2011A&A...532A.138T

V838 Mon erupted at the beginning of 2002. In the course of the outburst the object evolved to low effective temperatures and declined as a very late M-type supergiant. Among various scenarios proposed to explain the nature of the outburst, the most promising is a stellar merger event. We aim at studying the structure and evolution of the object in the decline from the 2002 eruption. We obtained spectroscopic observations of V838 Mon in January-March 2009 with UVES/VLT. The results are analysed and compared with similar observations obtained in October 2005 with HIRES/Keck. The most striking difference between 2009 and 2005 is a complete absence of the B3 V component and of the [FeII] emission lines in 2009. The present spectrum displays only the spectrum of the 2002 eruption remnant. It resembles that of an ∼M6 giant, although the molecular bands in V838 Mon are deeper than those in standard stellar spectra of a similar spectral class. Several atomic lines, which displayed P-Cyg profiles in 2005, are now dominated by pure absorptions. Some of these lines, however, show a narrow emission component, which, as we argue, measures the radial velocity of V838 Mon. The resulting heliocentric velocity is 71km/s, which very well agrees with the velocity obtained from SiO maser observations. The atomic lines and the molecular bands show very complex kinematics. In some atomic lines and high-excitation molecular bands we observe matter infalling in the object atmosphere. The infall components were already observed in 2005, but were less pronounced and present in fewer lines than in 2009. We argue that the most negative radial velocity components seen in the resonance atomic lines and in the low-excitation molecular bands were formed in the ejecta of the 2002 eruption. The B3 V companion most probably became engulfed in an opaque dusty matter of the 2002 V838 Mon ejecta.