Abstract
One of the seemingly strongest constraints on the fraction of dark matter in the form of primordial black holes (PBH) of 
(10) M⊙ relies on the merger rate inferred from the binary BH merger events detected by LIGO/Virgo. The robustness of these bounds depends however on the accuracy with which the formation of PBH binaries in the early Universe can be described. We revisit the standard estimate of the merger rate, focusing on a couple of key ingredients: the spatial distribution of nearest neighbours and the initial clustering of PBHs associated to a given primordial power spectrum. Overall, we confirm the robustness of the results presented in the literature in the case of a narrow mass function (which constrain the PBH fraction of dark matter to be fPBH≲ 0.001–0.01). The initial clustering of PBHs might have an effect tightening the current constraint, but only for very broad mass functions, corresponding to wide bumps in the primordial power spectra extending at least over a couple of decades in k-space.