The structure of the homunculus. III. Forming a disk and bipolar lobes in a rotating surface explosion

We present a semianalytic model for the shaping of the Homunculus Nebula around η Carinae that accounts for the simultaneous production of bipolar lobes and an equatorial disk through a rotating surface explosion. Material is launched normal to the surface of an oblate rotating star with an initial kick velocity that scales approximately with the local escape speed. Thereafter, ejecta follow ballistic orbital trajectories, feeling only a central force corresponding to a radiatively reduced gravity. Our model is conceptually similar to the wind-compressed disk model of Bjorkman & Cassinelli, but we modify it to an explosion instead of a steady line-driven wind, we include a rotationally distorted star, and we treat the dynamics somewhat differently. A continuum-driven explosion, where the radiation force is independent of velocity, avoids the disk inhibition mechanisms that normally operate in line-driven winds. This allows midlatitude material with appropriate initial specific energy to migrate toward the equator, where it collides with material from the opposite hemisphere to form a disk. Thus, our model provides a simple method by which rotating hot stars can simultaneously produce intrinsically bipolar and equatorial mass ejections, without relying on an aspherical environment or magnetic fields. Although motivated by η Carinae, the model may have generic application to episodic mass ejection where rotation is important, including other luminous blue variables, B[e] stars, the nebula around SN 1987A, or possibly even bipolar supernova explosions themselves. In cases where near-Eddington radiative driving is less influential, our model generalizes to produce bipolar pinched-waist morphologies without disks, as seen in many planetary nebulae. If rotating single stars can produce strongly axisymmetric ejecta by this mechanism, then the presence of aspherical ejecta by itself is insufficient justification to invoke close binary evolution.