When are LIGO/Virgo's Big Black-Hole Mergers?

We study the evolution of the binary black hole (BBH) mass distribution across cosmic time. The second gravitational-wave transient catalog (GWTC-2) from LIGO/Virgo contains BBH events out to redshifts $z \sim 1$, with component masses in the range $\sim5$--$80\,M_\odot$. In this catalog, the biggest black holes, with $m_1 \gtrsim 45\,M_\odot$, are only found at the highest redshifts, $z \gtrsim 0.4$. We ask whether the absence of high-mass BBH observations at low redshift indicates that the astrophysical BBH mass distribution evolves: the biggest BBHs only merge at high redshift, and cease merging at low redshift. Alternatively, this feature might be explained by gravitational-wave selection effects. Modeling the BBH primary mass spectrum as a power law with a sharp maximum mass cutoff (Truncated model), we find that the cutoff increases with redshift ($> 99.9\%$ credibility). An abrupt cutoff in the mass spectrum is expected from (pulsational) pair instability supernova simulations; however, GWTC-2 is only consistent with a Truncated mass model if the location of the cutoff increases from $45^{+13}_{-5}\,M_\odot$ at $z < 0.4$ to $80^{+16}_{-13}\,M_\odot$ at $z > 0.4$. Alternatively, if the primary mass spectrum has a break in the power law (Broken power law) at ${38^{+15}_{-8}\,M_\odot}$, rather than a sharp cutoff, the data are consistent with a non-evolving mass distribution. In this case, the overall rate of mergers, at all masses, increases with increasing redshift. Future observations will confidently distinguish between a sharp maximum mass cutoff that evolves with redshift and a non-evolving mass distribution with a gradual taper, such as a Broken power law. After $\sim 100$ BBH merger observations, a continued absence of high-mass, low-redshift events would provide a clear signature that the mass distribution evolves with redshift.