The Origin of Close Massive Binaries in the M17 Star-Forming Region

Context. Spectroscopic multiplicity surveys of O stars in young clusters and OB associations have revealed that a large portion (similar to 70%) of these massive stars (M-i > 15 M-circle dot) belong to close and short-period binaries (with a physical separation of less than a few astronomical units). Follow-up VLT(I) high-angular-resolution observations led to the detection of wider companions (up to d similar to 500 au), increasing the average companion fraction to >2. Despite the recent and significant progress, the formation mechanisms leading to such close massive multiple systems remain to be elucidated. As a result, young massive close binaries (or higher-order multiple systems) are unique laboratories for determining the pairing mechanism of high-mass stars. Aims. We present the first VLTI /GRAVITY observations of six young O stars in the M17 star-forming region (less than or similar to 1 Myr) and two additional foreground stars. VLTI /GRAVITY provides the K-band high-angular-resolution observations needed to explore the close environment of young O-type stars, and, as such, o ffers an excellent opportunity to characterise the multiplicity properties of the immediate outcome of the massive star formation process. Methods. From the interferometric model fitting of visibility amplitudes and closure phases, we search for companions and measure their positions and flux ratios. Combining the resulting magnitude di fference with atmosphere models and evolutionary tracks, we further constrain the masses of the individual components. Results. All six high-mass stars are in multiple systems, leading to a multiplicity fraction of 100% and yielding a 68% confidence interval of 94-100%. We detect a total of nine companions with separations of up to 120 au. Including previously identified spectroscopic companions, the companion fraction of the young O stars in our sample reaches 2:3 +/- 0:6. The derived masses span a wide range, from 2.5 to 50 M-circle dot, with a great tendency towards high-mass companions. However, we do not find a significant correlation between the mass of the companions and their separation. Conclusions. While based on a modest sample, our results clearly indicate that the origin of the high degree of multiplicity is rooted in the star formation mechanism of the sample stars. No clear evidence for one of the competing concepts of massive star formation (core accretion or competitive accretion) could be found. However, given that we find all of the companions within similar to 120 au, our results are compatible with migration as a scenario for the formation of close massive binaries.