Unveiling the stellar origins of high-redshift black hole mergers with next-generation gravitational-wave observatories

Next-generation (XG) gravitational-wave (GW) observatories will reach farther into the universe than currently possible, unveiling previously hidden stellar binary populations that lead to black hole (BH) mergers. Population III (Pop III) stars formed in the low-metallicity early universe are one such target whose astrophysical properties are currently poorly constrained. We develop a method to infer the initial mass function (IMF) and star formation rate density (SFRD) of Pop III stars directly from the population of gravitational-wave sources. Using machine learning, we build an astrophysics-informed mapping from the IMF to BH properties from population-synthesis simulations while accounting for training uncertainty in the model. Combined with a Gaussian-process prior on the SFRD, we analyze mock catalogs of sources detectable in XG that comprise subpopulations of Pop I/II and Pop III BBH. We demonstrate our ability to separate Pop I/II from Pop III BBH at redshifts where both contribute to the BH merger rate. Moreover, we show we can effectively constrain the Pop III progenitor properties straight from GW data. Our method incorporates astrophysical simulations directly into the GW inference pipeline and informs the potential for XG observatories to reveal the details of high-redshift stellar populations.