Most carbon-enhanced metal-poor ( CEMP ) stars are thought to result from past mass transfer of He-burning material from an asymptotic giant branch ( AGB ) star to a low-mass companion star , which we now observe as a CEMP star .
Because AGB stars of intermediate mass efficiently cycle carbon into nitrogen in their envelopes , the same evolution scenario predicts the existence of a population of nitrogen-enhanced metal-poor ( NEMP ) stars , with \mathrm { [ N / Fe ] } > 1 and \mathrm { [ N / C ] } > 0.5 .
Such NEMP stars are rare , although their occurrence depends on metallicity : they appear to be more common at \mathrm { [ Fe / H ] } < -2.8 by about a factor of 10 compared to less metal-poor stars .
We analyse the observed sample of metal-poor stars with measurements of both carbon and nitrogen to derive firm constraints on the occurrence of NEMP stars as a function of metallicity .
We compare these constraints to binary population synthesis calculations in which we vary the initial distributions of mass , mass ratio and binary orbital periods .
We show that the observed paucity of NEMP stars at \mathrm { [ Fe / H ] } > -2.8 does not allow for large modifications in the initial mass function , as have been suggested in the literature to account for the high frequency of CEMP stars .
The situation at lower metallicity is less clear , and we do not currently have stellar models to perform this comparison for \mathrm { [ Fe / H ] } < -2.8 .
However , unless intermediate-mass AGB stars behave very differently at such low metallicity , the observed NEMP frequency at \mathrm { [ Fe / H ] } < -2.8 appears incompatible with the top-heavy forms of the initial mass function suggested in the literature .