2022MNRAS.516.4052H

The most massive stars dominate the chemical enrichment, mechanical and radiative feedback, and energy budget of their host environments. Yet how massive stars initially form and how they evolve throughout their lives is ambiguous. The mass loss of the most massive stars remains a key unknown in stellar physics, with consequences for stellar feedback and populations. In this work, we compare grids of very massive star (VMS) models with masses ranging from 80 to 1000 M_⊙, for a range of input physics. We include enhanced winds close to the Eddington limit as a comparison to standard O-star winds, with consequences for present-day observations of ∼50-100 M_⊙ stars. We probe the relevant surface H abundances (X_s) to determine the key traits of VMS evolution compared to O stars. We find fundamental differences in the behaviour of our models with the enhanced-wind prescription, with a convergence on the stellar mass at 1.6 Myr, regardless of the initial mass. It turns out that X_s is an important tool in deciphering the initial mass due to the chemically homogeneous nature of VMS above a mass threshold. We use X_s to break the degeneracy of the initial masses of both components of a detached binary, and a sample of WNh stars in the Tarantula Nebula. We find that for some objects, the initial masses are unrestricted and, as such, even initial masses of the order 1000 M_⊙ are not excluded. Coupled with the mass turnover at 1.6 Myr, X_s can be used as a 'clock' to determine the upper stellar mass.