The spectroscopic binary fraction of the young stellar cluster M17

Context. Significant progress has been made toward understanding the formation of massive ( M > 8 M ⊙ ) binaries in close orbits (with periods of less than a month). Some of the observational studies leading to this progress are the detection of a very low velocity dispersion among the massive stars in the young region M17 and the measurement of a positive trend of velocity dispersion with age in Galactic clusters. The velocity dispersion observed in M17 could be explained either by the lack of binaries among the stars in this region, which implies the highly unlikely scenario of a different formation mechanism for M17 than for other Galactic regions, or by larger binary separations than typically observed, but with a binary fraction similar to other young Galactic clusters. The latter implies that, over time, the binary components migrate toward each other. This is in agreement with the finding that the radial velocity dispersion of young Galactic clusters correlates positively with their age. Aims. We aim to determine the origin of the strikingly low velocity dispersion by determining the observed and intrinsic binary fraction of massive stars in M17 through multi-epoch spectroscopy. Methods. We performed a multi-epoch spectroscopic survey consisting of three epochs separated by days and months, respectively. We complemented this survey with existing data covering timescales of years. We determined the radial velocity of each star at each epoch by fitting the stellar absorption profiles. The velocity shifts between epochs were used to determine whether a close companion is present. Results. We determined an observed binary fraction of 27% and an intrinsic binary fraction of 87%, consistent with that of other Galactic clusters. We conclude that the low velocity dispersion is due to a large separation among the young massive binaries in M17. Our result is in agreement with a migration scenario in which massive stars are born in binaries or higher-order systems at large separation and harden within the first million years of evolution. Such an inward migration may either be driven by interaction with a remnant accretion disk or with other young stellar objects present in the system, or by dynamical interactions within the cluster. Our results imply that possibly both dynamical interactions and binary evolution are key processes in the formation of gravitational wave sources.