A scaled-up planetary system around a supernova progenitor

Virtually all known exoplanets reside around stars with $M<2.3~M_\odot$; to clarify if the dearth of planets around more massive stars is real, we launched the direct-imaging B-star Exoplanet Abundance STudy (BEAST) survey targeting B stars ($M>2.4~M_\odot$) in the young (5-20 Myr) Scorpius-Centaurus association (Sco-Cen). Here we present the case of a massive ($M \sim 9~M_\odot$) BEAST target, $\mu^2$ Sco. Based on kinematic information, we found that $\mu^2$ Sco is a member of a small group which we label Eastern Lower Scorpius, refining in turn the precision on stellar parameters. Around this star we identified a robustly detected substellar companion ($14.4\pm 0.8 M_J$) at a projected separation of $290\pm 10$ au, and a probable second object ($18.5\pm 1.5 M_J$) at $21\pm 1$ au. The planet-to-star mass ratios of these objects are similar to that of Jupiter to the Sun, and their irradiation is similar to those of Jupiter and Mercury, respectively. The two companions of $\mu^2$ Sco are naturally added to the giant planet b Cen b recently discovered by BEAST; although slightly more massive than the deuterium burning limit, their properties resemble those of giant planets around less massive stars and they are better reproduced by a formation under a planet-like, rather than a star-like scenario. Irrespective of the (needed) confirmation of the inner companion, $\mu^2$ Sco is the first star that would end its life as a supernova that hosts such a system. The tentative high frequency of BEAST discoveries shows that giant planets or small-mass brown dwarfs can form around B stars. When putting this finding in the context of core accretion and gravitational instability, we conclude that the current modeling of both mechanisms is not able to produce this kind of companion. BEAST will pave the way for the first time to an extension of these models to intermediate and massive stars. (abridged)