Astron. Astrophys. 323, 853-875 (1997)

1. Introduction

Massive X-ray binaries were among the very first X-ray sources detected and optically identified more than 20 years ago. These systems consist of a compact object, a magnetized neutron star (X-ray pulsar) or a black hole, in orbit around a massive OB star. One usually distinguishes three subtypes depending on whether the X-ray source accretes matter through Roche lobe overflow via an accretion disc (e.g. LMC X-4), from the high velocity wind of an early type star (e.g. Vela X-1) or from a low velocity extended envelope around a Be star (e.g. A0535+26). Because Bondi-Hoyle accretion (Bondi & Hoyle 1944) is extremely sensitive to the relative velocity of the orbiting compact object with respect to the circumstellar material, the low velocities and high densities usually observed in Be envelopes provide rather favourable conditions for accretion and not surprisingly, the majority of the known massive X-ray binaries are in fact Be/X-ray systems (see Van den Heuvel & Rappaport 1987 and Apparao 1994 for recent reviews).

Sporadic ejection of matter often observed in single Be stars combined with large variations of the accretion radius along the eccentric orbit explain the fact that most of these objects appear to be highly variable or even transient sources. The known massive X-ray binaries exhibit a large range of luminosities (10 erg s^-1 ; 1-10 keV). In spite of their concentration at low galactic latitude, their hard X-ray spectra resulting from the accretion onto a highly magnetized young neutron star or in rare cases onto a black hole allows to detect them very deeply into the galactic plane especially during the outburst states. However, large interstellar absorption often renders optical identification very difficult.

As individual objects, their scientific importance is high since, for instance, a large fraction of our knowledge on the masses of neutron stars comes from the Doppler analysis of these X-ray pulsars. By studying the variation of the X-ray luminosity and absorption by the intervening medium along the orbit, i.e. by using the neutron star as a probe of the stellar environment, several characteristics of the circumstellar envelopes (e.g. Waters et al. 1989; Motch et al. 1991a) or of the stellar wind (e.g. Haberl et al. 1989, Haberl & White 1990) can be constrained.

As a class of objects, their scientific importance is also large since they could account for part, if not all, of the hard diffuse emission observed in our Galaxy (e.g. the hard X-ray ridge emission; Warwick et al. 1985) and more generally in starburst galaxies (e.g. Griffiths & Padovani 1990) where they may turn on as bright X-ray sources already 9 10⁶ yr after the onset of star formation. Finally, these binaries may end their evolution by forming high mass binary pulsars (see e.g. Verbunt & van den Heuvel, 1995 for a review).

With a 1-10 keV sensitivity of the order of 2.5 10^-11 erg cm^-2 s^-1, X-ray surveys carried out before ROSAT had detection threshold luminosities of 3 10³³ (d/1 kpc)² erg s^-1. Therefore, many high mass systems may still remain hidden in the galactic plane, especially those powered by high velocity wind accretion and those belonging to the low end of the Be/X-ray luminosity function. Although the low energy range of the ROSAT PSPC may not be the best suited for searching such objects in regions of high interstellar absorption, the ROSAT all-sky survey has nevertheless the capability to detect these systems up to distances 10 times larger than previous experiments in directions relatively clear of interstellar absorption.

In this paper we present the result of a search for new OB/X-ray systems using the cross-correlation in position of all O and B stars listed in the SIMBAD database with the source list of the ROSAT galactic plane survey (RGPS; Motch et al. 1991b). By definition the RGPS is the part of the ROSAT all-sky survey (RASS; Voges 1992) restricted to regions of absolute galactic latitudes below 20 . The ROSAT satellite and instrumentations are described in Trümper (1983) and Pfeffermann et al. (1986).

This work extends the preliminary search carried out by Meurs et al. (1992) for a subsample of early type stars. By selecting stars which apparently exhibited a / ratio in excess of what is expected from normal stellar OB X-ray emission we could isolate a subset of candidate objects. Follow-up optical and pointed ROSAT observations allowed to assess the reality of the X-ray luminosity excess and resulted in the probable discovery of five new massive X-ray binaries with an additional two likely candidates.

In Sect. 2 we present the methods and results of the initial selection using SIMBAD entries and the ROSAT survey source list. In Sects. 3 to 5 we analyze our optical and X-ray survey and pointed ROSAT observations and draw conclusions on the reality of each new candidate massive X-ray binary. We then discuss the nature of the new systems discovered and their relation to the already known population of massive X-ray binaries. A preliminary report on this work was given in Motch et al. (1996c, d).

© European Southern Observatory (ESO) 1997

Online publication: May 26, 1998

helpdesk.link@springer.de