The early evolution of young massive clusters II. The kinematic history of NGC 6618/M 17

Context. Characterising the outcome of the star formation process is key to understand and predict the evolution of stellar populations. Especially the fraction of massive stars in young stellar clusters is of importance as they are the dominant sources of both mechanical and radiative feedback, strongly influencing the thermal and dynamical state of their birth environments, and beyond. Their supernovae may trigger the formation of new generations of stars in neighbouring regions. It turns out that a significant fraction of massive stars escape from their parent cluster via dynamical interactions of single stars and/or multiple stellar systems.Aims. M 17 is the nearest giant H II region hosting a very young and massive cluster: NGC 6618. Our aim is to identify stars brighter than G less than or similar to 21 mag that belong to NGC 6618, including the (massive) stars that may have escaped since its formation, and to determine the cluster distance and age.Methods. The Gaia DR3 database was used to identify members of NGC 6618 based on parallax and proper motion within 9 ' from the cluster centre. We searched for nearby stars in a field of 5 degrees around the cluster centre that may have originated from the cluster, and we determined their transverse velocity, kinematic age, and impact parameter.Results. We identified 42 members of NGC 6618 of which eight have a spectral type of O, with a mean distance of 1675(-18)(+19) pc and a (transversal) velocity dispersion of about 3 km s(-1), and a radial velocity dispersion of similar to 6 km s(-1). Another ten O stars are associated with NGC 6618, but they cannot be classified as members due to poor astrometry and/or high extinction. We have also identified six O star runaways. The relative transverse velocity of these runaways ranges from 10 to 70 km s(-1) and their kinematic age ranges from about 100 to 750 kyr. Given the already established young age of NGC 6618 (less than or similar to 1 Myr), this implies that massive stars are being ejected from the cluster already directly after (or during) the cluster formation process.Conclusions. When constructing the initial mass function, one has to take into account the massive stars that have already escaped from the cluster, that is, about 30% of the O stars of the original population of NGC 6618. The trajectories of the O runaways can be traced back to the central 0.2-0.3 pc region of NGC 6618. The good agreement between the evolutionary and kinematic age of the runaways implies that the latter provides an independent way to estimate (a lower limit to) the age of the cluster.