With the remarkable success of the LVK consortium in detecting binary black hole mergers, it has become possible to use the population properties to constrain our understanding of the progenitor stars’ evolution. The most striking features of the observed primary black hole mass distributions are the extended tail up to 100M⊙ and an excess of masses at 35M⊙. Currently, isolated binary population synthesis have difficulty explaining these features. Using the well-tested bpass detailed stellar binary evolution models to determine mass transfer stability, accretion rates, and remnant masses, we postulate that stable mass transfer with super-Eddington accretion is responsible for the extended tail. These systems are able to merge within the Hubble time due to more stable mass transfer at higher donor masses with higher mass ratios and spin-orbit coupling allowing the orbits to shrink sufficiently. Furthermore, we find that in bpass the 35M⊙ excess is not due to pulsational pair-instability, as previously thought, but a feature caused by stable mass transfer, whose regime is limited by the mass transfer stability, quasi-homogeneous evolution, and stellar winds. These findings are at odds with those from other population synthesis codes but in agreement with other recent studies using detailed binary evolution models.