2021A&A...650A.128G

Context. Massive star evolution is still poorly understood, and observational tests are required to discriminate between different implementations of physical phenomena in stellar evolution codes.
Aims. By confronting stellar evolution models with observed properties of blue supergiants, such as pulsations, the chemical composition, and position in the Hertzsprung-Russell diagram, we aim to determine which of the criteria used for convection (Schwarzschild or Ledoux) is best able to explain the observations.
Methods. We computed state-of-the-art stellar evolution models with either the Schwarzschild or the Ledoux criterion for convection. Models are for 14 to 35M_☉ at solar or Large Magellanic Cloud metallicity. For each model, we computed the pulsation properties to know when radial modes are excited. We then compared our results with the position of blue supergiants in the Hertzsprung-Russell diagram, with their surface chemical composition and with their variability.
Results. Our results at Large Magellanic Cloud metallicity shows only a slight preference for the Ledoux criterion over the Schwarzschild one in reproducing, at the same time, the observed properties of blue supergiants, even if the Schwarzschild criterion cannot be excluded at this metallicity. We checked that changing the overshoot parameter at solar metallicity does not improve the situation. We also checked that our models are able to reproduce the position of Galactic blue supergiants in the flux-weighted-gravity-luminosity relation.
Conclusions. We confirm that overall, models computed with the Ledoux criterion are slightly better in matching observations. Our results also support the idea that most Galactic α Cyg variables are blue supergiants from group 2, that is stars that have been through a previous red supergiant phase where they have lost a large amount of mass.