4. Criteria for the identification of OB/X-ray candidates
The goal of the present work is merely to extract from the ROSAT Galactic Plane Survey the most obvious OB/X-ray candidates. As stated in Sect. 2.2, we decided to ignore for the moment all associations of a survey source with groups of OB stars and all cases where there was a possible alternative candidate to the OB identification. We applied the same strategy during the follow-up observations, dropping the candidate as soon as another possible optical counterpart was found.
Active coronae constitute up to 85% of the ROSAT survey sources at b = 0 (Motch et al. 1996b) and over 50% in the overall RGPS region (Motch et al. 1991b). Although no direct conspicuous signature of coronal activity exists in the optical spectra of late type stars, it is known that chromospheric and coronal activities tightly correlate in the Sun and F-M stars in general. In particular, Einstein observations have shown that the luminosity of the chromospheric Balmer and Ca II H&K emission lines is well correlated with the soft X-ray luminosity in Me stars (e.g. Fleming et al. 1988) and in F-G stars (e.g. Maggio et al. 1987). Using medium resolution blue optical spectra of 100 late type stars identified with RGPS sources with measured Ca II re-emission, Guillout (1996) find a very good correlation between Ca II H&K and soft X-ray emission with . Because the slope in luminosity is close to 1, there also exists a useful relation between chromospheric fluxes and PSPC count rates r ; with a 0.1 PSPC cts s-1 corresponding to 2 10-13 erg cm-2 s-1. These relations hold over 2 decades in flux and over 3 decades in luminosity. A rms scatter of about a factor 2 is present in flux and luminosity around the mean relation with a maximum deviation of a factor 10 above or below. The intrinsic long (solar cycles) and short (flares) time scales of variability and the non-simultaneity of the optical follow-up / X-ray survey observations probably account for the rather large scatter. We checked that the measured chromospheric fluxes and PSPC count rates of all late type stars proposed as alternative identification to the early type star were compatible with the relation of Guillout (1996). Because of the sometimes inaccurate spectrophotometry we re-calibrated our blue medium resolution spectra using either magnitudes extracted from the literature or derived from our CCD images.
In the absence of alternative candidates, we used the X-ray luminosity as additional criterion. Although active coronae with X-ray luminosities above 1031 erg s-1 are usually not observed in late type stars (see e.g. Rosner et al. 1985) some RS CVns (Ottmann & Schmitt 1992) and extreme T Tauri stars (e.g. Montmerle et al. 1983) do exhibit on occasions soft X-ray luminosities above 1031 erg s-1. Accordingly, candidate stars with excess in the range of 1031-32 erg s-1 were indicated as deserving further attention but are not considered as very good candidates.
The last criterion considered is the hardness of the source. Normal OB stars usually exhibit HR2s 0.3 whereas most HMXBs known prior to the launch of ROSAT have hard HR2s 0.4. However, the separation power of this X-ray spectral criterion is decreased by the presence of a soft excess in a small fraction of these OB/X-ray systems (e.g. LMC X-4). Furthermore, some very active coronae may also display hard spectra (e.g. Schmitt et al. 1990) and a fraction of T Tauri stars exhibit hardness ratios HR2 larger than 0.4 (e.g. Neuhäuser et al. 1995). We did not consider the presence or absence of Balmer emission lines as a criterion since some early type companions of known massive sources lack these spectral features.
Finally, in spite of the large galactic absorption expected in the directions occupied by OB stars, a few bright extragalactic emitters may still shine in X-ray through the galactic fog. These sources will also display a hard HR2 as a result of interstellar absorption and will be optically faint (V 18). The detection of characteristic features in their optical spectra may be beyond the capability of the instrumentation used for this project. Motch et al. (1996b) estimate that the spatial density of extragalactic emitters in the RGPS Cygnus test region is 2.7 10-2 extragalactic sources per square degree brighter than 0.02 cts s-1 for b in the range of -5 to +5 . If these densities are typical of the average OB star directions, we expect 0.1 spurious coincidence within a typical 95% confidence radius between an extragalactic emitter and an OB star. Therefore, we cannot completely rule out the possibility that at least one of the investigated candidate sources is actually extragalactic.
© European Southern Observatory (ESO) 1997
Online publication: May 26, 1998