSC 08546+1732, which is detected here (2MASSI J0857258+172052). One object, 2MASSI J0401097+182807, featured an overlying low‐excitation emission‐line spectrum with neutral alkalis (K i, Rb i, and Cs i), which disappeared after 19 months. This is argued to be due to vigorous mass loss probably associated with Mira‐like pulsations. These initial results suggest that 2MASS can be used to define a useful sample at high Galactic latitude as tracers of the Galactic halo out to distances comparable to those of the Magellanic Clouds.

Received 2000 April 28; accepted 2000 May 26

1. INTRODUCTION

 

Luminous carbon stars in the asymptotic giant branch (AGB) phase of stellar evolution are apparently rare in the Galactic halo. This is due to the fact that their progenitors, which are of intermediate mass (≳2 M) and relatively young (≲1 Gyr), are also exceptional in the halo. At their high luminosities, such objects offer the potential to trace the spheroid mass to great distances, as shown by Mould et al. (1985) and Bothun et al. (1991). Totten & Irwin (1998) are assembling a sizeable sample from scanned photographic Schmidt plates and suggest a surface density of ∼1 per square degree. Finding a sample of them over a large solid angle might also reveal the distribution of a relatively young population at high Galactic latitudes. If all such stars originate in the Galactic disk, one desires to know the maximum z‐distances at which they can appear or whether they are runaway objects from the younger Galactic disk. If such stars can be part of the halo population, are they part of a population of intermediate‐mass stars that formed in the halo, are they just the progeny of relatively massive blue stragglers, or could they be part of the dynamical interaction with the Magellanic Clouds or dwarf galaxies? Carbon giants have unique colors and spectra and are easily identified. When they lose mass at a high rate, they become enshrouded in circumstellar dust, causing reradiation in the thermal infrared (see the excellent review by Wallerstein & Knapp 1998). These are especially suitable for detection in an infrared sky survey. Embedded carbon stars, many discovered by detection at far‐infrared wavelengths by the Infrared Astronomy Satellite (IRAS), now account for some 25% of the known carbon stars (Guglielmo et al. 1993). However, nearly all of the known, dust‐enshrouded objects are believed to be in the Galactic disk population.

The Two Micron All Sky Survey (2MASS; Skrutskie et al. 1997) began in 1997 June at the Fred L. Whipple Observatory (Mount Hopkins, Arizona) and in 1998 April at the Cerro Tololo Inter‐American Observatory in Chile. A deep, wide‐angle survey like 2MASS presents the opportunity to find interesting, rare objects, including the type discussed in this paper. We report here the discovery of four new carbon stars at high Galactic latitude. Two of these are detected by IRAS at 12 and 25 μm and, hence, are likely to have circumstellar dust. We also detect a similar object (SSC 08546+1732) previously discovered with IRAS (Cutri et al. 1989)—our J2000 designation is 2MASSI J0857258+172052. Spectrophotometry of these five stars presented in § 2 shows all to be cool carbon stars.

The full 2MASS designations, photometric magnitudes, and errors for the five stars discussed here are listed in Table 1. Those designated 2MASSI are part of the Second Incremental Point Source Catalog;2 the 2MASSW star was drawn from the working point source database. In the remainder of this paper, objects will be referred to using truncated names, for conservation of page charges. Two measures for 2M0703 are listed, as it fell in the overlap region between two adjacent survey tiles and thus had two independent observations. Finding charts are not posted here, since all five stars can be seen using the 2MASS Survey Visualizer server at the Infrared Science Archive,3 even the SSC object which was not detected on the POSS I.

TABLE 1
TABLE 1 2MASS Magnitudes and Errors

Open New Window

Very red color was the general basis for selection of carbon star candidates, and our five confirmed stars have J between 2.3 and 4.7. In general these sources are significantly redder in JHK than our latest L dwarfs (Kirkpatrick et al. 1999), though overlapping with redder AGNs (Nelson et al. 1998). The carbon stars presented here are all that we have confirmed with spectra up to this time, but the assessment of very red point sources has not yet been systematic or extensive enough for general conclusions about the surface or space densities of such objects. Our principal purpose here is to characterize the sample of carbon stars that 2MASS can select.

2. SPECTRA AND ENERGY DISTRIBUTIONS

 

Spectrophotometric observations of the 2MASS carbon stars at red wavelengths were made with the Low Resolution Imaging Spectrograph (LRIS; Oke et al 1995) at the 10 m W. M. Keck Observatory (Keck II) and/or with the double spectrograph (DSpec; Oke & Gunn 1982) on the Hale 5 m telescope, as listed in Table 2. With LRIS a 400 groove mm−1 grating blazed at 8500 Å was used with a 1 slit and a Leach CCD to yield 1.9 Å pixel−1 (9 Å resolution) spectra covering the wavelength range 6300–10100 Å. An OG570 blocking filter was utilized to exclude second order from the blue standards. Further details of data acquisition and reduction are as given by Kirkpatrick et al. (1999, 2000). The Palomar spectra used a 300 groove mm−1 grating in the blue camera to cover the range 3375–6825 Å at 10 Å resolution. A 316 line mm−1 grating was placed in the red camera for coverage from 6800 to 9150 Å at a resolution of 10 Å. A 2 0 slit and dichroic beam splitter that splits the light near 6800 Å were used.

TABLE 2
TABLE 2 Log of Spectroscopic Observations

Open New Window

All red spectra are dominated by the red CN (2, 0) and (3, 0) band systems in the 7000–8500 Å wavelength range. In Figure 1 the spectra of the objects are ordered by decreasing strengths of these bands from bottom to top, likely caused by increasing circumstellar extinction and spectral veiling. The 2M0401 object shows an even redder energy distribution at these far‐red wavelengths than the IRAS(SSC) 2M0857 source. 2M0401 and a third carbon star (2M0326) are detected in the shortest two IRAS wave bands. The emission lines in the spectra will be discussed below.

Fig. 1.— Red spectrophotometry of five 2MASS carbon stars, in order of decreasing dust veiling (top to bottom), as described in the text.

Open New Window

These objects are dramatically cooler and/or more heavily extincted than the carbon stars found in published surveys at optical wavelengths. This is shown in the JHK two‐color diagram of Figure 2, where the high‐latitude carbon star (HLCS) samples of Mould et al. (1985) and Bothun et al. (1991) are shown for comparison. Also shown for comparison are Large and Small Magellanic Cloud (MC) carbon stars which from their absolute values must be luminous AGB stars (Mould & Aaronson 1983). In contrast, a sample of (warmer) R‐ and CH‐type carbon stars from Dominy et al. (1986), and four dwarf carbon stars from Green, Margon, & MacConnell (1991) and Dearborn et al. (1986), are shown at much bluer JHK colors. Indeed, it may be seen that only a few of the latest HLCSs and MC stars have JHK colors similar to the bluest 2MASS stars. The redder 2MASS sources resemble embedded Galactic disk AGB stars, as we show below using more complete energy distributions.

Fig. 2.— JHK two‐color diagram showing the five 2MASS carbon stars (filled circles) studied in this paper. For comparison are shown optically selected, high‐latitude carbon stars (HLCSs), luminous Magellanic Cloud AGB stars, four dwarf carbon stars, and warmer R and CH stars; the references from which these magnitudes were taken are given in the text.

Open New Window

The distinction between N and R classes of carbon stars can be ambiguous (Keenan 1993), though the real ambiguity appears to be between early N and late R at least at bluer wavelengths. Keenan (see his Table 1) considers the latest R and CH subtypes equivalent to early M oxygen type, while it clear from Figure 2 and the preceding discussion that the carbon stars discussed here are comparable at least to late M giants. Given also the dominance in the red spectra of the CN band systems, we believe it is appropriate to classify them as late N type, or more properly late C‐N type in the revised system of Keenan (1993). However, Keenan’s spectral classification system is defined at bluer wavelengths and higher resolution. Spectroscopic atlases and catalogs for carbon stars are published at much higher spectral resolution (Barnbaum 1994) or bluer wavelengths (Barnbaum, Stone, & Keenan 1996). Spectral classification of these objects may be confused by the fact that they are heavily embedded and the optical spectra appear to be veiled. There is also evidence that they are spectrum variables.

On two occasions, 2M0401 exhibited an impressive emission‐line spectrum, with resonance transitions of easily ionizable alkali metals. We obtained three spectra of this object to search explicitly for variability, all displayed in Figure 3. The C2 Phillips band near 8750 Å reverts from normal photospheric absorption into emission. We cannot exclude the possibility that 2M0401 is some kind of symbiotic binary with a warm (but not hot) companion. We argue based on the documentation below that episodes of enhanced mass loss are capable of generating weak shocks to produce such low‐excitation emission lines. The last spectrum of 2M0401 indicates that mass loss can weaken enough for a more normal photospheric spectrum to appear. Curiously, the bottom three stars of Figure 1 appear to show Hα in emission, though lacking the lower excitation features, while 2M0401 does not show Hα.

Fig. 3.— Three red spectra of 2M0401 show the weakening of the emission spectrum over a 19 month period. It is possible that this is linked to changes in the mass‐loss rate over part of a Mira pulsation cycle.

Open New Window

Barnbaum (1992) has published multiepoch, higher spectral resolution studies of bright carbon stars which include those which are Mira and semiregular variables. She finds some 33 stars, mostly Mira variables, which show Hα in emission. Moreover, she finds that K i is seen in two Mira variables with emission components—P Cygni or inverse P Cygni profiles—in a manner that seems to correlate with the pulsation phase. However, inspection of Figure 5 of Barnbaum (1992) suggests that the EW at maximum emission strength for the active objects she observed is no more than 1.5 Å. The K i lines reached several Å equivalent widths in the top spectrum of Figure 3.

More evidence for mass loss is found from the far‐infrared fluxes. The energy distributions of the carbon stars with IRAS detections are plotted in Figure 4. Shown for comparison is a normalized energy distribution of the prototypical embedded carbon star IRC +10216, adapted from Figure 2 of Wallerstein & Knapp (1998). The energy distribution of IRC +10216 has been argued to be a composite of (1) the heavily extinguished photosphere of a ∼2000 K AGB star and (2) a ∼700 K component of reprocessed dust radiation. A 700 K blackbody energy distribution normalized to the 12 μm flux of 2M0401 also provides a fair fit to the far‐IR fluxes of the 2MASS objects. However, the 2MASS stars show stronger excesses at shorter wavelengths than does IRC +10216, due to reddened photospheres. Again, the three 2MASS objects with IRAS detections, which also have the reddest J colors, also exhibit the weakest photospheric absorption in Figure 1.

Fig. 4.— Energy distributions of the three carbon stars with detections in the IRAS 12 and 25 μm bands, plus 2MASS fluxes (JHK) and L band (open squares: 2M0401; filled triangles: 2M0326; and filled circles: 2M0857). The solid curve is the approximate energy distribution of IRC +10216, and the dashed curve is a 700 K blackbody, both normalized approximately to the 12 μm IRAS flux of 2M0401.

Open New Window

To sum up, the evidence is consistent with the assumption that the 2MASS‐selected carbon stars are luminous AGB giants. The dust is indicative of the formation of a circumstellar envelope as the AGB star increases greatly in radius and luminosity and loses mass at a prodigious rate. Carbon‐rich envelopes form dust especially efficiently, and such a dense envelope reprocessing the photospheric radiation releases most of the stellar luminosity as mid‐ to far‐infrared radiation. The process may end in the formation of a planetary nebula heated by the exposed, hot stellar core after it begins the post‐AGB phase.

3. ABSOLUTE MAGNITUDES AND DISTANCES

 

For the first time, measured trigonometric parallaxes exist for a large number of carbon giants, thanks to the Hipparcos mission. Absolute magnitudes determined for N‐type stars are reviewed in Wallerstein & Knapp (1998, their Table 1). The means for SRb and Lb variables are similar at ; the Mira variables and SRa means are −7.6, both with more than 1 mag dispersions. Note that the pulsational variability of typical Mira variables accounts for a 1–1.5 mag K range by itself. Since the IRAS and 2MASS fluxes were obtained at different times, we could not try to use these data to study the relationship between dust temperature and luminosity even if we knew the latter. We have already shown that the canonical 700 K blackbody is not a bad approximation. It is generally believed that the dust temperature is warmer near maximum luminosity of the pulsation cycle.

In order to estimate roughly the distance to each object and its height above the Galactic plane, we adopt the assumption of Claussen et al. (1987) that the embedded carbon AGB stars lie generally between and −8.5, corresponding to a luminosity range of near the AGB tip. The recent evidence cited above is generally supportive of this choice. In Table 3 we present distance moduli and heights (in kiloparsecs) above the Galactic plane for the stars. The latter indicate strongly that most are members of the Galactic halo, or possibly even beyond the Galaxy.

TABLE 3
TABLE 3 Galactic Coordinates, Distance Moduli, and z‐Distance

Open New Window

The finding of carbon giants at likely distance moduli of 18–20 indicates that the 2MASS survey has the chance to give a complete map of the Galactic halo out to the distances similar to those of the Large and Small Magellanic Clouds. It is already clear that 2MASS detects a large number of similar carbon stars in and near the Magellanic Clouds (Nikolaev & Weinberg 2000).

This publication makes use of data products from the Two Micron All Sky Survey, a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation. We thank the referee, Jill Knapp, for several helpful criticisms and suggestions. This research is supported by a NASA JPL grant (961040NSF). R. M. C., B. N., J. D. K., and J. E. G. acknowledge the support of the Jet Propulsion Lab, California Institute of Technology, which is operated under contract with NASA.

REFERENCES

 
  • 1 Portions of the data presented here 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 R. M. Cutri et al. 2000, Explanatory Supplement to the 2MASS Second Incremental Data Release (http://www.ipac.caltech.edu/2mass/releases/second/doc/explsup.html).

  • 3 Infrared Science Archive (http://irsa.ipac.caltech.edu).

© 2000. The Astronomical Society of the Pacific. All rights reserved. Printed in U.S.A.