5. Relative abundances
The mean abundance of iron, [Fe/H]=, is in close agreement with other spectroscopic and photometric determinations: [Fe/H]=-0.05 (Canterna et al. 1986), [Fe/H]=0.06 (Nissen et al. 1987), [Fe/H]=-0.07 (Anthony-Twarog 1987), [Fe/H]=0.04 (Garcia Lopez et al. 1988), [Fe/H]=-0.08 (Friel & Janes 1991), [Fe/H]=0.02 (Friel & Boesgaard 1992), [Fe/H]=-0.09 (Friel & Janes 1993). Along with iron, the abundances of other heavy chemical elements are also very close to solar, as it ought to be in a cluster of almost the same age as the Sun (4.0 Gyr, Dinescu et al. 1995; 4.3 Gyr, Carraro et al. 1996; 4.0 Gyr, Boyle et al. 1998) and located only about 800 pc from it. The small underabundance of zirconium is probably caused by effects of the hyperfine structure.
In Fig. 2 we display the relative abundances of some chemical elements (for stars F84, F141 and F151, the mean values are plotted with double weight for results obtained from the spectra). We also display results published for M 67 by other authors. A detailed spectroscopic analysis for IV-202 was done by Griffin (1975) and for T626 by Griffin (1979), for four stars (F105, F170, F224, and F231) by Cohen (1980) and for F164 and F170 by Foy & Proust (1981). The large scatter of the abundance ratios in these early analyses, when present, is most probably due to observational errors caused by photographic data used.
In our study, among mixing-sensitive chemical elements alterations of carbon, nitrogen and sodium are noticeable. In Sects. 5.1 and 5.2 we describe them in more detail.
5.1. Carbon and nitrogen
There are several possibilities to investigate carbon abundances in stars: a low excitation [C I] line at 8727 Å, a number of high excitation C I lines and molecular lines of CH and . Due to its low excitation potential, the [C I] line should not be sensitive to non-LTE effects and to uncertainties in the adopted model atmosphere parameters, contrary to what may be expected for the frequently used high excitation C I and CH lines. As it can be seen from results by Clegg et al. (1981), carbon abundances obtained using the high excitation C I lines and CH molecular lines are systematically lower than abundances obtained from the 8727 Å [C I] line. The agreement is present only for results obtained from molecular lines. A comprehensive discussion on this subject is presented by Gustafsson et al. (1999) and Samland (1998).
Unfortunately, the [C I] 8727 Å line is not suitable for the analysis in spectra of red giants, especially when a probability of increased strengths of CN molecular lines is present. Lambert and Ries (1977) examined CNO abundances in K-type giants and found that the [C I] line was not trustworthy as a C abundance indicator, devoting an Appendix to the problem. Consequently, we decided to use the Swan (0,1) band head at 5635.5 Å for the analysis. This feature is strong enough in our spectra and is quite sensitive to changes of the carbon abundance (see Fig. 3 for illustration).
The previous high-resolution determination of [C/H] in M 67 giants (F105, F108 and F170) was carried out by Brown (1985), using C2 lines at around 5110 Å and 4730 Å, and gave the mean value [C/H]=-0.26. Our result for these stars is slightly higher [C/H]=-0.19. The mean value for the clump stars is -0.21 dex.
The available study of carbon abundances in dwarf stars of M 67 is that by Friel & Boesgaard (1992). Six high excitation C I lines in the spectral region from 7100 Å to 7120 Å were analysed in three F dwarfs and the mean [C/H]= was obtained. This value, probably, would be by about 0.1 dex higher if the low excitation [C I] line would be used instead. Then the [C/H] value in dwarfs of M 67 would be equal to the solar value. Thus, compared with the Sun and with dwarf stars of M 67, the carbon abundance might be depleted by about 0.2 dex in the stars we analysed.
Another evaluation of the carbon abundance can be done by the comparison with carbon abundances determined for dwarf stars in the galactic disk. Gustafsson et al. (1999), using the forbidden [C I] line, have performed an abundance analysis of carbon in a sample of 80 late F and early G type dwarfs. As is seen from Fig. 4, the ratios of [C/Fe] and [C/O] in our stars lie by about 0.2 dex below the trends obtained for dwarf stars in the galactic disk.
The wavelength interval 7980-8130 Å, with 65 CN lines selected, was analysed in order to determine the nitrogen abundances. The mean nitrogen abundance, as determined from the giants, is [N/H]= and from the clump stars [N/H]=. Neither the carbon depletion, nor the nitrogen enrichment are as large as was reported by Brown (1985). Consequently, the C/N ratios obtained in our work do not request large extra-mixing processes in order to be explained. The mean C/N ratios are lowered to the value of 1.72 in the giants and to the value of 1.44 in the clump stars.
In our work, ratios were determined for all programme stars from the (2,0) line at 8004.728 Å with a laboratory wavelength adopted from Wyller (1966). Fig. 5 illustrates the enhancement of line at 8004 Å in spectra of F164 and F224. The star F224 has the lowest value =8, among stars investigated in our work. For F84, F141 and F151, the line at 8381.06 Å were observed at higher resolution, but due to the weakness of the line we can for F84 and F141 only confirm that the ratios are consistent with the values derived from the 8004.7 Å feature, while for F151 we are not able to derive any useful information of the isotope ratio from this line.
Ratios of were investigated for stars in M 67 by Gilroy (1989) and Gilroy & Brown (1991). There are six stars in common with Gilroy (1989) and one with Gilroy & Brown (1991). Except for two stars (F84 and F170), our ratios agree within errors of uncertainties. The mean difference between our values and these by Gilroy & Brown is equal to . In their study, Gilroy & Brown rule out mixing during the He-core flash because the two stars, F108 and F170, had ratios similar to the clump stars. In our study however, a small difference can be suspected. We find the mean ratios lowered to the value of in the giants and to the value of in the clump stars.
The standard theoretical evolution of the carbon isotopic ratio and carbon to nitrogen ratio along the giant branch was homogeneously mapped by Charbonnel (1994) for stellar masses between 1 and , and for different values of metallicity. For a 1.25 star (approximately the turn-off mass of M 67), with initially solar composition the predicted and ratios at the end of the first dredge-up phase are about 22.7 and 1.6, respectively (Charbonnel 1994, Fig. 2 and Fig. 4 (it is not explaned in the paper why values in Table 2 and Fig. 7 and Fig. 8 are different from those presented in Fig. 2 and Fig. 4, so we decided to use homogeneous ones)). The predicted values are in good agreement with our results for the giant stars. The clump stars, however, may well show an additional decrease caused by an extra mixing process (see Fig. 6).
The stars in our sample, as determined from the Na I lines 5682.64, 6154.23 and 6160.75 Å show a slight overabundance of sodium (see Fig. 7).
Overabundances of sodium in red giants have long been considered as being of a primordial origin (see, e.g., Cottrell & Da Costa 1981). However, the star-to-star variations of Na, the existence of Na versus N correlations, and Na versus O anticorrelations in globular cluster red giants have revealed a possibility of sodium to be produced in red giant stars (Cohen 1978, Peterson 1980, Drake et al. 1992, Kraft et al. 1995, 1997 and references therein). Norris & Da Costa (1995) have concluded that Na variations exist in all clusters, while Al variations are greater in the more metal-poor clusters. Theoretical explanations for the production of Na and Al have been proposed by Sweigart & Mengel (1979), Denissenkov & Denissenkova (1990), Langer & Hoffman (1995), Cavallo et al. (1996) and other studies; however, the origin and extent of the phenomenon is not well understood.
For the stars in our sample, the overabundance of sodium is not followed by a noticeable overabundance of aluminium and underabundance of oxygen. This confirms the conclusion by Norris & Da Costa (1995) that Al variations are not that great as of Na in metal-rich stars. The overabundance of Na could appear due to the deep mixing from layers of the NeNa cycle, which lie higher than ON-processed regions in red giants (cf. Cavallo et al. 1996). Shetrone (1996), from the analysis of red giants in the globular cluster M 71, also concluded that either Al and Na are created in different nucleosynthesis processes, or the NeNa cycle can occur without the ON or MgAl cycles.
5.3. Final remarks
The change in the surface composition of a star ascending the giant branch is predicted by theoretical calculations. When a star evolves up the giant branch its convective envelope deepens and CN-cycle products are mixed to the surface of the evolving star, causing the surface and ratios to drop. These ratios decrease with increasing stellar mass and decreasing metallicity. Extra-mixing processes may become efficient on the red giant branch when stars reach the so-called luminosity function bump and modify the surface abundances (see Charbonnel et al. 1998 for more discussion). In case of M 67, this may happen starting from log=1.64 (Charbonnel 1994), however the first and only evidence on the evolutionary state at whish this non-standard mixing actually becomes effective has to come from observations. The giants F108 and F170 (log), observed in our work, do not show obvious effects of the extra mixing. Other bright M 67 giants such as T626 and IV-202 are not investigated since have quite low membership probabilities (19% and 51%, respectively, as quoted by Sanders 1977) and are rather cool (continuum placement in their spectra would be at best difficult). In our work, the extra-mixing processes seem to show up in the clump stars observed (Fig. 6), however here the He-core flash may be responsible.
The role that the He-core flash may play in producing surface abundance changes still has to be investigated. The theoretical calculations indicate that the nature of nucleosynthesis and mixing depend upon the degree of degeneracy in the He-core and, hence, intensity of the explosion: intermediate flashes produce the most mixing (Despain 1982, Deupree 1986, Deupree & Wallace 1987, Wallace 1988 and references therein). Due to the numerical difficulties in treating such a violent event, the He core-flash remains an event of interest to theorists. The observational data for giants and clump stars of M 67 in our work show a slight increase of abundance changes in more evolved clump stars. New precise observational data are necessary in understanding effects of the He-core flash and other open questions of chemical evolution of stars.
© European Southern Observatory (ESO) 2000
Online publication: August 17, 2000