The first phase of mass transfer in low-mass binaries: Neither stable nor a common envelope

Context. The masses of the white dwarfs in a binary carry information about previous mass-transfer phases. The core mass – radius relation of low-mass giants gives the size of the progenitor of a helium white dwarf at the moment it last filled its Roche lobe. Previously, we used this information for a few observed systems to propose a new mass-transfer type based on an angular momentum balance. Aims. Our aim is to investigate if stable mass transfer instead of the angular-momentum prescription is consistent with the observed double-helium white-dwarf masses. Methods. We reconstructed the progenitor evolution of observed double-helium white dwarfs using the core mass – radius relation and evaluated if the periods at the start of the second phases of mass transfer are consistent with the outcome of stable mass transfer. More generally, we calculated the mass distribution of double-helium white dwarfs for three different progenitor scenarios: double common envelope (with parameter αλ ), angular-momentum prescription (with parameter γ ), and stable mass transfer. Results. We find that the observed systems are generally not consistent with stable mass transfer. Stable mass transfer leads to a tight correlation between the two white dwarf masses in a binary that is not consistent with the observed mass distribution.Double-common-envelope evolution is a particularly poor fit to the observations. The angular-momentum prescription can populate the observed mass distribution, but not perfectly. Conclusions. We conclude that the first phase of mass transfer initiated on the red giant branch in low-mass systems does not generally proceed as stable mass transfer nor as a common envelope, and thus it is poorly understood. This may be related to the fact that for many observed binaries that have finished the first phase of mass transfer the orbit is eccentric, which is an unexpected outcome of mass transfer.