The impact of black hole seeding in cosmological simulations

Most cosmological simulations of galaxy evolution include active galactic nucleus (AGN) feedback, typically seeding black holes with masses of $\geq 10^5\, h^{-1}\, \rm{M}_{\odot}$ when the dark matter halo exceeds a given threshold mass. Taylor & Kobayashi (2014) introduced a new model, which seeds black holes at $10^3\, h^{-1}\, \rm{M}_{\odot}$ based on gas properties alone, and motivated by the channel of black hole formation due to the collapse of the most massive first stars in the Universe. We compare the black hole mass when the dark matter halo mass is $10^{10}\, h^{-1}\, \rm{M}_{\odot}$ between the different seeding methods. We find that seeding based upon gas properties gives a distribution of black hole masses with $\langle \log M_{\rm{BH}} {/ \rm{M}_{\odot}} \rangle = (5.18 \pm 0.54)$ when dark matter halo mass is $10^{10}\, h^{-1}\, \rm{M}_{\odot}$, consistent with the {seeding criteria} used in other simulations. However, the evolution of individual galaxies can be strongly affected by the different seeding mechanisms. We also find that the mean value of the distribution of black hole masses at a given halo mass evolves over time, with higher masses at higher redshifts, indicative of downsizing. Our results can inform more physically motivated black hole and AGN feedback models in cosmological simulations and semi-analytic models.