The abundance spread in the giants of NGC 6752.

A spectroscopic survey has been performed of 69 stars on or near the giant branches of the metal-poor globular cluster NGC 6752. Our basic results are: (i) There is a large range in the strength of the violet cyanogen bands on the red giant branch, with the available evidence strongly suggesting that the distribution is bimodal. (ii) The cyanogen variations on the giant branch appear to be accompanied by an anticorrelated variation in the abundance of the CH molecule. Spectrum synthesis analysis of a (CN strong)/(CN weak) pair of stars for which relatively high resolution data are available shows that there is a variation of Δ[N/A] ∼+0.9, and Δ[C/A] ∼-0.3, indicative of the CN cycle. (iii) On the red giant branch there are variations in the strength of the lines of Al I which correlate positively with the cyanogen variations. The size of the variations is consistent with the hypothesis that the same phenomenon has occurred in NGC 6752 and ω Centauri, but to a much smaller extent in the former. (iv) On the asymptotic giant branch (AGB), the features of CH are weaker than on the red giant branch at the same color or magnitude, and there are no examples of stars in the strong CN group. Spectrum synthesis suggests that the behavior of the CH features is consistent, on the average, with the effective temperature and gravities of the AGB stars, but that the absence of strong CN stars cannot be explained in this way. We set an upper limit of Δ[C/H] ∼0.3 to the possible range of carbon on the AGB at log L/L -- stars: individual: ∼2.3, and between this group and stars of similar color on the red giant branch. (v) Most of the stars on the anomalously low luminosity end of the AGB are not members of NGC 6752. Two stars, (CS 41 and CS 44), however, deserve further study, since they could be examples of partially mixed stars.

No definitive statement can be made concerning the origin of the abundance anomalies. if mixing is responsible, the data require this process to have operated in stars on the giant branch at a luminosity of log L/L_sun ∼1.6 (the level of the horizontal branch), and that at least half the stars have mixed. The Al I variations are difficult to explain as the result of mixing. if, on the other hand, the variations are primordial, it is difficult to envisage a succession of stellar generations which can lead to the two observed, roughly equal, populations in which the nitrogen strong, Al I strong group is deficient in carbon by a factor of roughly 2. One somewhat ad hoc primordial model which is consistent with the observations is that proton NGC 6752 was made up of two cells of different chemical histories which merged to leave the cluster as we see it today.

We consider two hypotheses which are consistent with the present observations and the existence of the well known gap on the horizontal branch of NGC 6752. in the first we suggest that, when star formation ceased in the cluster, there were two groups of stars having not only the observed carbon and nitrogen properties, but also a difference in helium abundance, ΔY ∼0.05, in the sense that the nitrogen strong group has enhanced helium. This difference in helium leads to a mass difference of ∼0.07 M_sun at the main sequence turnoff, which, together with our current knowledge of horizontal branch morphology, provides an explanation of both the gap on the horizontal branch and the lack of CN strong stars on the AGB. (The high helium, high CN group does not ascend the giant branch for a second time.) The second hypothesis supposes that mixing is responsible for the abundance anomalies, and that this process is associated with greater mass loss, leading to two mass groups on the horizontal branch. (Here, too, the CN strong, low mass group does not ascend the giant branch for a second time.)