next previous
Up: Search for young stars cloud


   
8 Completeness of our survey

In Fig. 10, we compare the complete log N-log S curve with those for stellar, extra-galactic, and unidentified sources. At a PSPC count rate threshold of 0.1 cts  ${\rm s}^{-1}$, $90\%$ of the RASS sources have optical counterparts, most of them being stars. At lower count rates the fraction of unidentified sources increases from $\sim 30\%$ at 0.07 cts  ${\rm s}^{-1}$ to more than $50\%$ at 0.02 cts  ${\rm s}^{-1}$, a level at which all curves flatten because of the incompletness of the RASS itself. Above a PSPC count rate of 0.03 cts s-1 we have optically identified 75 among 136 RASS X-ray sources (i.e. $55\%$), most of them being stars. The small number of extra-galactic object identified at this threshold (only two) is simply due to our follow-up observation strategy. According to the results of Guillout (1996) and Zickgraf et al. (1997), we conclude that the extra-galactic population in the R CrA region is likely to account for about $25\%$ of the unidentified X-ray sources detected above 0.03 cts s-1, the rest (i.e. $20\%$) probably being optically faint active K- and M-type stars.


  \begin{figure}
\par\includegraphics[width=9cm,clip]{ds1875_f10.ps}
\end{figure} Figure 10: The log N-log S curves in the CrA region. Number of X-ray sources per deg2 detected by ROSAT as a function of PSPC count rate for all RASS sources listed in Table 2 (solid bold line), for the optically identified stellar (star symbols) and extra-galactic (diamonds) populations as well as for the unidentified sources ($\times $)

We now focus on the identified stellar population and define two sub-regions within our field, namely the on-cloud region (from $\alpha $ = 18 h 56 m to 19 h 24 m and from $\delta = -38^{\circ}$ to $-36^{\circ}$, i.e. 14 deg2) and the off-cloud region (complementary to the on-cloud region, i.e. 112 deg2).


  \begin{figure}
\par\includegraphics[width=9cm,clip]{ds1875_f11.ps}
\end{figure} Figure 11: The stellar log N-log S curves in the CrA region. Number of stellar X-ray sources per deg2 optically identified on-cloud (star symbol) and off cloud (plus symbol) regions as well as predictions from stellar X-ray population model. Bold curve is for all model age bins, dash-dotted line for stars younger than 150 Myrs, dashed line for stars from 150 to 1000 Myrs old and dotted line for stars older than 1000 Myrs

We have plotted in Fig. 11 the observed on-cloud and off-cloud stellar log N-log S curves as well as the predictions of the stellar X-ray population model from Guillout et al. (1996). Computations were run for |b| = 15$^\circ$ and l = 180$^\circ$although the galactic longitude is irrelevant at the RASS sensitivity. Results are summarized in Table 7.


 
Table 7: X-ray source density. Surface density of stellar X-ray sources observed on ( $\rho _{\rm On}$) and off ( $\rho _{\rm Off}$) R CrA cloud, model predictions ( $\rho _{\rm model}$) as well as observed-prediction ratio as a function of PSPC count rate threshold S in cts/s
S $\rho _{\rm On}$ $\rho_{\rm off}$ $\rho _{\rm model}$ $\rho _{\rm On}$/ $\rho _{\rm model}$ $\rho _{\rm Off}$/ $\rho _{\rm model}$
0.10 0.179 0.167 0.113 1.58 1.47
0.05 0.626 0.414 0.273 2.29 1.51
0.03 1.073 0.682 0.482 2.22 1.41

First we note that at any PSPC count rate the on-cloud stellar density is significantly higher (by a factor 2) compared to the model predictions, as expected for a region with ongoing star formation. On the other hand, we expect that the off-cloud log N-log S curve lies within $15\%$ of the model prediction, which is clearly not the case.

In order to check the relevance of the model prediction, we compare with the so-called RasTyc sample (Guillout et al. 1999), a sample of all objects included in both RASS and Tycho, i.e. the largest sample of stellar X-ray sources with homogeneous and accurate data constructed so far. In order to account for the magnitude and X-ray flux limited biases of of RasTyc, we ran a specific model optically limited at 10.5 mag (plus $20\%$ of stars down to 11.5 mag). We then compared the expected number of stars per deg2 with the observed one computed in two regions extending $10^{\circ}$ wide all around the sky and centered at |b| = 15$^\circ$. At a PSPC count rate threshold of 0.03 cts s-1, there are 1819 RasTyc stars detected within these two regions amounting to 6946 deg2, i.e. 0.26 stars per deg2. At this level, the model predicts a stellar surface density of 0.23 stars per deg2, in very good agreement with the observations. We are thus confident that the theoretical log N-log S curves plotted in Fig. 11 give a good estimation of the ambient galactic plane stellar population at the R CrA cloud galactic latitude.

We then conclude that in the region surrounding the CrA molecular cloud our observations reveal $\sim 40\%$ excess of stellar X-ray sources with respect of a "pure'' galactic plane population (see Table 7). According to the expected contribution of extra-galactic sources to the unidentified population, $40\%$ is a lower limit on the excess. Such excesses were also detected around other star forming regions (see Neuhäuser 1997 and references therein). However, contrary to some other star forming regions like Lup-Sco-Cen (Guillout et al. 1998a,b), the Gould Belt can hardly be an explanation because of the position of the CrA molecular cloud projected well below the Gould Belt plane.

Also around the Chamaeleon clouds (Alcála et al. 1995) and south of the Taurus clouds (Neuhäuser et al. 1997), many new pre-MS stars were found, although there is no Gould Belt in that directions. As far as the Chamaeleon off-cloud TTS are concerned, Mizuno et al. (1998) found new, previously unknown, small cloud-lets near one third of the off-cloud TTS, which may be the birth places of those seemingly off-cloud TTS. If one can explain off-cloud TTS around the CrA and Cha clouds by cloud-lets rather than by the Gould Belt, at least some of the Lup-Sco-Cen, and Orion off-cloud TTS may also have originated in such small cloud-lets, as originally proposed for the Chamaeleon off-cloud TTS by Feigelson (1996).

The question now is whether we found all young, i.e. coronally active stars (inside and) around the CrA dark cloud. This can be investigated by optical follow-up observations of additional unidentified X-ray sources found in deep ROSAT PSPC and HRI pointed observations (Walter et al., in preparation).

Whether all young stars were found among all RASS sources can be investigated in the following way: Sterzik et al. (1995) have shown that it is possible to pre-select TTS candidates from the RASS using four criteria, namely the two hardness ratios, the X-ray count rate, and the optical magnitude of the nearest (if any) counterpart (within, say, $40^{\prime \prime}$). Then, TTS candidates are those RASS sources which resemble best previously known RASS-detected bona-fide TTS according to the same properties. The parameter which describes how well a particular RASS source resembles the typical TTS properties is called discrimination probability P, described in detail in Sterzik et al. (1995).


  \begin{figure}
\par\includegraphics[angle=90,width=8cm,clip]{ds1875_f12.ps}
\end{figure} Figure 12: Completeness of our follow-up observations. This histogram shows the number of X-ray sources per discrimination probability P for PMS stars (17 RASS sources identified with new PMS stars plus four RASS sources identified with previously known PMS stars), otherwise coronally active stars (nine ZAMS stars plus seven dKe/dMe stars), 68 X-ray sources identified with other objects, and all 206 RASS sources in CrA

In Fig. 12, we plot the number of CrA RASS sources per discrimination probability P, namely for PMS stars, otherwise active stars, other objects, and unidentified RASS sources. If we would have pre-selected TTS candidates using the Sterzik et al. (1995) method, i.e. if we would have done optical follow-up observations only for RASS sources with a discrimination probability of, say, $P \ge 0.45$, we would have obtained a high success rate by loosing only one TTS.

Now, for a discrimination probability of, say, P=0.5, the reliability (rel) of the TTS candidate selection is 0.45, based on the classified sub-sample. The reliability number gives the fraction of real TTS (real according to our spectroscopy) among those X-ray sources with discrimination probability above some threshold, e.g. $P \ge 0.5$. The fraction of lost unidentified TTS is 0.17, which is the number of real TTS to be expected (according to their discrimination probability values) among those X-ray sources not observed by optical spectroscopy. Because there is a total of N = 46 sources with $P \ge 0.5$and 160 below this threshold, the expected number of TTS hidden in the RASS sample is

\begin{displaymath}\quad \le~N(P \ge 0.5) \cdot {\rm rel} + N(P < 0.5) \cdot {\rm loss}~=~48.
\end{displaymath}

Hence, there should be a total of $\le 48$ X-ray sources with TTS as true counterparts (with an X-ray flux above the RASS flux limit). The lower limit can be found when considering only those RASS sources which are unclassified, but do have an optical counterpart; there are $\tilde N = 46$ such sources with $P \ge 0.5$ and 46 with P < 0.5. Hence, there should be

\begin{displaymath}\quad \ge~~\tilde N(P \ge 0.5) \cdot {\rm rel} + \tilde N(P < 0.5) \cdot {\rm loss}~=~29
\end{displaymath}

RASS sources with TTS as true counterparts. Of those 29 to 48 sources, 21 are identified as such, namely as previously known or newly found TTS. The remaining eight to 27 still unknown TTS should be found among as yet unidentified RASS sources. According to Fig. 10, most of those missing TTS identifications are due to the magnitude limit in the catalogs used here (e.g. the GSC).


next previous
Up: Search for young stars cloud

Copyright The European Southern Observatory (ESO)