Collaborateurs: Paul Green (Harvard). Lars Winter (Hamburg).
Project description: Carbon (C) stars, readily recognizable from their strong C$_2$ and CN absorption bands, have been detected as faint as $V\approx18$ on objective prism plates (e.g., Sanduleak \& Pesch 1988, ApJS, 66, 387; MacAlpine \& Lewis 1978, ApJS, 36, 587), and in the CCD survey of Green et al. (ApJ, 1994, 434, 319). These faint high galactic latitude carbon (FHLC) stars, assumed to be giants at galactocentric distances as large as 50 - 100$\,$kpc, have been sought as dynamic tracers of the outer halo, invaluable for understanding the structure, dynamics, and evolution of the Galaxy (e.g., Mould et al. 1985, PASP, 97, 130; Bothun et al. 1991, AJ, 101, 2220). Except for one odd dwarf, \hbox{G77--61}, it was long assumed that carbon stars are always giants, as carbon is thought to reach the photosphere only during dredge-up in AGB stars. However, Green et al. (1992, ApJ, 400, 659; hereafter G92) discovered four further dwarf carbon (dC) stars, each found due to high proper motion (hereafter, p.m.). At least 2 more dCs have subsequently been discovered by similar means (Warren et al. 1993, MNRAS, 261, 185). In addition, Heber et al. (1993, A\&A, 267, L31) report that PG0824+289, a hot DA white dwarf, is in a double-line spectroscopic binary with a dC companion of absolute magnitude very similar to G77--61 ($M_V\approx+10$). Another DA/dC composite system, CBS311, was recently discovered by Liebert et al. (1993, ApJ, 421, 733). The most reasonable explanation of the prominent carbon bands in these dwarf spectra is mass transfer from a now unseen companion during the companion's second ascent of the giant branch (AGB). Enhanced \hbox{$s$-process} element abundances (Green \& Margon 1994, ApJ, 423) lend further credence to this scenario. With about 10 dC stars now known from an incomplete hodgepodge of surveys and serendipity, the conclusion seems inescapable that many more C giants than C dwarfs are presently known only by virtue of the much greater luminosities of the former class. From a p.m. survey of 39 FHLC stars, G92 have estimated that 13\% are spheroid dwarfs. They conclude that the local space density of dCs probably surpasses that of all other types of C star combined. This conclusion is {\it conservative}. First, the samples of giants and subgiants used for comparison are, to typical survey magnitude limits, far more complete than samples of the much fainter dwarfs. Second, disk dCs may remain in the FHLC sample, with p.m.s below the detection threshold. Thus, contrary to the formerly prevailing paradigm, {\it dwarf {\rm C} stars are likely to be the numerically dominant type of carbon star in the Galaxy.} The discovery of so many dCs, and the remarkable similarity of their spectra to those of C giants means that care must be taken to distinguish dwarfs from giants in FHLC star samples intended for dynamics (G92). As innocent bystanders in a mass transfer binary (MTB) system, dC's spectroscopic and orbital properties provide valuable fossil records of the history and evolution of an extinct population of AGB stars. In many cases, substantial amounts of mass must have been accreted to drive C$>$O in the deep convective envelopes of these rather low mass dwarfs. Therefore, the initial masses of many dCs may have been near or below the hydrogen-burning limit (Liebert et al. 1994, ApJ, 421, 733). Even using conservative constraints on the shape of the IMF and binary fractions in the disk and spheroid, we may expect mass transfer to have occurred in a large number of such systems (DeKool \& Green 1995, ApJ, 449, 236). The majority of these low mass, post MTB stars could be M dwarfs, in which the evidence for past mass transfer may be subtle, if not undetectable. In dCs, the signature of extensive mass transfer is glaringly obvious. Intrinsically luminous C stars have been detected from R=12 to 16 in the outer halo to galactocentric distances as large as 50$\,$kpc from objective prism plates (Mould et al., 1985, PASP, 97, 130; Bothun, G. et al. 1991, AJ, 101, 2220; Sanduleak \& Pesch 1988, ApJS, 66, 387; Totten \& Irwin 1998, MNRAS, 294, 1). However, on the Hamburg/ESO survey (HES) objective prism plates an automated, more sensitive and efficient selection of such stars is possible. The spectral resolution of the digitized HES spectra is $\sim 15$\,{\AA} at H$\gamma$, and the wavelength range is $3200\,\mbox{\AA}<\lambda<5300\,\mbox{\AA}$, so that several strong C$_2$ bands are easily detectable (see Fig. 1), and a further spectroscopic confirmation of the carbon star candidates is not necessary. Additionally, our sample is complementary to that of Totten \& Irwin (1998), which was biased towards red stars. For the HES, the C star selection function can be quantified, as CCD or model spectra can be converted to objective prism spectra via rebinning and convolution with spectral sensitivity curves of HES plates (for examples of ``simulated'' spectra and comparison with ``real'' spectra see also Fig. 1). The HES covers 10\,000 sq.deg. in the southern extragalactic sky. On a subset of 104 plates (effective area $\sim 2100$ sq.deg.) we have identified 96 carbon star candidates by using two line indices placed at the positions of strong C$_2$ bands (cf. Fig. 1). The usage of {\em two\/} indices prevents confusion with plate artifacts (e.g. scratches). A selection box in the {\tt C2idx1} vers. {\tt C2idx2} plane has been choosen such that stars with obvious C$_2$ bands are selected by the automatic procedure. The selection region is very well seperated from the ``normal'' star region (see Fig. 2). All selected spectra are inspected individually on the computer screen to identify any remaining plate artifacts and to verify their C star nature. The selection procedure has been tested by 13 known C stars present on HES plates (cf. Fig. 2). The stars have been taken from Stephenson's carbon star catalogue (1989, Publ.W.\&S.Obs., 3, No. 2), Totten \& Irwin (1998) and an compilation of Green. Additionally, 5 ``simulated'' C star spectra have been used. Of these 18 spectra, 17 (or 95\%) have been re-discovered by the selection procedure. The only object not selected is entry \#3180 in Stephenson's catalogue, which is saturated in the HES ($B_J=12.8$). The number of candidates found by our selection is interesting in itself, suggesting a high level of completeness of our HES sample. Green et al. (1994) estimated the surface density of FHLCs to $0.02$ per sq.deg. for a survey limit of $V=18$, whereas we find $0.045$ candidates per sq.deg. within the HES limit for FHLCs of $V=16.5$! Proper motions will be derived by comparison of POSS positions from the Digital Sky Survey (epochs 1975--84 for the relevant plates) to the CCD positions, using a large number of control stars, as in Deutsch 1994 (PASP, 106, 1134). This method will enable us to rapidly measure proper motions for all program stars to within $\sim 30$ milliarcsec/yr for the shortest epoch-to-epoch baseline, more than adequate to select {\em all} halo dCs to within the HES limit ($d\sim 200$\,pc for the $M_V=10$ dCs). Not all dCs will have detectable p.m.s, so other luminosity/distance indicators are needed. Comparison of high resolution optical spectra to model atmospheres might seem promising, but no reliable models of the complicated spectra of dCs have yet been generated. In addition, most dCs have $V\geq 15$, so spectroscopy of sufficient resolution for luminosity estimates is difficult. Other luminosity indicators are clearly needed. G92 have shown that high p.m. FHLC stars may exhibit distinctive $JHK$ colors appropriate to late-type dwarfs, suggesting IR colors as a possible luminosity indicator. However, the sample of known dCs is still too small to draw any definitive conclusions. In period 65 we will propose to obtain $JHK$ photometry using SOFI and the NTT for all new dCs from the campaign proposed here. Their infrared colours will then be compared with those of known non-p.m. C stars from this survey and from the literature. Harris et al. (1998, ApJ, 502, 437) has recently shown that the visual ($BVI$) colors of (3) dCs differ from all known red dwarfs, subdwarfs, and white dwarfs. Even more rapid progress in our understanding will be made if we can publish $BVI$ photometry for a larger sample of dCs. Recent progress in theoretical model atmospheres for dCs (Tsuji 1996, A\&A, 308, 29) require matching to optical/IR photometry. Thus, a very small additional overhead to our primary $V$ imaging campaign for astrometry, will add significantly to C star research, and quite possibly enable detection of {\em disk} dC stars. Most FHLC stars we find wil not be dCs, but rather distant giants. These provide excellent probes of the structure, composition, and dynamics of the outer Galactic halo between 2 and 30\,kpc. Put simply, a sample of about 100 distant halo C giants will yield the mass of the Milky Way, including all the dark matter! Combined samples can be used for both dynamical analysis and tests of C to late M star ratios at different radii in the Galactic halo, a key observation for constraining theories of the collapse and formation of the Milky Way (Stephenson 1986, ApJ, 301, 927). The search for dwarf carbon stars will be extended to another $\sim 100$ HES fields in period 64. This will yield another $\sim 100$ candidates. According to the estimates of G92, 13\,\% of the total candidate sample of $\sim 200$ stars will be dCs, so that we expect the sample of known dCs to be increased by approximately a factor of three by our program.
Last modified 15.1.1999 by
nchristlieb@hs.uni-hamburg.de