Line‐blanketed Spherically Extended Model Atmospheres of Hot Luminous Stars with and without Winds

Abstract

We have found that our spherical, hydrostatic, nonlocal thermodynamic equilibrium (non-LTE) metal line-blanketed model atmospheres are able to reproduce the spectral energy distributions of the bright B giants b and e Canis Majoris, including the extreme ultraviolet, where previous models have failed (J. P. Aufdenberg, P. H. Hauschildt, & E. Baron 1999, MNRAS, 302, 599; J. P. Aufdenberg, P. H. Hauschildt, S. N. Shore, & E. Baron 1998, ApJ, 498, 837). The brightest stellar source of EUV radiation longward of 400 A, e CMa is an extremely important star because it produces more hydrogen-ionizing flux than all nearby stars combined. Because of their proximity to the Sun, these two stars provide the rare opportunity to directly test predictions of theoretical stellar continua below the Lyman limit. These stars have precisely measured angular diameters, trigonometric parallaxes, and bolometric fluxes, and we have shown that the best-fitting models fall on the Hertzsprung-Russell diagram within the small error boxes provided by these observations. Within these constraints, we were able to compare our synthetic spectra to the observed spectral energy distributions of these B stars in absolute units, a rarity in stellar atmosphere analyses. Observations show a large EUV flux excess compared with plane-parallel model atmospheres (J. P. Cassinelli et al. 1995, ApJ, 438, 932; J. P. Cassinelli et al. 1996, ApJ, 460, 949). This led to the suggestion that various scenarios such as X-ray heating and the presence of a stellar wind were responsible for the mismatch between theory and observation. We discovered that a combination of metal line blanketing and spherical geometry present in our model atmospheres produced significantly more EUV flux than (otherwise similar) plane-parallel atmospheres without the need for these ad hoc mechanisms. These findings are in agreement with studies of B star H ii regions which also show evidence for excess EUV flux from B stars relative to plane-parallel model predictions (R. J. Reynolds 1985, AJ, 90, 92; R. J. Reynolds & S. L. Tufte 1997, ApJ, 439, L17). Our models for b and e CMa represent the only spherical, non-LTE line-blanketed models of B stars in the literature. Surprisingly, we have found that the combination of spherical geometry, line blanketing, and surface gravities as high as will produce significantly different model temlog g p 3.5 perature structures and synthetic EUV spectra relative to otherwise similar plane-parallel models over a range in effective temperature of 18,000–28,000 K. It is the combination of the spherical geometry and the line blanketing, not the non-LTE aspects, that is the most important factor creating this effect. We believe an important factor which shifts the surface cooling/ backwarming to greater depths in the spherical models is the reduced optical depth seen by photons in a spherical geometry relative to that in a semi-infinite plane-parallel geometry, for a given physical depth. As a result, in the spherical models, backwarming begins at a greater depth than in the plane-parallel models. These deeper layers are intrinsically warmer, so that the temperature of the EUV continuum-forming layers are at a higher temperature than the EUV continuum-forming layers in the plane-parallel models. We encourage other researchers to reproduce our findings with an independent spherical lineblanketed model atmosphere code. Our models for hot, luminous stars and their winds are thus far the only ones that include metal line blanketing and combine the inner hydrostatic layers with the outer dynamic layers of the atmosphere in a single structure. This is a significant step, particularly regarding the inclusion of full non-LTE metal line blanketing in both the computation of the model atmosphere plus wind and the synthetic spectrum. Model approaches to date have been limited by the use of a spherical gray temperature structure (not fully converged), the lack of full line blanketing, and the Sobolev approximation, which cannot treat overlapping lines or multiple scattering. In our models, the temperature structure is computed from the condition of energy conservation in the comoving frame, full line blanketing is included, and the Sobolev approximation is not used. Moving beyond these models, we have begun to develop a theoretical framework that will self-consistently obtain the velocity structure and the mass-loss rate from radiation-driven wind theory given the effective temperature, radius, and gravity of the star. We have developed a module for the PHOENIX