On Two Theories of the Cyclical Outbursts of η Carinae

The star η Carinae is an exceptionally luminous, blue variable that is at least partially evolved and undergoes major outbursts once every 3-6 yr on the average. On the basis of a new grid of stellar evolutionary tracks, we estimate that η Car possessed an initial mass of 150-300 M ☉ and is still burning hydrogen in its core. We test a hypothesis that η Car is repeatedly encountering ionization-induced dynamical instability within its outer envelope. In that case, the star has probably already lost a third of its initial mass, while its surface hydrogen abundance has fallen to about half its original value. Its effective temperature, defined from quasi-static models in the conventional way, is predicted to be 8000 ± 2000 K, but theoretically it might be as high as 200,000 K and so is not accurately known from the models. Although our assumption of hydrostatic equilibrium in the envelope of such an extremely luminous star is crude, the predicted cycles of mass loss agree well with these actually observed. Nevertheless, the models impose requirements before instability first breaks out that seem to be at variance with observations of extremely massive stars: stellar wind mass-loss rates falling too far on the low side and a significant amount of time spent as a B-type supergiant. Our models for η Car itself may be too cool. An alternative destabilizing mechanism, based on an assumed supercritical radiative acceleration of material outward from the photosphere, applies potentially to very luminous, hot stars like η Car, but produces no sustained cycles in our models. However, our models are only quasi-static, not hydrodynamic. For luminous blue variables generally, we confirm that the mean cycle time is, to a rough approximation, inversely proportional to luminosity. Our results, therefore, justify the use of luminous blue variables as extragalactic distance indicators. Analysis of a simple one-zone model supports our adopted criterion for dynamical instability and hence at least our theoretical results for ordinary luminous blue variables with low and moderate luminosities.