HIGH-RESOLUTION CCD SPECTRA OF STARS IN GLOBULAR CLUSTERS-V - CARBON, NITROGEN, AND OXYGEN ABUNDANCES IN STARS IN 47-TUC, M4, AND M22

We have obtained spectra of signal-to-noise 60-160 and resolution 20 000 in the blue, red, and infrared of four stars in 47 Tue, four stars in M4, and seven stars in M22. The data have been analyzed using model atmospheres to derive abundances of C, N, O, and Fe. For 47 Tue we find [Fe/H] = -0.85 to -0.9 and [O/Fe] = 0.43 ± 0.07. The carbon and nitrogen abundances show log (N/C) = 0.9 which is indicative of the mixing of substantial amounts of CN-processed matter to the surface. For M4 we find [Fe/H] = -1.3 and [O/Fe] = 0.31 ± 0.06. Excluding star 1412, whose spectrum is peculiar, we find log (N/C) = 0.95 again indicative of deep mixing. For M22 the situation is more complicated due to the likely presence of nonuniform reddening across the cluster. Four of our program stars show a high velocity component of interstellar potassium [also seen in sodium by Cohen (1981)]. They indicate additional reddening of 0.18 in (B - V) but that value is uncertain because the ratio of gas to dust in high velocity clouds may be large. Using a schematic treatment of nonuniform reddening we find [Fe/H] = -1.59 but with the possibility that the difference of 0.25 dex between stars III-3 and IV-102 is actually real. Our mean value of [O/Fe] is ≤ 0.35 but the range is from +0.6 to ≤0.1 which may be real. For uniform reddening, i.e., ignoring the extra cloud, we find [Fe/H] = -1.75 ± 0.06 and a range of -1.6 to -2.0 which indicates that some of the spread is intrinsic. However for uniform reddening the spread in [O/Fe] is reduced slightly to 0.45 with a mean value of ≤0.29. The uncertainties in the reddening, temperature, and model atmospheres introduce uncertainties that are difficult to quantify. A good guess is ±0.10 dex, which implies maximum errors of ±0.2 dex. Our observed nitrogen abundances in M22 range from log ∈(N) = 8.3 to 7.0 which cannot be explained by CNO cycling and mixing. An intrinsic spread of nitrogen abundances and possibly also of carbon must have been present when these stars were on the main sequence. We discuss the possibility that winds from massive WR stars or, more likely, cool supergiants provided self-enrichment during the star-formation period in M22. Finally we discuss the likelihood that the high values of [O/Fe] found by Abia and Rebolo for metal-poor subdwarfs could have characterized the stars in M4 and M22 now seen as giants. The evidence is against it but not very strongly except for two stars in M22, whose sum of C + N + O is 0.55 dex below their suggested initial oxygen abundance. In appendices we discuss the inhomogeneous reddening in M22, the peculiar spectrum of M4 star 1412, and the emission at Ha in all the program stars.