Fine structure of the red giant clump from Hipparcos data, and distance determinations based on its mean magnitude

The I-band brightnessMI of clump stars is a possible distance indicator for stellar populations. Investigations have shown that MI is almost insensitive to the (V − I) colour within the clump. Based on this, it was assumed that MI was insensitive to age and composition of the stellar population and therefore an ideal standard candle, which could be calibrated with local clump stars, whose absolute brightness is known from Hipparcos parallaxes. This resulted in a distance to the LMC about 15% shorter than usually determined. In the present paper we show that with a population synthesis approach we can reproduce the constancy of MI with colour for the local Hipparcos clump sample. Nevertheless, MI is not a constant among different populations, but depends on metallicity. As a result, the determined distance modulus to the LMC of 18.28±0.18 mag is in better agreement with standard values. This resolves, at least partially, the controversial result obtained by the assumption of a universal value for MI . Particularly remarkable is our prediction that stars slightly heavier than the maximum mass for developing degenerate He cores, MHef , should define a secondary, clumpy structure, about 0.3 mag below the bluest extremity of the red clump. Both features are well separated in the MI vs. V − I diagram of metal-rich stellar populations. Indeed, this secondary clump can be clearly identified in the Hipparcos database of stars with reliable I photometry and parallax errors smaller than 10%. Since the stars in this feature should represent a narrow range of masses, their mass determination, e.g. by the use of binary systems, can provide information about the efficiency of convective overshooting from stellar cores. Our investigation demonstrates that the RGB clump cannot be used as a distance indicator without proper knowledge and modelling of the population under investigation. In addition, there remain unsolved problems in the models, such as correct bolometric corrections and colour transformations.