Abstract

Context. Many massive stars have nearby companions. These hamper a characterization of massive stars through spectroscopy.Aims. We continue to obtain spatially resolved spectroscopy of close massive visual binaries to derive their spectral types.Methods. We used the lucky spectroscopy technique to obtain a large number of short long-slit spectroscopic exposures of 19 close visual binaries under good seeing conditions. We selected those with the best characteristics, extracted the spectra using multiple-profile fitting, and combined the results to derive spatially separated spectra. The results were analyzed in combination with data from lucky imaging, regular intermediate-resolution single-order spectroscopy, and echelle high-resolution spectroscopy.Results. The new application of lucky spectroscopy has allowed us (among other results) to [a] spatially disentangle two O stars (FN CMa B and 6 Cas B) with brighter BA supergiant companions for the first time; [b] determine that two B stars (alpha Sco B and HD 164 492 B) with close and more massive companions are fast rotators (in the second case, solving a case of mistaken identity); [c] extend the technique to cases with extreme magnitude differences (the previous two cases plus CS Cam A,B), shorter separations (HD 193 443 A,B), and fainter primary magnitudes down to B=11 (HD 219 460 A,B); [d] spatially disentangle the spectra of stars with companions as diverse as an A supergiant (6 Cas A), a Wolf-Rayet star (HD 219 460 B = WR 157), and an M supergiant (alpha Sco A); [e] discover the unexpected identity of some targets such as two previously unknown bright O stars (HD 51 756 B and BD +60 544) and a new member of the rare OC category (HD 8768 A); and [f] identify and classify (in some cases for the first time) which of the components of four visual binaries (sigma Ori, HD 219 460, HD 194 649, and HD 191 201) is a double-lined spectroscopic binary. For another seven systems (FN CMa, sigma Sco, HD 51 756, HD 218 195, HD 17 520, HD 24 431, and HD 164 492), we detect signs of spectroscopic binarity using high-spectral-resolution spectroscopy. We also determine the limits of the technique.