2018A&A...613A..12M

Context. Massive stars play a key role in the evolution of galaxies and our Universe. Aims. Our goal is to compare observed and predicted properties of single Galactic O stars to identify and constrain uncertain physical parameters and processes in stellar evolution and atmosphere models. Methods. We used a sample of 53 objects of all luminosity classes and with spectral types from O3 to O9.7. For 30 of these, we determined the main photospheric and wind parameters, including projected rotational rates accounting for macroturbulence, and He and N surface abundances, using optical spectroscopy and applying the model atmosphere code FASTWIND. For the remaining objects, similar data from the literature, based on analyses by means of the CMFGEN code, were used instead. The properties of our sample were then compared to published predictions based on two grids of single massive star evolution models that include rotationally induced mixing. Results. Any of the considered model grids face problem in simultaneously reproducing the stellar masses, equatorial gravities, surface abundances, and rotation rates of our sample stars. The spectroscopic masses derived for objects below ∼30M_☉ tend to be smaller than the evolutionary ones, no matter which of the two grids have been used as a reference. While this result may indicate the need to improve the model atmosphere calculations (e.g. regarding the treatment of turbulent pressure), our analysis shows that the established mass problem cannot be fully explained in terms of inaccurate parameters obtained by quantitative spectroscopy or inadequate model values of V_rot on the zero age main sequence. Within each luminosity class, we find a close correlation of N surface abundance and luminosity, and a stronger N enrichment in more massive and evolved O stars. Additionally, we also find a correlation of the surface nitrogen and helium abundances. The large number of nitrogen-enriched stars above ∼30M_☉ argues for rotationally induced mixing as the most likely explanation. However, none of the considered models can match the observed trends correctly, especially in the high mass regime. Conclusions. We confirm mass discrepancy for objects in the low mass O-star regime. We conclude that the rotationally induced mixing of helium to the stellar surface is too strong in some of the models. We also suggest that present inadequacies of the models to represent the N enrichment in more massive stars with relatively slow rotation might be related (among other issues) to problematic efficiencies of rotational mixing. We are left with a picture in which invoking binarity and magnetic fields is required to achieve a more complete agreement of the observed surface properties of a population of massive main-sequence stars with corresponding evolutionary models.