Unveiling early black hole growth with multifrequency gravitational wave observations

Third-generation ground-based gravitational wave interferometers, like the Einstein Telescope (ET), Cosmic Explorer, and the Laser Interferometer Space Antenna (LISA), will detect coalescing binary black holes over a wide mass spectrum and across all cosmic epochs. We track the cosmological growth of the earliest light and heavy seeds that swiftly transit into the supermassive domain using a semi-analytical model for the formation of quasars at z = 6.4, 2, and 0.2, in which we follow black hole coalescences driven by triple interactions. We find that light-seed binaries of several 10(2) M-circle dot are accessible to ET with a signal-to-noise ratio (S/N) of 10-20 at 6 < z < 15. They then enter the LISA domain with larger S/N as they grow to a few 10(4) M-circle dot. Detecting their gravitational signal would provide first time evidence that light seeds form, grow, and dynamically pair during galaxy mergers. The electromagnetic emission of accreting black holes of similar mass and redshift is too faint to be detected even for the deepest future facilities. ET will be our only chance to discover light seeds forming at cosmic dawn. At 2 < z < 8, we predict a population of 'starved binaries', long-lived marginally growing light-seed pairs, to be loud sources in the ET bandwidth (S/N > 20). Mergers involving heavy seeds (similar to 10(5)-10(6) M-circle dot) would be within reach up to z = 20 in the LISA frequency domain. The lower z model predicts 11.25 (18.7) ET (LISA) events per year, overall.