2000A&A...364..563V

The recent finding by Chevalier & Ilovaisky (1998A&A...330..201C) from Hipparcos observations that OB-supergiant X-ray binaries have relatively large runaway velocities (mean peculiar tangential velocity <v_tr≥42±14km/s), whereas Be/X-ray binaries have low runaway velocities (<v_tr≥15±6km/s), provides confirmation of the current models for the formation of these two types of systems. These predict a difference in runaway velocity of this order of magnitude. This difference basically results from the variation of the fractional helium core mass as a function of stellar mass, in combination with the conservation of orbital angular momentum during the mass transfer phase that preceded the formation of the compact object in the system. This combination results into: (i) Systematically narrower pre-supernova orbits in the OB-supergiant systems than in the Be-systems, and (ii) A larger fractional amount of mass ejected in the supernovae in high-mass systems relative to systems of lower mass. Regardless of possible kick velocities imparted to neutron stars at birth, this combination leads to a considerable difference in average runaway velocity between these two groups. If one includes the possibility for non-conservative mass transfer the predicted difference between the runaway velocity of the two groups becomes even more pronounced. The observed low runaway velocities of the Be/X-ray binaries confirm that in most cases not more than 1 to 2M_☉ was ejected in the supernovae that produced their neutron stars. This, in combination with the -on average- large orbital eccentricities of these systems, indicates that their neutron stars must have received a velocity kick in the range 60-250km/s at birth. The considerable runaway velocity of Cygnus X-1 (v_tr=50±15km/s) shows that also with the formation of a black hole considerable mass ejection takes place.