Constraints on the initial mass function of the first stars

Motivated by theoretical predictions that the first stars were predominantly very massive, we investigate the physics of the transition from an early epoch dominated by massive Pop III stars to a later epoch dominated by familiar low-mass Pop II/I stars by means of a numerically generated catalogue of dark matter haloes coupled with a self-consistent treatment of chemical and radiative feedback. Depending on the strength of the chemical feedback, Pop III stars can contribute a substantial fraction (several per cent) of the cosmic star formation activity even at moderate redshifts, z≈ 5. We find that the three z≈ 10 sources tentatively detected in Near Infrared Camera and Multi-Object Spectrometer (NICMOS) Ultra Deep Fields (UDFs) should be powered by Pop III stars, if these are massive; however, this scenario fails to reproduce the derived Wilkinson Microwave Anisotropy Probe (WMAP) electron scattering optical depth. Instead, both the UDFs and WMAP constraints can be fulfilled if stars at any time form with a more standard, slightly top-heavy, Larson initial mass function.