Common envelope (CE) phases in binary systems where the primary star reaches the tip of the red giant branch are discussed as a formation scenario for hot subluminous B-type (sdB) stars. For some of these objects, observations point to very low-mass companions. In hydrodynamical CE simulations with the moving-mesh code AREPO, we test whether low-mass objects can successfully unbind the envelope. The success of envelope removal in our simulations critically depends on whether or not the ionization energy released by recombination processes in the expanding material is taken into account. If this energy is thermalized locally, envelope ejection eventually leading to the formation of an sdB star is possible with companion masses down to the brown dwarf range. For even lower companion masses approaching the regime of giant planets, however, envelope removal becomes increasingly difficult or impossible to achieve. Our results are consistent with current observational constraints on companion masses of sdB stars. Based on a semi-analytic model, we suggest a new criterion for the lowest companion mass that is capable of triggering a dynamical response of the primary star thus potentially facilitating the ejection of a CE. This gives an estimate consistent with the findings of our hydrodynamical simulations.
