Discovery of Low-Luminosity Carbon Stars in the Fornax Dwarf
Spheroidal Galaxy with the MOS at CFHT

Marc Azzopardi, Gérard Muratorio

Observatoire de Marseille
2, place Le Verrier, F-13248 Marseille Cedex 04, France
Electronic-mail: azzopardi@observatoire.cnrs-mrs.fr, muratorio@observatoire.cnrs-mrs.fr

Abstract:

We report circumstances and observational techniques used with the CFHT Multi-Object Spectrograph that led us to the discovery of low-luminosity carbon stars in the Fornax dwarf spheroidal galaxy. We review the carbon star surveys carried out with this system and present our main results. A more detailed study based on our CFHT and ESO observations will be published in the proceedings of the IAU Symposium 192, ``The Stellar Content of the Local Group''.

Introduction

  Within the framework of our studies of the nearby system stellar contents, we have been engaged, for some time, in the search for carbon (C) stars in Local Group galaxies in order to investigate the possible correlation of the carbon star population luminosity function with the metallicity of the related parent system. Earlier slitless spectroscopic photographic searches at the prime foci of the ESO and CFH 3.6-m telescopes equipped with wide-field correctors and very low dispersion transmission gratings (grism or grens) allowed us to identify C-star candidates (see Azzopardi (1994) and references quoted therein) in the Magellanic Clouds and the dwarf spheroidal galaxies in the galactic halo. The efficiency of our surveys was mainly due to the detection technique used: first we adopted the Swan C₂ bands at 4737 and 5165 Å in the blue-green spectral domain for C-star selection criteria - a method innovated by Sanduleak & Philip (1977) - while we restricted the instrumental spectral range to a useful domain (4350-5300 Å), by combining a IIIa-J emulsion with a Schott GG435 filter, in order to keep the number of the spectrum overlaps as low as possible. In addition to the identification of a large number of objects that escaped previous near-infrared photometric and/or spectroscopic surveys, a noteworthy result was the detection in the Small Magellanic Cloud (SMC) of a sample of faint and relatively blue C-star candidates. Subsequent slit spectroscopy and JHK photometry of these objects resulted in the discovery of low-luminosity C stars with $M_{\rm bol}$ ranging from -3.0 to -1.4 (Westerlund et al. 1992, 1995) (assuming (m-M)₀=18.8 mag) which are in many ways similar to those found at CFHT in the Galactic bulge by Azzopardi et al. (1991) and are among the least luminous C-stars ever found in an external system.

C-Star Surveys in the Fornax Dwarf Spheroidal Galaxy

Spectroscopic observation of seven objects selected by Aaronson & Mould (1980), among the 66 very red stars that had been discovered by Demers & Kunkel (1979) by blinking a pair of B, V plates obtained with the CTIO 1.5-m telescope, revealed the presence of the first five C-stars in the Fornax dwarf spheroidal. From these results, Aaronson & Mould claimed that ``for B-V greater than $\sim$2.1, the giant branch of Fornax may consist entirely of C-stars''. Effectively, the subsequent near-infrared grism searches for late-type stars, carried out by Blanco and McCarthy (Frogel et al. 1982) at the prime focus of the CTIO 4-m reflector (one 23 arcmin diameter field) and by Westerlund (Richer & Westerlund 1983) at the prime focus of the ESO 3.6-m telescope (three 16 arcmin diameter fields), were meant to increase the number of the C-star candidates known in this galaxy. The ESO survey (Westerlund et al. 1987 ; Lundgren 1990) provided a list of 48 probable C-stars, 23 of them being new identified objects.

CFHT photographic wide-field camera grens survey

In December 1983, during our C-star galactic halo dwarf spheroidal surveys (Azzopardi et al. 1985, 1986), two plates (the same one sq. degree field of view) were obtained by Lequeux and Azzopardi in the direction of the Fornax galaxy with the 3.6-m CFH Telescope equipped with the wide-field corrector and the blue-green grens providing 2000 Å/mm dispersion. A visual scrutiny of these plates by James Lequeux resulted in the detection of a number of bona fide C-star candidates. Subsequent ESO spectroscopic observations led us to a discovery of 29 new C-stars, 17 of them being outside the previous near-infrared grism survey limits. Obviously, this result is due, for the most part, to the high quality of the images provided by the CFHT - which is a basic parameter for slitless spectroscopy - in spite of the unfavourable declination of the target (-30$^{\circ}$) for the Mauna Kea site. Those that escaped detection in the near-infrared are relatively blue C-stars, hence the strength of the Swan C₂ bands with regard to the CN ones more than compensates the loss of flux in the blue-green spectral domain - most likely - for this kind of object. Our medium resolution slit-spectroscopy with various ESO telescopes to confirm the carbon star nature of a large number of objects coming from the complete C-star candidate sample known in Fornax at that time, allowed us to estimate at about 74 the total number of C-stars belonging to this system.

CFHT MOS slitless prism survey

From the experience we gained with the photographic field spectroscopy technique, we have been engaged for some time in a CCD camera low dispersion slitless spectroscopy using suitable interference filters (see Muratorio & Azzopardi 1997). In November-December 1994, the C-star population of the Fornax system was revisited by us with the CFHT, as a back-up programme, at the end of three observing nights devoted to a search for new Wolf-Rayet stars in M31. We used the Multi-Object Spectrograph (MOS) in the slitless spectroscopy mode equipped with an interference filter having a 1025 Å bandwith centered at 4850 Å. For this purpose, a prism giving about 800 Å/mm dispersion at 4800 Å was designed by Derrick Salmon and Gregory Barrick in order to avoid the parasitic images corresponding to the different orders provided by the low-resolution transmission gratings. Five contiguous slitless spectroscopy images (10$\times$10 arcmin² each) covering the central region of the Fornax galaxy were obtained with the 2048$\times$2048 pixel Loral 3 chip (15 $\mu$m/0.31 arcsec pixel) in 1200s integration each. Several new C-star candidates were identified both by visual inspection of the frames using the SAO image tool and by means of a semi-automatic procedure developed by Gérard Muratorio (Muratorio & Azzopardi 1994) in the Munich Image Data Analysis System (MIDAS) environment. Subsequent multi-object spectroscopy with the Meudon ESO Fiber Object Spectrograph (MEFOS) mounted on the ESO 3.6-m telescope led us to confirm the C-star nature of 29 additional very faint C-stars. All these new objects, which are inside the limits of the previous transmission grating surveys, are too faint to be detected by the photographic cameras used. Actually, the C-star census of the Fornax system can be considered achieved, at least inside the boundaries of our CFHT MOS survey.

C-Star Population in the Fornax Dwarf Spheroidal Galaxy

With more than one hundred confirmed C-stars, not taking into account the ``continum'' (Ctm) stars, Fornax contains by far the largest number of these objects known among the galactic halo dwarf spheroidals. Near-infrared JHK photometry of nearly all the C-stars identified in the Fornax dwarf, mainly carried out by us with IRAC2 at the ESO 2.2-m telescope, led to determination of the virtually entire C-star luminosity function for this system. The bolometric magnitudes, derived following Wood et al. (1983), point out the presence of low-luminosity C-stars as faint as $M_{\rm bol} \simeq -1.2$, assuming (m-M)₀=21.0 mag as the absorption-free distance modulus for the Fornax galaxy. The latter resembles that we had previously found in the SMC (Westerlund et al. 1992, 1995).

Our discovery of low-luminosity C-stars in two metal-poor galaxies corroborates the theoretical predictions about the dependence of mean C-star luminosity on metallicity, leading to the expectation that, for two systems with similar star-formation histories (i.e., similar initial stellar mass functions and star formation rates), the mean luminosity of C-stars would be lower in the system with lower metallicity. As a matter of fact, given two stars of equal mass but different metallicities, the metal-poor one will become a C-star at a lower luminosity than the metal-rich one since less carbon has to be dredged up to the surface to get C/O>1. However, the faint-ends of the C-star luminosity functions of more nearby extragalactic systems need to be taken into consideration to first confirm this trend and then to better constrain the aforementioned theory.