Which massive stars are gamma-ray burst progenitors?

The collapsar model for gamma-ray bursts requires three essential ingredients: a massive core, removal of the hydrogen envelope, and enough angular momentum in the core. We study current massive star evolution models of solar metallicity to determine which massive star physics is capable of producing these ingredients. In particular, we investigate the role of hydrodynamic and magnetic internal angular momentum transport and binary mass and angular momentum transfer. We follow the evolution of rotating single stars and of binary systems that include rotational processes for both stars. Neglecting magnetic fields, we show that the cores of massive single stars can maintain a high specific angular momentum ( j similar to 10(17) cm(2) s(-1)) when evolved with the assumption that mean molecular weight gradient suppresses rotational mixing processes. In binary systems that undergo mass transfer during core hydrogen burning the mass receiving star accretes large amounts of high angular momentum material, leading to a spin-up of the core. We find, however, that this merely compensates for the tidal angular momentum loss due to spin-orbit coupling, which leads to synchronous rotation before the mass transfer event. Therefore the resulting cores do not rotate faster than in single stars. We show that some accreting stars become Wolf-Rayet stars at core helium exhaustion and form CO-cores that are massive enough to form a black hole. We also present models that include magnetic fields generated by differential rotation and we consider the internal angular momentum transport by magnetic torques. Though magnetic single star models are known to develop rather slowly rotating cores with specific angular momenta at the end of the evolution close to those in observed young pulsars ( j similar to 10(14) cm(2) s(-1)), we investigate the capability of magnetic torques to efficiently pump angular momentum into the cores of accreting stars. Despite our finding that this mechanism works, the magnetic coupling of core and envelope after the accreting star ends core hydrogen burning leads to slower rotation ( j similar to 10(15-16) cm(2) s(-1)) than in the non-magnetic case. We conclude that our binary models without magnetic fields can reproduce stellar cores with a high enough specific angular momentum ( j >= 3 x 10(16) cm(2) s(-1)) to produce a collapsar and a GRB. If magnetic torques are included, however, GRBs at near solar metallicity need to be produced in rather exotic binary channels, or current dynamo model overestimates the magnetic torques. But then the problem is that significant angular momentum loss from the iron core either during core collapse or from the proto-neutron star would be required.