EDP Sciences
Free Access
Issue
A&A
Volume 544, August 2012
Article Number A10
Number of page(s) 6
Section Stellar structure and evolution
DOI https://doi.org/10.1051/0004-6361/201219108
Published online 19 July 2012

© ESO, 2012

1. Introduction

Carbon stars represent the brightest members of the intermediate-age population. Their evolutionary status and age were established by Catchpole & Feast (1973), who observed a few intermediate-age globular clusters in the Small Magellanic Cloud containing carbon stars. They are also seen in massive dwarf spheroidal galaxies, such as Fornax (Demers & Kunkel 1979; Aaronson & Mould 1980), Leo I (Lee et al. 1993), or Sagittarius (Sgr) (Whitelock et al. 1999).

Carbon stars are not found in Galactic globular clusters. In principle, they should also not be seen in the halo of the Galaxy, which is made up of old stars. However, extremely red stars, observed at high Galactic latitudes, were confirmed to be C stars by Totten & Irwin (1998), and subsequently, over one hundred halo C stars, selected from their 2MASS colours, were spectroscopically confirmed by Mauron et al. (2004, 2005, 2007) and Mauron (2008).

If we understand the star formation history of the Galactic halo sufficiently well, then these stars must be alien trespassers. Some of them are faint enough to suggest that they have distances from the Sun of greater than 50 kpc. The Totten & Irwin (1998) carbon star survey led Ibata et al. (2001) to trace, for the first time, the tidal stream of the Sagittarius dwarf spheroidal galaxy orbiting the Galaxy. Ibata et al. (2001) concluded, however, that roughly half of the halo carbon stars, known at the time, do not seem to belong to the Sgr stream. These stars could belong to some other unidentified streams, although this has not yet been established. Accurate distances and velocities would be required to assess their origin more accurately.

It is known that the majority of carbon stars are variable. All the R and N-type C stars observed by Hipparcos indeed, turned out to be either irregular, semi-regular, or nearly periodic variables (Grenon et al. 2000). CH stars originate from mass transfer in binary systems and with colours (J − K) < 0.8 (Goswami et al. 2010) they form a class of objects photometrically distinct from our N-type stars, which have (J − K) > 1.4. Since C stars are thin-disk objects, variable star surveys have been limited to low Galactic latitudes. However, two high Galactic latitude C star variables were detected by Meusinger & Brunzendorf (2001) and Meusinger (2002), which are both close to the direction of the globular cluster M92. One with a 2MASS (J − K) = 1.35 has a period of ~0.81 yr, while the second one is bluer, (J − K) = 0.67, and has no known period.

Evidence has accumulated suggesting that the period of a Mira, regardless of whether it is O-rich or C-rich, is related to its mass and age. These clues were briefly summarized by Feast (2009). Evidence has come from the kinematics of Miras around the Sun (Feast et al. 2006) and the study of LMC globular clusters, of known age, containing Miras (Nishida et al. 2000; van Loon et al. 2003; Kamath et al. 2010). Periods of Miras in dwarf spheroidal galaxies are also compatible with the known star formation history of these galaxies (Menzies et al. 2011). Essentially, low-mass old giants have short periods while young ones (<1 Gyr) have periods, well over 500 days. Thus, the detection of the variability and a period determination of a halo C star provide important clues to their origin.

A more clearly defined age for the halo C stars would provide insights into the merging history of the hypothetical “victim” that left its debris around our Galaxy. The recent identification of a group of younger globular clusters by the HST ACS Survey, led to similar conclusion. It is indeed very tempting to argue that their origin is related to their formation within Milky Way satellite galaxies that were later accreted (Marin-Franch et al. 2009).

2. Targets and observations

The 25 carbon stars selected for monitoring are given in Table 1. The id numbers are from Mauron’s lists. We list their equatorial J2000.0 coordinates and both their Galactic longitude and latitude along with their Ks magnitude from 2MASS. Since they were to be observed from Chile, we selected targets with δ  <  −20° in order to be able to observe them over several months.

Table 1

id numbers, coordinates and Ks magnitudes of selected targets.

The observations, presented in this paper, were secured with rapid eye mount (REM), which is a robotic 60 cm telescope located on La Silla. The telescope hosts two instruments: REMIR, an infrared imaging camera with a 512 × 512 array giving a scale of 1.221′′ per pixel, and ROSS, a visible imager, equipped with a 1024 × 1024 CCD giving a scale of 0.575′′ per pixel. The two cameras can observe simultaneously thanks to a dichroic placed before the telescope focus. The field of view is 10′ × 10′. The observatory is operated for INAF by the REM Team1.

Observations were obtained, every ~15 days, weather permitting, during the long observing season of each target. Targets were followed during four semesters, from February 2010 to January 2012. One observation consists of two I exposures of 60 s and two K′ and J exposures of 30 s each. The images were analysed with SExtractor (Bertin & Arnouts 1996). Since the differences between K′ and Ks are of the order of a few hundredths, we omit the prime and the “s”.

The NIR instrumental magnitudes are calibrated with 2MASS point source observations seen in the field. The I images are calibrated with DENIS (Epchtein et al. 1997) data, which, unfortunately, are not available for every field. The photometric accuracy in J and K depends on the apparent magnitudes, as shown in Fig. 1. We have calculated the standard deviation in two calibrated J and K measures obtained on the same night. The means of the two measures were sorted by brightness and binned into groups of eight. We plot the average standard deviation versus the average magnitude in each bin. We see that for K > 12. and J > 13.4 the photometric errors become significant.

thumbnail Fig. 1

Standard deviations of two observations as a function of magnitude. Each point represents the mean of eight measures.

Open with DEXTER

Table 2

Mean magnitudes and colour of program stars.

thumbnail Fig. 2

J magnitude variations of the 23 program stars. Note that the magnitude axis covers two magnitudes. For m11, we plot the K mag.

Open with DEXTER

3. Results

In Table 2, we present the arithmetic mean magnitudes and colours of our target stars. The numbers of observations in each band are also listed. The small number of I magnitudes precludes their use for a period search. Stars with missing I magnitudes are those without DENIS stars and I magnitudes in their field. We also include the reddening for (J − K) taken from Schlegel et al. (1998) and adopt their transformations E(J − K) = 0.526E(B − V) and AK = 0.367E(B − V). We note that for some targets, seen toward the Galactic centre, the reddening is not negligible. Three of our program stars, listed in Table 1, are too faint to be properly observed with the 60 cm telescope. m84 was withdrawn from the observing program because it was too faint; m11 is very red, 2MASS colour (J − Ks) = 2.79, and we obtained only one J magnitude value which we give in Table 2; m54 is quite faint and often invisible, even when the two combined K 30 s exposures are used. We have few observations of these last two targets.

3.1. Light variations

The J magnitudes of the program stars are plotted as a function of time in Fig. 2. To facilitate the amplitude comparison, the Y axis covers two magnitudes in J, excepted for m18, which has a J amplitude larger than 2 mag. For m11, we plot K magnitudes.

Inspection of the figures reveals three types of variations: a large amplitude as in the case of m18; low amplitude as for m50, and a very irregular variation as for m37. The light variations of faint stars such as m37, m48 and m26 display a large scatter.

To enable us to classify the low-amplitude variables, we inspected their I mag observations, when available. The I magnitudes of five of these stars are shown in Fig. 3. The inspection of the I mag variations neither reveals any periodicity nor provides significant additional clues about the nature of the stars.

thumbnail Fig. 3

I magnitude light variations for irregular or low-amplitude stars.

Open with DEXTER

3.2. Period search

We selected the J magnitudes to search for periodicity because the variables have a larger amplitude in this band than in the K band. I magnitudes might have been useful in this respect but we have fewer I measures because we started to obtain I observations only during the second semester. A simple sine curve is fitted to the data points. The quality of the fit is defined by the χ2 parameter, which corresponds essentially to the mean squared deviation between the observed points and the fitted sine curve. We adopted the period yielding the smallest χ2. We are confident that the accuracy of the adopted periods is smaller than 10 days. Periods are seen to fluctuate within this range as we add a new data point. A second method was also attempted, the Phase Dispersion Minimization code (Stellingwerf 1978). Periods found with this approach did not produce better fits to the light curves. We are unable of course to determine periods longer than the time interval covered by our observations. We see no long secular trends, which would suggest periods of 600+ days. In Table 3, we give the time interval (days) covered by the observations (ΔT), the periods found in days, the χ2 parameter indicating the quality of the sine fit, and the K, J, and I amplitudes of the light curve along with remarks or classification. Even for genuine large-amplitude variables, the χ2 can be large if the light curve does not repeat itself from cycle to cycle. This is the case for both m06 also m35, which have two maxima that are not at the same luminosity. As a rule, we can says that if χ2 is greater than ≈ 0.15 the light curve shows a substantial amount of scatter.

Soszyński et al. (2009) used the I-band amplitude to distinguish Miras and semi-regular variables. Miras are defined as stars with an I-band amplitude larger than 0.8 mag, while Whitelock (2012) also takes into account the K amplitude, ΔK  >  0.4 mag. On the basis of this criterion, we have 14 Miras in our sample. We include m24 in this class even if its ΔI is quoted as 0.6 mag because we do not have I observations for the dates of the first J magnitudes (see Fig. 2). Slightly more than 60% of the observed C stars are found to be Miras. In Fig. 4, we show two examples of light curves plotted against the phases. We note that the x axis covers more than one cycle, some of the points being repeated.

Table 3

Period, amplitude, and classification.

thumbnail Fig. 4

Examples of light curves with different χ2, 0.15 for m41 (top) and 0.04 for m31 (bottom).

Open with DEXTER

4. Discussion

It is well-established that the period of a Mira represents a good indicator of the stellar population to which it belongs. Long-period Miras are expected to have high mass progenitors, thus belong to a younger population (Iben & Renzini 1983). Therefore, the period of a Mira can be used as a proxy of its age, as suggested by Habing & Whitelock (2004). For example, Feast et al. (2006) inferred, from the velocity dispersion of Galactic Miras, that those with periods longer than 500 days are ≈ 1 Gyr old, while those with P ≈ 375 days are ~3 Gyr. Hereafter, we adopt the Habing & Whitelock (2004) period/age groups: Miras with P < 250 days are older than ~5 Gyr; variables with periods between 250 days and 550 days are of intermediate-age, between 1 and 3 Gyr; longer periods correspond to Miras of 1 Gyr old or younger.

The effect of metallicity on the period of a Mira is at the present time somewhat uncertain: very little has been done to investigate this relationship. Using oxygen-rich Miras and semi-regular variables in globular clusters, Feast & Whitelock (2000) established a “maximum period” versus metallicity relation. Since globular cluster stars are old, 96% of Feast & Whitelock (2000) sample have P < 250 days. We do not know whether the trend extends to longer periods and whether carbon-rich variables also follow such a trend. We are therefore obliged to neglect the metallicity effect for the time being.

In Fig. 5, we compare the period distribution of the halo Miras with Miras in dwarf galaxies without current star formation. Data for the dwarf spheroidal galaxies are for Leo I, from Menzies et al. (2010), Fornax, from Whitelock et al. (2009),and both NGC 185 and NGC 147, from Lorenz et al. (2011), where we consider as Miras all variables with ΔI > 0.8.

The bulk of the Miras in NGC 185 and, to a lesser extent, NGC 147 are old stars, while most of the Miras in the Galactic halo and Fornax are of intermediate ages. In Leo I, about 60% of the Miras are of intermediate-age; since there are currently only seven known Miras in this galaxy, its age distribution is poorly defined. Fornax has had a long and complex star formation history. Intermediate-age stars are abundant in its inner and intermediate fields (Held et al. 2009), which suggests that the C star Miras located in the Galactic halo originate from a population somewhat similar to Fornax. The Sgr galaxy could be a possible candidate. Unfortunately, its Mira population has not yet been surveyed, but Bellazzini et al. (2006) estimated that the mean age of its central population is 8 Gyr. We should then expect its period distribution to differ from that of Fornax, and for it to contain many short-period Miras.

thumbnail Fig. 5

Comparison of the period distributions of Miras. The dashed lines define the three age groups.

Open with DEXTER

4.1. Distances of Miras

It has been known for more than 30 years (Glass & Lloyd Evans 1981) that Mira variables, discovered by optical surveys, follow a period-luminosity relation in the near infrared. However, surveys performed in the near-infrared also identify dusty variables with circumstellar envelopes that are missed by optical surveys. The dust that affects the K magnitudes is responsible for the large scatter seen in the K − log P plane, as in the case, for example, of the Miras in NGC 6822 (Battinelli & Demers 2011) observed during a survey done in the near-infrared.

To properly determine the MK of the Miras, regardless of whether they are dusty, we adopted the approach of Ita & Matsunaga (2011) in using a P-L-C relation. Their colour correction was given as: We used the MK period-luminosity relation of LMC carbon stars determined by Whitelock et al. (2008), but instead of adopting their LMC distance of 18.39 we used a modulus of 18.50 (Koerwer 2009; Pietrzyński et al. 2009; Laney et al. 2012): In Table 4, we list the estimated distances of the 14 Miras identified.

Table 4

Distances of Miras.

4.2. Association with either the Sagittarius galaxy or stream

Only one of our target stars, m50, is clearly associated with the Sgr galaxy. It is within its elongated central region, eight degrees from its centre. However, 4 other C stars are probably members of the “southern arc” stream, which was identified by Majewski et al. (2003), namely m17, m18, m20, and m22. Both m17 and m18 are Miras and their calculated distances closely match the Sgr distance of 24.8 kpc determined by Kunder & Chaboyer (2009). However, none of our targets are seen toward the Sgr stream in the southern hemisphere,  ~ 160° and b ~ −50°, mapped by Koposov et al. (2012). This lack of association is not too surprising since Ibata et al. (2001) deduced that only half of the Totten & Irwin (1998) halo C stars belong to the Sgr Stream. Since our list includes none of the Totten & Irwin (1998) halo C stars, it represents an independent sample. Whitelock et al. (1999) estimated that the Sgr galaxy contains at least 100 C stars. They mention that several C stars have large amplitude variations and are probably Miras. However, only 4 of them have known periods.

If the bulk of the Sgr stars were indeed older than 5 Gyr (Bellazzini et al. 2006), then most of the Miras associated with Sgr should have periods shorter than 250 days. Unfortunately, any analysis at present is strongly affected by small number statistics. Four Sgr Miras have periods determined by Whitelock et al. (1999), only one of which has a P < 250 days. From Fig. 5, we see that barely two of the 13 newly discovered halo Miras have such a short period. Therefore, we tentatively propose that the majority of the halo Miras may have an origin that is unrelated to Sgr. We are conducting, however, an ongoing survey of the Sgr carbon stars designed to determine whether they do indeed have a short-period excess, as implied by most of their ages. We are currently surveying a second sample of halo C stars to enhance our data set for the analysis of the period distributions.


Acknowledgments

This publication makes use of data products from the Two Micron All Sky Survey, which is 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. The DENIS project has been partly funded by the SCIENCE and the HCM plans of the European Commission under grants CT920791 and CT940627. It is supported by INSU, MEN and CNRS in France, by the State of Baden-Württemberg in Germany, by DGICYT in Spain, by CNR in Italy, by FFwFBWF in Austria, by FAPESP in Brazil, by OTKA grants F-4239 and F-013990 in Hungary, and by the ESO C&EE grant A-04-046.

References

All Tables

Table 1

id numbers, coordinates and Ks magnitudes of selected targets.

Table 2

Mean magnitudes and colour of program stars.

Table 3

Period, amplitude, and classification.

Table 4

Distances of Miras.

All Figures

thumbnail Fig. 1

Standard deviations of two observations as a function of magnitude. Each point represents the mean of eight measures.

Open with DEXTER
In the text
thumbnail Fig. 2

J magnitude variations of the 23 program stars. Note that the magnitude axis covers two magnitudes. For m11, we plot the K mag.

Open with DEXTER
In the text
thumbnail Fig. 3

I magnitude light variations for irregular or low-amplitude stars.

Open with DEXTER
In the text
thumbnail Fig. 4

Examples of light curves with different χ2, 0.15 for m41 (top) and 0.04 for m31 (bottom).

Open with DEXTER
In the text
thumbnail Fig. 5

Comparison of the period distributions of Miras. The dashed lines define the three age groups.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.