A spherical non-LTE line-blanketed stellar atmosphere model of the early B giant ε Canis Majoris

We use a spherical non-LTE fully line-blanketed model atmosphere to fit the full multiwavelength spectrum, including the extreme-ultraviolet (EUV) continuum observed by the Extreme Ultraviolet Explorer, of the B2 II star epsilon Canis Majoris (CMa). The available spectrophotometry of epsilon CMa from 350 Angstrom to 25 mu m is best fitted with model parameters T-eff = 21,750 K, log g = 3.5, and an angular diameter of 0.77 mas. Our best-fit model predicts a hydrogen ionizing flux, q(0), of 1.59 x 10(21) photons cm(-2) s(-1) at the star's surface and 2290 photons cm(-2) s(-1) at the surface of the Local Cloud. The close agreement between the model and the measured EUV flux from epsilon CMa is a result of the higher temperatures at the formation depths of the H I and He I Lyman continua compared with other models. The realistic model treatment of early B giants with spherical geometry and non-LTE metal line-blanketing results in the prediction of significantly larger EUV fluxes compared with plane-parallel models. We find that our metal line-blanketed spherical models show significantly warmer temperature structures, 1-3 kK at the formation depth of the Lyman continua, and predict stronger EUV fluxes, up to a factor of 5 in the H I Lyman continuum, compared with plane-parallel atmospheres that have identical model parameters. In contrast, we find that spherical and plane-parallel models that do not include metal line blanketing are nearly identical. Our T-eff = 21,000 K, log g = 3.2, spherical non-LTE model predicts more than twice as many hydrogen ionizing photons and over 200 times more neutral helium ionizing photons than a standard hydrostatic plane-parallel LTE model with the same stellar parameters. Our synthetic spectra are in reasonably good agreement with observed continuum and line fluxes from echelle spectra obtained with the Goddard High Resolution Spectrograph. While we find agreement between the absolute UV flux of epsilon CMa as measured by GHRS and our model atmosphere, these fluxes are similar to 30% higher in the UV than those measured by IUE, OAO 2, and TD-1, in excess of the published errors in the absolute calibration of these data.