3 The nature of the X-ray sources in NGC 6530

We have plotted the X-ray sources with a single optical counterpart in a (B-V,V) colour-magnitude diagram (Fig. 7). For this purpose, we adopted <E(B-V)>  = 0.30, R_V = A_V/E(B-V) = 3.1 and DM = 11.25 (van den Ancker et al. 1997). Although they found that some highly reddened stars show anomalous extinction, van den Ancker et al. (1997) proposed that the average extinction law of the intercluster material is normal (R_V = 3.1).

We caution that there could be a number of foreground (or background) sources that contaminate our sample. Nevertheless, for the sake of simplicity, we have adopted the distance of NGC 6530 for all the sources. In Table 1, we include the membership probability from the proper motion study of van Altena & Jones (1972) whenever this information exists. However, Sung et al. (2000) caution that the proper motion study might be problematic since low membership probabilities were assigned to some early-type stars. In their study, Sung et al. therefore favour purely photometric criteria.

Apart from a group of rather bright ($V \leq 12$) stars that are probably either early-type OB stars, Herbig Ae/Be stars (see below) or foreground objects, we find a group of objects lying to the right of the ZAMS (by about 0.5 in B-V). Note that this result is preserved if we adopt the reddening properties suggested by Sung et al. (2000) instead. Five of these objects display H$\alpha$ emission and are classified as PMS stars (Sung et al. 2000). It seems therefore very likely that this family of X-ray sources corresponds to intermediate- and low-mass PMS objects.

3.1 A population of pre-main sequence stars?

Low-mass PMS stars are classified according to their infrared properties into classes 0, I, II and III (see e.g. Feigelson & Montmerle 1999). Class 0 and I objects are deeply embedded in their nascent molecular cloud and due to their heavy circumstellar extinction, the X-ray emission from these objects is strongly attenuated. These protostars have X-ray luminosities that range from 10²⁸ to 10³⁰ erg s^-1 with occasional flares of order 10³¹ erg s^-1 (Carkner et al. 1998) and are best detected in the hard X-rays (e.g. Feigelson & Montmerle 1999). Class II pre-main sequence stars are surrounded by a thick disk. The most distinctive feature of these so-called classical T Tauri stars is their H$\alpha$ emission. Sung et al. (2000) used this criterion to identify 58 PMS stars in NGC 6530 with strong H$\alpha$ emission and 17 PMS candidates with weak H$\alpha$ emission. As the PMS stars evolve, it is thought that their disks dissipate and the PMS stars become Class III objects (or weak-line T Tauri stars) that have weaker or no H$\alpha$ emission. At least in the Taurus-Auriga-Perseus complex, the weak-line T Tauri stars are found to be X-ray brighter than the classical T Tauri stars (Stelzer & Neuhäuser 2001).

The X-ray emission of T Tauri stars is probably due to solarlike magnetic activity or a magnetic interaction with the surrounding protostellar disk. However, the X-ray luminosities are up to 10³ times larger than that of the Sun (Neuhäuser 1997). Low-mass PMS stars in the T Tauri stage have multi-temperature thermal X-ray spectra with a soft component in the range $kT_1 \in [0.2, 0.45]$ keV and a harder component with $kT_2 \in [1.3, 2.6]$ keV. These objects are usually variable X-ray emitters that exhibit flare events with a fast rising curve followed by a slower decay (see e.g. Feigelson & Montmerle 1999 for a review on their high-energy properties).

In the lower panel of Fig. 7, we have plotted the unabsorbed luminosities (in the 0.5-5.0 keV band) as a function of V. For the brightest sources, we used the luminosities inferred from the spectral fits in Table 4 and assuming a distance modulus of 11.25. For those sources for which we were not able to perform a spectral fit, we have converted the observed count rates into unabsorbed fluxes assuming a thermal plasma model with kT = 2 keV, an interstellar column density of $N_{\rm H} = 5.8 \times 10^{21}~E(B-V) = 0.17 \times 10^{22}$ cm^-2 (Bohlin et al. 1978) and DM = 11.25. Note that kT = 2 keV is in better agreement with the mean temperature inferred from our spectral fits than the "usual'' kT = 1 keV. Adopting kT = 1 keV instead of 2 keV would reduce the luminosities by a factor 0.56 (-0.25 dex).

In summary, among the objects with an intrinsic luminosity exceeding 10³¹ erg s^-1 and having a single optical counterpart, there are 57 X-ray sources with an optical counterpart fainter than 12th magnitude and we suggest that most of these objects are good candidates for T Tauri stars. The strong flare that we observe in the light curve of SCB 731 provides further evidence that at least some of these sources are related to PMS stars. The fact that only a few X-ray selected PMS candidates display an H$\alpha$ emission is in agreement with the suggestion that weak-line T Tauri stars are intrinsically X-ray brighter than the classical T Tauri stars (Stelzer & Neuhäuser 2001).

The unabsorbed X-ray luminosities of the brightest sources are found to cluster around a few times 10³¹ erg s^-1. Let us emphasize that these values are slightly larger than expected for weak-line T Tauri stars and much larger than those of classical T Tauri stars (Neuhäuser 1997). It seems unlikely that this result is due to uncertainties on the distance of NGC 6530; most of the recent determinations of the cluster distance agree nicely (Sung et al. 2000; van den Ancker et al. 1997). In any case, a sizeable fraction of our sources have $L_{\rm X}/L_{\rm bol}$ exceeding 10^-3. The most likely explanation for these extreme values is that our data only reveal the tip of the X-ray luminosity function of PMS stars in NGC 6530. The cluster could harbour many more PMS stars with X-ray fluxes below our detection threshold ($\sim$4- $8 \times 10^{30}$ erg s^-1). Assuming $L_{\rm X}/L_{\rm bol} \sim 10^{-3}$, this X-ray luminosity threshold corresponds to $M_{\rm bol} \sim 4.3$. Most PMS stars less massive than 1.0 $M_{\odot }$ were therefore not detected in our observation.

We have constructed the Hertzsprung-Russell diagram of the X-ray selected stars in NGC 6530 using the $T_{\rm eff}$ versus V - I calibration and the bolometric corrections for main-sequence stars from Kenyon & Hartmann (1995). Figure 8 compares the location of the objects with the pre-main sequence evolutionary tracks of Siess et al. (2000) for Z = 0.02 and without overshooting (note that these evolutionary models include neither rotation nor accretion).

We find that the bulk of the X-ray selected PMS stars appear to have masses between 1.0 and 2.0 $M_{\odot }$. A comparison with the isochrones indicates that most of these objects have ages ranging from 4 to 20 Myr. Only a few more massive objects appear to have ages below 1 Myr. We caution that a comparison of these results with other age determinations of PMS stars in NGC 6530 would be difficult because of the use of different calibrations and to the uncertainties related to the evolutionary calculations (see e.g. the discussion in Siess et al.). However, in Fig. 8, we have also included the PMS stars and candidates selected from their H$\alpha$ emission and which do not appear as X-ray sources in our data. The two samples of PMS stars do not reveal significant age differences, most of the H$\alpha$ selected objects also fall between the 4 and 20 Myr isochrones. This result is quite interesting: at least in NGC 6530 there is no clear evidence for an age difference between weak-line and classical T Tauri stars. This conclusion is in line with the finding of Stelzer & Neuhäuser (2001) that the overall age distribution of weak-line and classical T Tauri stars in the Taurus-Auriga-Perseus region is mixed.

Let us emphasize that there could be a difference in the mass distribution of these two categories of PMS stars. First of all, with the present data sets it is impossible to compare the lower mass limit of the two samples since the H$\alpha$ selection criterion allows to identify fainter (and hence less massive) PMS objects than the X-ray criterion. However, we note that while both categories contain more or less the same number of objects in the 1.0-1.5 $M_{\odot }$ range, there are many more X-ray selected PMS stars in the 1.5-2.0 $M_{\odot }$ interval.

Figure 8: Hertzsprung-Russell diagram of the X-ray selected stars in NGC 6530. Evolutionary tracks from Siess et al. (2000) for masses of 0.5, 0.7, 1.0, 1.5, 2.0, 2.5, 3.0, 4.0, 5.0, 6.0 and 7.0 $M_{\odot }$ are overplotted. Filled symbols indicate X-ray detected objects without H$\alpha$ emission, whereas encircled asterisks stand for PMS stars selected from their H$\alpha$ emission. The open circles and open triangles stand for H$\alpha$ selected PMS and PMSc stars from Sung et al. (2000) respectively. The thick solid line shows the ZAMS, while the dashed lines correspond to isochrones for ages of 0.5, 1.5, 4.0, 10.0 and 20.0 Myr (for clarity only the first two isochrones are labelled).

3.2 Early-type stars

Five Be stars in NGC 6530 were found to exhibit a strong IR excess and were suggested to be Herbig Ae/Be stars by van den Ancker et al. (1997): LkH$\alpha$ 108 (SCB 240), LkH$\alpha$ 112 (SCB 583), LkH$\alpha$ 115 (SCB 879), Walker 29 (SCB 427) and Walker 303. These stars are believed to be intermediate-mass ($\sim$5.5-14 $M_{\odot }$, van den Ancker et al.) young stellar objects (YSOs). X-ray emission from Herbig Ae/Be YSOs has been reported by Hamaguchi et al. (2001). The objects observed by Hamaguchi et al. (2001) were found to display high temperature thermal emission ($kT \sim1$-5 keV) with luminosities in the range 10³⁰-10³² erg s^-1.

Walker 29 and Walker 303 are not detected in our soft-band X-ray images. LkH$\alpha$ 112 is marginally detected (with a combined likelihood of 9.9), but another star (SCB 581) lies also inside the 9 arcsec radius around the position of the XMM source. LkH$\alpha$ 108 lies inside the "error box'' of source # 75, together with SCB 228 and 229. The only unambiguous detection of X-ray emission from a Herbig Ae/Be star in our sample concerns LkH$\alpha$ 115 (source #72), which is detected with an unabsorbed X-ray luminosity of $3.9 \times 10^{31}$ erg s^-1. We also note that the X-ray spectrum of source #11 (SCB 182, B2.5 V) yields a surprisingly high temperature ( $kT \sim 3.2$ keV) for a B-star. This suggests that SCB 182 could also be an intermediate-mass YSO, despite the fact that van den Ancker et al. (1997) did not classify it as a Herbig Be star.

High-mass young stellar objects could also emit X-rays. Chandra observations of the Orion Nebula Cluster revealed a huge number of X-ray emitting PMS stars over a broad range of masses (Garmire et al. 2000) including the detection of emission from massive stars that are just settling down on the ZAMS. Rho et al. (2001) reported on ROSAT and ASCA observations of the Trifid Nebula (M 20) which is thought to be in a "pre-Orion'' star forming stage. Two massive YSOs as well as one low-mass PMS star were identified as counterparts of ROSAT sources. Kohno et al. (2002) obtained a Chandra observation of the Monoceros R2 cloud and detected X-ray emission from four massive YSOs which were found to have a mean temperature of $kT \sim 3$ keV and to display flare-like events, much like their low-mass counterparts. This led Kohno et al. (2002) to suggest that the massive YSOs produce X-rays through the same magnetic activity as low-mass PMS stars. This activity would continue until the ZAMS phase and the onset of the stellar wind.

In NGC 6530 it seems rather unlikely to find high-mass YSOs that have not yet reached the ZAMS (at least outside the HG region). In fact, the earliest star in our field of view (apart from 9 Sgr and H 36) is HD 164816 (O9.5 III-IV) which has probably already evolved off the main sequence. Accordingly, van den Ancker et al. (1997) suggested that the formation of stars more massive than 8 $M_{\odot }$ has already stopped in NGC 6530. An exception to this rule could be the Hourglass Region which we shall discuss in the next section.

Finally, it is quite remarkable that some of the optically brightest ( $V \leq 10.0$) stars in NGC 6530 are not detected in our EPIC data. These include the late-type stars HD 164584, HD 164948 and SCB 444, as well as a number of main-sequence B-type stars (HD 164865, SCB 466, 588, 599, 647, 667 and 708) that have spectral types B1 or slightly later according to van den Ancker et al. (1997). The latter result is somewhat surprising. In fact, ROSAT all-sky survey results reported by Berghöfer et al. (1997) suggest that stars of spectral type B1 can have X-ray luminosities of order a few times 10³¹ erg s^-1 and most of them should therefore be detected in our data. We conclude that most of the early B-type main sequence stars in NGC 6530 may have X-ray luminosities on the lower side of the $L_{\rm X}/L_{\rm bol}$ relation proposed by Berghöfer et al. (1997).