Direct collapse to supermassive black hole seeds: the critical conditions for suppression of ${\mathrm{H}}_2$ cooling

Abstract

Observations of high-redshift quasars imply the presence of supermassive black holes already at z~ 7.5. An appealing and promising pathway to their formation is the direct collapse scenario of a primordial gas in atomic-cooling haloes at z ~ 10 - 20, when the ${\mathrm{H}}_2$ formation is inhibited by a strong background radiation field, whose intensity exceeds a critical value, $J_{\mathrm{crit}}$. To estimate $J_{\mathrm{crit}}$, typically, studies have assumed idealized spectra, with a fixed ratio of ${\mathrm{H}}_2$ photo-dissociation rate $k_{{\mathrm{H}}_2}$ to the ${\mathrm{H}}^{-}$ photo-detachment rate $k_{{\mathrm{H}}^{-}}$. This assumption, however, could be too narrow in scope as the nature of the background radiation field is not known precisely. In this work we argue that the critical condition for suppressing the ${\mathrm{H}}_2$ cooling in the collapsing gas could be described in a more general way by a combination of $k_{{\mathrm{H}}_2}$ and $k_{{\mathrm{H}}^{-}}$ parameters. By performing a series of cosmological zoom-in simulations with an encompassing set of $k_{{\mathrm{H}}_2}$ and $k_{{\mathrm{H}}^{-}}$, we examine the gas flow by following evolution of basic parameters of the accretion flow. We test under what conditions the gas evolution is dominated by ${\mathrm{H}}_2$ and/or atomic cooling. We confirm the existence of a critical curve in the $k_{{\mathrm{H}}_2}-k_{{\mathrm{H}}^{-}}$ plane, and provide an analytical fit to it. This curve depends on the conditions in the direct collapse, and reveals domains where the atomic cooling dominates over the molecular cooling. Furthermore, we have considered the effect of ${\mathrm{H}}_2$ self-shielding on the critical curve, by adopting three methods for the effective column density approximation in ${\mathrm{H}}_2$. We find that the estimate of the characteristic length-scale for shielding can be improved by using $\hat{I}{»}_{Jeans25}$, which is 0.25 times that of the local Jeans length.

10 pages, 5 figures, accepted for publication in MNRAS