The recent LIGO detection of gravitational waves (GW150914), likely originating from the merger of two $sim 30 M_odot$ black holes suggests progenitor stars of low metallicity ($[Z/Z_odot] lesssim 0.3$), constraining when and where the progenitor of GW150914 may have formed. We combine estimates of galaxy properties (metallicity, star formation rate and merger rate) across cosmic time to predict the low redshift black hole - black hole merger rate as a function of present day host galaxy mass, $M_mathrm{gal}$, and the formation redshift of the progenitor system $z_mathrm{form}$ for different progenitor metallicities $Z_mathrm{c}$. At $Z_mathrm{c}=0.1 Z_odot$, the signal is dominated by binaries in massive galaxies with $z_mathrm{form}simeq 2$, with a small contribution from binaries formed around $z_mathrm{form}simeq 0.5$ in dwarf galaxies. For $Z_mathrm{c}=0.01Z_odot$, fast mergers are possible and very recent star formation in dwarfs likely dominates. Additional gravitational wave detections from merging massive black holes will provide constraints on the mass-metallicity relation and massive star formation at high redshifts.
Download