Most stars in the Galaxy are supposed to form in OB associations (Miller & Scalo 1978). The knowledge of the stellar content of OB associations is a key towards an understanding of the star formation process and the origin of the initial mass function. One important point is the comparison of the low-mass populations in OB associations and T associations. This can yield information on whether star formation is bimodal (cf. Walter & Boyd 1991) and to what extent the environment influences the star formation process.
While the low-mass stellar population in T associations (like in Taurus) is rather well-known, not much is known about the low-mass stellar contents of OB associations. This is due to our extremely poor knowledge of membership for all but the brightest O and early B stars. The low-mass population remains largely unexplored, because most of the low-mass pre-main sequence (PMS) stars cannot be easily distinguished from normal field stars. Only the classical T Tauri stars can be rather easily found by their strong H emission, e.g. by objective prism surveys. However, the PMS population is dominated by the weak-line T Tauri stars (cf. Krautter 1996), which lack such an easily detectable signature and thus are very hard to find among the many thousands of foreground and background field stars in the huge area on the sky (several hundred square-degrees) covered by nearby OB associations.
Interestingly, most PMS stars are strong X-ray emitters, probably due to their rapid rotation (cf. Montmerle 1996). Since their X-ray luminosities are about 2 - 3 orders of magnitude above those of main sequence stars of similar spectral type, X-ray observations have proven to be extremely efficient in discovering the PMS stars among the older field stars (e.g. Walter et al. 1988; Neuhäuser 1997; Wichmann et al. 1997).
The Scorpius Centaurus association is the OB association nearest to the Sun. It contains several hundred B stars which cluster in the three subgroups Upper Scorpius, Upper Centaurus Lupus, and Lower Centaurus Crux (cf. Blaauw 1964). With an age of about 5 Myrs, Upper Sco is the youngest subgroup. De Geus et al. (1989) list 98 B and early A stars as probable association members and claim completeness to spectral type B9.
The Scorpius Centaurus association was very well investigated by the astrometry satellite Hipparcos. Astrometric data from Hipparcos are now available for 1215 stars in Upper Sco. De Bruijne et al. (1997) present first results of kinematic membership determinations making full use of the proper motions and parallaxes measured by Hipparcos. For of the probable members from the list of de Geus et al. (1989) membership is confirmed by the Hipparcos data. Furthermore, de Bruijne et al. (1997) could identify 115 new members, and their analysis of the trigonometric parallax distribution gives a mean distance of pc for Upper Sco.
Immediately to the east of Upper Sco lie the Ophiuchus dark clouds. The center of this cloud complex, the Oph cloud core, contains many PMS stars, X-ray sources, embedded infrared sources, and displays several signposts of very recent star forming activity (cf. Wilking 1991, Casanova 1995). In contrast to its spatial proximity to the Ophiuchus dark clouds, Upper Sco is essentially free of dense interstellar matter and there is no evidence for ongoing star formation in Upper Sco. This is probably due to the strong stellar winds of the numerous B stars, which have dispersed the original molecular cloud. Furthermore, a supernova explosion presumably has swept out the molecular gas a few million years ago. This is the reason for the generally quite low extinction of the young stars in Upper Sco ( mag; cf. de Geus et al. 1989, Walter et al. 1994).
In contrast to the well studied population of high mass stars in Upper Sco, virtually nothing was known about the population of the low-mass PMS stars, until recently. Only 6 PMS stars in Upper Sco are listed in the catalog of Herbig & Bell (1988). The first systematic search for PMS stars in Upper Sco was performed by Walter et al. (1994, W94 hereafter), who observed the optical counterparts of X-ray sources detected in 7 individual EINSTEIN fields and could detect 28 PMS stars. A quite surprising result of this study was that the PMS stars seem to have isochronal ages of 1 - 2 Myrs with a very small dispersion. This is significantly younger than the well established age of the B stars ( Myrs) and was interpreted as an indication that the formation of these PMS stars was triggered.
Since this result was based on the assumption of a common distance of 160 pc for all PMS stars, it should be treated with some caution, because the new Hipparcos data imply a slightly smaller distance of 145 pc and probably there is a spread in the individual distances of the PMS stars. We can gain some information about the spread in distances for the early type members from the histogram of parallaxes presented in Fig. 2 of de Bruijne et al. (1997). The central 90% range of the distribution of parallaxes extends from 4 mas to 9 mas. We must take into account that this distribution of parallaxes is broadened by measurement errors. The typical measurement errors are for parallaxes near 9 mas and for parallaxes near 4 mas. If we correct the range of parallaxes by this amount, we find that the data are consistent with the assumption that the distances of the individual stars vary between pc and pc.
Recently, Martin (1998) disputed the results of W94 and argued that there is a large age spread among the PMS stars. Using the 6708 Å lithium line data from W94, he classified 8 of the 28 PMS stars, i.e. a fraction of about 30%, as "post T Tauri stars" with ages of Myrs, what would imply that these stars are at distances of only pc. However, we note that these stars are radial velocity members of Upper Sco (W94), and, as we will show below, the population of late type PMS stars seems to be spatially coincident with the early type members. Thus there is no expectation that the late type members should have a much broader distribution of distances than the pc range we have estimated for the early type members above. So the scenario of Martin (1998) does not fully convince us.
In any way, it must be kept in mind that the EINSTEIN observations covered only about 7 square-degrees, a tiny fraction of the whole association (cf. Fig. 1), and the sample of 28 PMS stars is clearly far from being complete. Further conclusions about the star formation history of the association require a much larger sample of PMS stars and thus the observation of many more stars in Upper Sco. A first step in this direction was a study by Kunkel (1998, K98 hereafter), who investigated more than 200 ROSAT All Sky Survey X-ray sources in a square-degree area in Upper Sco. He claimed the detection of 93 new PMS stars. 35 of these stars are located in the area investigated in our study and satisfy our classification criteria for PMS stars (see Sect. 4.1).
The main intention for the study presented in this paper was to provide a more complete and homogeneously selected sample of PMS stars in a large area around the previously known PMS stars. This requires optical spectroscopy of several hundred targets in an area of several hundred square-degrees. Therefore we have used the wide-field multiobject spectrograph FLAIR at the UK Schmidt telescope, which is ideally suited for this project.
The very efficient observing mode possible with FLAIR even allowed us to address another important question: How complete is the X-ray selected sample of PMS stars? Since nearly all known PMS stars in Upper Sco have been found by follow-up observations of X-ray sources, this question is of fundamental importance for any conclusions. The significance of this question has recently been demonstrated in a study of PMS stars in the Orion nebula region. Wolk (1996) found PMS stars not only among X-ray selected stars but also among non -X-ray selected stars with similar positions in the color-magnitude-diagram. This means that a rather large fraction of the PMS stars in this region have much lower X-ray luminosities than typical for the X-ray detected PMS stars of similar bolometric luminosity. This surprising result implies the existence of a population of "X-ray quiet" PMS stars and shows that the X-ray selected sample of PMS stars in Orion is probably considerably incomplete.
We have investigated this question for Upper Sco with a slightly different approach. In addition to the X-ray selected candidates, we also studied a large sample of stars which were not detected as X-ray sources, but which we regarded as possible members of the Upper Scorpius association because of their proper motions being similar to the ones of the known early-type members.
The region we have studied has a total area of about 160 square-degrees and is defined by a mosaic of 6 individual UK Schmidt Telescope survey fields (see Fig. 1). Our area contains all the EINSTEIN fields analyzed in W94 and partially overlaps with the region studied by K98. Before our study, 69 PMS stars were known in our area: Six from Herbig & Bell (1988), 28 from W94, and 35 from K98. Since all the PMS candidates we have observed are listed in the Guide Star Catalog (GSC; Lasker et al. 1990), we use GSC numbers to identify the stars. Throughout this paper we use the B magnitudes as given in the GSC.
© European Southern Observatory (ESO) 1998
Online publication: April 20, 1998