Statistical Analyses of Massive Stars and Stellar Populations

Abstract

Massive stars, i.e. stars more massive than about ten times that of the Sun, are key agents in the Universe. They synthesise many of the chemical elements that are so important for life on Earth, helped reionising the early Universe and end their lives in spectacular supernova explosions that are visible out to large distances. Because of their important role for much of astrophysics, accurate and reliable stellar evolution models are essential. However, recent developments regarding wind mass loss rates, internal mixing processes and duplicity seriously challenge our understanding of massive stars and stellar populations.<br /> It is now established that most, if not all, massive stars reside in binaries or higher order multiple systems such that more than two-thirds of all massive stars are expected to interact through mass transfer with a binary companion during their lives. We investigate the consequences of this finding for coeval stellar populations and show that the most massive stars in star clusters are likely all rejuvenated binary products that may seriously bias the determination of cluster ages. We further find that wind mass loss from stars and binary mass transfer leave their fingerprints in the high mass end of stellar mass functions. Using these fingerprints, we are able to age-date the young Arches and Quintuplet star clusters with far reaching consequences for the stellar upper mass limit that we revise to be in the range 200–500 solar masses. Such an upper mass limit would allow for pair-instability supernovae in the local Universe.<br /> Large spectroscopic surveys such as the VLT-FLAMES Tarantula Survey (VFTS) deliver many atmospheric parameters of hundreds of massive stars that are ideal to probe and calibrate the physics used in stellar models. To make use of such data, we develop the Bayesian code BONNSAI and make it available through a web-interface. With BONNSAI we are able to match all available observables of stars including their uncertainties simultaneously to stellar models to determine fundamental stellar parameters like mass and age while taking prior knowledge such as initial mass functions into account. A key aspect of BONNSAI is that it allows us to identify stars that cannot be reproduced by stellar models. We use BONNSAI to test the Milky Way stellar models of Brott et al. (2011) with eclipsing binaries and find good agreement.<br /> We further use BONNSAI in combination with data from the VFTS to study the massive O and WNh stars in one of the largest starburst regions known to date, 30 Doradus. In particular we investigate their age distributions to learn about their formation history. The VFTS stars in our sample are mostly found outside clusters and associations and we do not find spatially coherent age patterns. The stars either formed continuously over the 30 Doradus field or in clusters and associations from where they were ejected to their current positions. The age distributions of our sample stars are consistent with the existence of at least two to four coeval stellar populations which would imply that most of the VFTS stars in our sample formed in clusters and associations.