Modelling the formation of double white dwarfs

We investigate the formation of the ten double-lined double white dwarfs that have been observed so far. A detailed stellar evolution code is used to calculate grids of single-star and binary models and we use these to reconstruct possible evolutionary scenarios. We apply various criteria to select the acceptable solutions from these scenarios. We confirm the conclusion of Nelemans et al. (2000) that formation via conservative mass transfer and a common envelope with spiral-in based on energy balance or via two such spiral-ins cannot explain the formation of all observed systems. We investigate three different prescriptions of envelope ejection due to dynamical mass loss with angular-momentum balance and show that they can explain the observed masses and orbital periods well. Next, we demand that the age difference of our model is comparable to the observed cooling-age difference and show that this puts a strong constraint on the model solutions. However, the scenario in which the primary loses its envelope in an isotropic wind and the secondary transfers its envelope, which is then re-emitted isotropically, can explain the observed age differences as well. One of these solutions explains the DB-nature of the oldest white dwarf in PG 1115+116 along the evolutionary scenario proposed by Maxted et al. (2002a), in which the helium core of the primary becomes exposed due to envelope ejection, evolves into a giant phase and loses its hydrogen-rich outer layers.