Context. Stellar mergers are expected to take place in numerous circumstences in the evolution of stellar systems. In particular, they are considered as a plausible origin of stellar eruptions of the V838 Mon type. V1309 Sco is the most recent eruption of this type in our Galaxy. The object was discovered in September 2008. Aims. Our aim is to investigate the nature of V1309 Sco. Methods. V1309 Sco has been photometrically observed in course of the OGLE project since August 2001. We analyse these observations in different ways. In particular, periodogram analyses were done to investigate the nature of the observed short-term variability of the progenitor. Results. We find that the progenitor of V1309 Sco was a contact binary with an orbital period of ∼1.4 day. This period was decreasing with time. The light curve of the binary was also evolving, indicating that the system evolved towards its merger. The violent phase of the merger, marked by the systematic brightenning of the object, began in March 2008, i.e. half a year before the outburst discovery. We also investigate the observations of V1309 Sco during the outburst and the decline and show that they can be fully accounted for within the merger hypothesis. Conclusions. For the first time in the literature we show from direct observations that contact binaries indeed end up by merging into a single object, as was suggested in numerous theoretical studies of these systems. Our study also shows that stellar mergers indeed result in eruptions of the V838 Mon type.

We propose that the energy source of the outburst of V838 Mon and similar objects is an accretion event, i.e., gravitational energy rather than thermonuclear runaway. We show that the merger of two main-sequence stars, of masses and , can account for the luminosity, large radius, and low effectivetemperature of V838 Mon and similar objects. By varying the masses and types of the merging stars, and by considering slowly expanding, rather than hydrostatic, envelopes, this model can account for a large range in luminosities and radii of such outburst events.

Abstract. By fitting the available photometric data on V838 Mon with standard supergiant spectra we have derived principal stellar parameters, i.e. effective temperature, radius and luminosity, and followed the evolution of the object since its discovery in early January 2002. Our analysis shows that the 2002 outburst of V838 Mon consisted of two major phases: pre-eruption, which was observed in January 2002 and a major outburst, the eruption, which started in the beginning of February 2002. During pre-eruption the object seemed to be relaxing after an initial event which had presumably taken place in last days of December 2001. The eruption phase, which lasted until mid-April 2002, resulted from a very strong energy burst, which presumably took place in last days of January at the base of the stellar envelope inflated in pre-eruption. The burst produced an energy wave, which was observed as a strong luminosity flash in the beginning of February, followed by a strong mass outflow in the form of two shells, which was observed as an expanding photosphere in later epochs. In mid-April, when the outflow became optically transparent and most of its energy radiated away, the object entered the decline phase during which V838 Mon was evolving along the Hayashi track. This we interpret as evidence that the main energy source during decline was gravitational contraction of the object envelope inflated in eruption. Late in 2002, dust formation started in the expanding shells which gave rise to a strong infrared excess observed in 2003.

We discuss various models and scenarios proposed to explain the nature of the V838 Mon type eruptions. In this class of eruptive objects we include: M 31 RV (erupted in 1988), V4332 Sgr (erupted in 1994) and V838 Mon (erupted in 2002). We concentrate on three models: (i) thermonuclear runaway on an accreting white dwarf (nova-like event); (ii) He-shell flash in a post asymptotic giant branch star (born-again AGB); and (iii) merger of stars. We show that models (i) and (ii) cannot account for the majority of the observed properties of the objects. Most significantly, in both nuclear burning type models the object is expected to heat up before declining and fade as a very hot compact star. In the observed eruptions the objects declined as very cool giants or supergiants. We show that the stellar merger model can account for all the observed properties and conclude that presently this is the most promising model to explain the eruptions of the V838 Mon type.

No abstract