Runaways electron beams formed during tokamak disruptions pose a significant threat for future tokamak operations due to the highly localized heat loads they impart on plasma-facing components at impact. A method to dissipate runaway beams safely was recently discovered on DIII-D [1] and JET [2]. I consists in injecting large amounts of hydrogen or deuterium in a runaway beam. This provokes the recombination of the cold companion plasma coexisting with the beam. The recombination leads to a fast MHD instability when the beam collapses, and the absence of regeneration of secondary runaway electrons thanks to the low impurity content. The conditions under which the scenario is achievable need to be understood to apply it to ITER and beyond. Recent experiments at JET have shown that the composition of the companion plasma is essential for a benign termination. Too much argon (used to generate the beam in the first place) leads to regeneration of small beams during the final collapse or an incomplete dissipation of the runaway beam by the fast MHD event. Incomplete dissipations where too much argon is injected are found to be similar to unmitigated references, and sometimes even worse. They are sometimes associated with reionization of the companion plasma prior the final collapse. This is sufficient for non-benign termination, but is not a necessary condition: regeneration of small beams leading to significant heat loads are observed even in the absence of pre-collapse reionization. Injecting more hydrogenic material can restore the benign character, up to a certain point. It is indeed shown that an upper limit for the hydrogenic pressure exists, reverting the effect beyond a certain point. Mechanisms are proposed to explain this behaviour. Additionally, high pre-disruption current is found to require more hydrogenic material to achieve benign termination. The picture is however complex, as higher pre-disruption current comes with larger early current quench MHD, higher vertical instability, higher avalanche gain and higher runaway electron energy. Runaway regeneration physics during the collapse and the characterization of MHD events leading to runaway dissipation are discussed. [1] Paz-Soldan et al., PPCF 61 054001 (2019), [2] Reux et al., PRL 126 175001 (2021).