A new look at the binary characteristics of massive stars

We constrain the properties of massive binaries by comparing radial velocity data on early-type stars in Cygnus OB2 with the expectations of Monte Carlo models. Our comparisons test several popular prescriptions for massive binary parameters. We explore a range of true binary fraction, F, a range of power-law slopes, alpha, describing the distribution of companion masses between the limits q(low) and 1, and a range of power-law slopes, alpha, describing the distribution of orbital separations between the limits r(in) and rout. We also consider distributions of secondary masses described by a Miller-Scalo type IMF and by a two-component IMF that includes a substantial twin'' population with M-2 similar or equal to M-1. Several seemingly disparate prescriptions for massive binary characteristics can be reconciled by adopting carefully chosen values for F, rin, and rout. We show that binary fractions F < 0.7 are less probable than F >= 0.8 for reasonable choices of r(in) and r(out). Thus, the true binary fraction is high. For F = 1.0 and a distribution of orbital separations near the canonical Opik's law distribution (i.e., flat; beta = 0), the power-law slope of the mass ratio distribution is alpha = -0.6 to 0.0. For F similar or equal to 0.8, alpha is somewhat larger, in the range -0.4 to 1.0. In any case, the secondary star mass function is inconsistent with a Miller-ScaloYlike IMF unless the lower end is truncated below similar to 2-4 M-circle dot. In other words, massive stars preferentially have massive companions. The best-fitting models are described by a Salpeter or Miller-Scalo IMF for 60% of secondary star masses with the other similar to 40% of secondaries having M-2 similar or equal to M-1, i.e., twins.'' These model parameters simultaneously predict the fraction of Type Ib/c supernovae to be 30%-40% of all core-collapse supernovae, in agreement with recent observational estimates.