The Carbon, Nitrogen, and Oxygen Abundances in Four K Giants

The abundances, relative to hydrogen, of carbon, nitrogen, and oxygen are determined in the K giants a Boo, a Ser, /3 Gem, and e Peg. Use is made of the lines of C2, CN, C i, and [0 i] from spectrograms with dispersions of 2 and 6 A mm' Well-iterated LTE model atmospheres are employed in the reduction, together with a method of analysis which is independent of oscillator strengths. The Sun is used as the comparison star Weak-line theory is used to predict band intensities and equivalent widths of lines. A graphic-solution technique is employed to obtain the carbon and oxygen abundances, and the nitrogen abundance is then deduced from the CN data. Very rough lower bounds on the 12C/'~C ratio are de- termined for each of the four stars by means of rotational lines of the `3CN red (4,0) or (2,0) band. A discrepancy between the observed intensities of the CN red bands and those anticipated on the basis of available oscillator strengths is noted. The final atmospheric abundances, particularly the nitrogen enhancements, are in good agreement with recent predictions based on detailed calculations of stellar evolution, and hence they constitute further evidence that our understanding of stellar nuclear burning and evolution is correct. I. INTRODUCTION In recent years considerable progress has been made on the problem of determining metal abundances in K giants. However, even for stars as hot as the Sun, molecular formation has begun to deplete free C, N, and 0 (Lambert 1968; Schadee 1968), and the result is that no reliable determination of C, N, and 0 (hereinafter, CNO) abundances are available for stars of type K and later. In view of the strong evolutionary implications of the CNO abundances, this gap in our knowledge must be filled in. The present study is a step to this end. In the past, various approximations have been made regarding the gas pressures, temperatures, and abundances of band-forming regions, and considerable information has been deduced (see most recently Tsuji 1964; Dolan 1965; Spinrad and Vardya 1966; Conti et al. 1967; Morris and Wyller 1967; Ball and Pagel 1967). Schadee's (1968) study carries the problem to a higher level of accuracy in that he employs a grid of model atmospheres in deducing the band intensities which he has chosen to investigate. His study again emphasizes that shifts in the depths of formation of the different features due to variations in temperature, pressure, and composition introduce considerable uncertainty in one-slab analyses. Thus it is apparent that the oniy approach of real value will be to employ accurate model atmospheres. One difficulty in this approach is that good model atmospheres are not available for stars later than type K. However, there is the range of stars between solar-type and late type K which may be handled, and for these temperatures a further simplification arises from the fact that at most the ten diatomic intercombinations of H, C, N, and 0 need be considered in calculating the depletion of CNO, as long as the abundances are not too peculiar. Further, under most circumstances, several of these ten may be neglected. In the analysis reported here we took advantage of the above situation. We used spectrograms with dispersions of 2 and 6 A mm1 taken primarily by Dr. G. Wallerstein at the Lick Observatory, together with results from Griffin (1968), well-iterated LTE atmospheres generated with a model-atmosphere program kindly made available by * Present address: Space Physics Group, The Boeing Company, Seattle, Washingto