With approximately 50 binary black hole events detected by LIGO/Virgo to date and many more expected in the next few years, gravitational-wave astronomy is shifting from individual-event analyses to population studies. We perform a hierarchical Bayesian analysis on the GWTC-2 catalog by combining several astrophysical formation models with a population of primordial black holes. We compute the Bayesian evidence for a primordial population compared to the null hypothesis, and the inferred fraction of primordial black holes in the data. We find that these quantities depend on the set of assumed astrophysical models: the evidence for primordial black holes against an astrophysical-only multichannel model is decisively favored in some scenarios, but it is significantly reduced in the presence of a dominant stable-mass-transfer isolated formation channel. The primordial channel can explain mergers in the upper mass gap such as GW190521, but (depending on the astrophysical channels we consider) a significant fraction of the events could be of primordial origin even if we neglected GW190521. The tantalizing possibility that LIGO/Virgo may have already detected black holes formed after inflation should be verified by reducing uncertainties in astrophysical and primordial formation models, and it may ultimately be confirmed by third-generation interferometers.
Download