STIS Spectral Imagery of the OB Stars in NGC 604. II. The Most Luminous Stars

We present results using two-dimensional spectral imagery and photometry obtained with the Hubble Space Telescope (HST) for the starburst H II region NGC 604, in the nearby galaxy M33. The spectral imagery was acquired with the Space Telescope Imaging Spectrograph (STIS) using the MAMA-G140L configuration, which provided wavelength coverage spanning 1170–1730 A. From a single 1720 s STIS exposure, we have extracted spectra for 49 stars and derived individual UV spectral types for 40 stars in the crowded 25'' × 2'' stellar field sampled by the STIS aperture. These stars represent a significant fraction of the young, luminous O and B stars in NGC 604. Three objects have pronounced He II λ1640 emission, the signature of Wolf-Rayet (W-R) or luminous Of stars. By combining UV fluxes with HST WF/PC-1 and WFPC2 photometry at visible wavelengths, we derive the extinction curve for NGC 604. We use this extinction curve, together with the available accurate distance for M33, derived UV spectral types, and HST photometry, to determine positions of the luminous stars in the upper Hertzsprung-Russell diagram for NGC 604. The revision to the O star effective temperature scale by Martins et al., based on non-LTE, line-blanketed model atmospheres, is essential in obtaining reliable positions in the log L*–log Teff plane. These stars are quite young, with a characteristic age of ~3 Myr. The spectra and photometry indicate that three objects are exceedingly luminous. Their inferred locations in the H-R diagram relative to theoretical evolutionary tracks indicate stellar masses ≥120 M⊙. High spatial resolution HST imagery provides no evidence of multiple stars composing these objects. Still, we cannot eliminate the possibility that these objects are not unresolved multiple stars of lower mass, possibly W-R stars. Simple tests demonstrate that the 10 most luminous stars predominantly determine the UV spectral features seen in the total light of NGC 604. We conclude that the interpretation of spectral fitting of more distant starburst galaxies, where individual stars are not resolved, must be done with extreme care.