3 Results

3.1 Source detections and identification

Figure 1 (right) shows that in the very young massive star-forming region IRAS 19410+2336 hard X-ray emission from several point sources is detected. No evidence for extended or diffuse emission is found within our detection limits (Sect. 2.1). At mm wavelengths, the two massive cores detected with the IRAM 30 m single-dish observations at an angular resolution of 11''(Beuther et al. 2002a) split up into many sub-sources when observed at higher spatial resolution (Fig. 1, right), confirming that a cluster of stars is forming. The interferometric data highlight the real massive cores, whereas the large-scale surrounding core emission is filtered out by the interferometric observing technique. A detailed analysis of the protocluster properties will be presented elsewhere. Here we are focusing on the spatial associations of X-ray, near-infrared and mm sources.

Thirteen X-ray sources are detected within the field of interest (Table 1). Remarkably, nearly all X-ray sources are, within the positional uncertainty, associated with near-infrared counterparts observed in the K-band (X3 and X5 only tentatively). Because the K-band extinction is approximately only $10\%$ of the visual extinction, in the K-band we can observe more embedded regions of star-forming cluster than possible in the optical. We recall that the extinction in the soft X-ray regime is similar to the K-band extinction, and that it drops even further going to higher X-ray energies (Casanova et al. 1995; Ryter 1996). In contrast to the K-band counterparts, only 2 sources (X2 and X8) are found in the direct vicinity of mm dust cores that are tracing the coldest and deepest embedded regions where presumely the youngest sources are found. The whole region is also covered by the 2 Micron All Sky Survey (2MASS) in the near-infrared bands J, H and K. Except for sources X2, X3, X5 and X11, which are detected by our K-band Omega Prime observation but which are too faint to be detected by the 2MASS survey, all other X-ray sources are also 2MASS point sources. A color-color diagram helps to determine the characteristics of the associated near-infrared counterparts (Fig. 2). While a few of the near-infrared sources are located around the unreddened main sequence in the color-color diagram, many sources are reddened and show near-infrared excess, suggesting that they are pre-main-sequence objects surrounded by circumstellar material. This is expected in view of the early evolutionary stage, the large core masses and the high column densities of the region (Beuther et al. 2002a). Table 1 compiles reddening estimates of the X-ray near-infrared counterparts. The $A_{\rm {v}}$ of X1 is most uncertain because X1 has the largest near-infrared excess which makes the $A_{\rm {v}}$determination more difficult. Using the reddening scale (Fig. 2), we can still estimate an approximate value of the visual extinction $A_{\rm {v}}$ of X1.

Figure 2: Near-infrared color-color diagram of X-ray associated sources based on 2MASS data. The numbers correspond to the sources numbered also in Fig. 1. The full line presents main sequence colors with three labeled spectral star types (Ducati et al. 2001), and the dashed lines show the reddening band for main sequence colors (Rieke & Lebofsky 1985). The dotted line gives the reddening scale with encircled visual extinctions ([mag]). Source X1 shows the strongest near-infrared excess.

   
3.2 Spectral information of the X-ray data

   
3.2.1 X-ray spectra and X-ray luminosities

The number of photons per detected source ranges from 7 to 19 counts within the 20 ksec observation time (Table 1). We note that we are dealing with low-number statistics, and this has to be taken into account when deriving the physical parameters of this region. Figure 3 presents the photon energies of the 13 X-ray sources versus time. As expected, we detect emission mainly between 2 keV and 6 keV in the near-infrared reddened and thus embedded sources, whereas the sources, that are near main sequence colors in the color-color diagram, emit more in the soft regime below 2 keV (see also mean values in Fig. 3 and Table 1). The deficiency of soft photons (<2 keV) in the direction of the embedded sources is most probably due to the high column densities that absorb the soft X-rays. The lack of photons above 6 keV may be a real feature of the X-ray sources, but it has also to be noted that the sensitivity of Chandra above 6 keV decreases significantly. The effective collection area of Chandra at 8 keV drops to less than a quarter of that at 4 keV. The fact that we detect hard X-ray emission (>3 keV) of the embedded sources fits in the general finding that the younger the sources are, the harder the X-ray emission is (e.g., Feigelson & Montmerle 1999 or Sect. 1). The X-ray spectra of X1 to X7 show all such hard X-ray signatures, which makes them likely to belong to the forming cluster. The distribution of photon energies cannot be explained by a low-temperature-Raymond-Smith plasma of 1-2 keV (Raymond & Smith 1977), usually used to fit the X-ray spectra of low-mass T Tauri stars (Feigelson & Montmerle 1999), because the column densities $N_{\rm {H}}$, that are necessary to attenuate all the emission below 2 keV ( $N_{\rm {H}}$ a few times 10²³ cm^-2), would imply average X-ray luminosities as high as 10³⁴ erg s^-1. This value exceeds the known luminosities of young stellar objects by orders of magnitude (Sect. 1), and it could hardly be reached even during strongest flaring events (e.g., Preibisch et al. 1993; Grosso et al. 1997; Tsuboi et al. 2000). Contrasting these hard spectra, X8 to X13 show softer spectra. While some of them might be unrelated objects, others could belong to the cluster but be in a more advanced state of evolution (for details see Sect. 3.4).

Figure 3: Photon energy versus time for the 13 detected sources within the field of interest. The dotted lines mark the mean detected photon energy in each source (see also Table 1).

In order to derive estimates of the X-ray plasma temperatures and the X-ray luminosities, we performed a detailed spectral analysis. We extracted individual pulse-height spectra and built the corresponding redistribution matrices and ancillary response files for each source. We fitted the ungrouped pulse-height spectra with XSPEC, using a single-temperature thermal plasma model (model "raymond'', Raymond & Smith 1977) plus the absorption model "wabs''. The plasma temperatures and column densities derived in these fits are compiled in Table 1. The fitting results were then used to integrate (in XSPEC) the model fluxes over the 0.2-10 keV band and to compute the X-ray luminosities. In some cases the low number of detected photons allowed only to derive lower limits to the plasma temperature. We are fully aware that, given the rather low number of photons per source, the derived spectral parameters inherit a relatively large statistical uncertainty. Nevertheless, we believe that the fitting results contain valuable information, for the following three reasons:

(1) The spectral parameters derived in the fits allow a clear distinction of two different groups among the sources: sources X1-X7 show high plasma temperatures (kT>3 keV) and high colum densities ( $N_{\rm {H}}>10^{22}$ cm^-2); in contrast, the sources X8-X13 show low plasma temperatures (kT<1.5 keV) and (with the exception of X8) low colum densities ( $N_{\rm {H}}<10^{22}$ cm^-2). These two groups exactly reflect the spatial distribution of the sources: X1-X7 are located inside the central region of the molecular clump, while X8-X13 lie outside the molecular clump (Fig. 1).

(2) The column densities derived in the spectral fits agree in nearly all cases rather well with expectations from the infrared colors of the sources (Table 1, $A_{\rm {v}}=N_{\rm {H}}/2\times10^{21}$).

(3) Getman et al. (2002) have performed fitting simulations of pulse-height spectra with low numbers of counts to estimate the statistical uncertainties of the fitting results. They found that even if the column densities and temperatures derived in the fits are subject to rather large uncertainties, the broad-band X-ray luminosities derived from the fitting results are quite reliable (statistical error $\le$35% for spectra with 15 counts).

Very likely, the X-ray emission of these objects originates from a hot thermal plasma typical for very young embedded objects (e.g., Feigelson & Montmerle 1999; Preibisch & Zinnecker 2002; Kohno et al. 2002). But as the number of photons and by that the spectroscopic information is small, other X-ray emission processes have to be taken into account, e.g., non-thermal power-law distributions as a signature of synchrotron emission, which has been observed in the radio regime by Reid et al. (1995) in the massive star-forming region W3(OH). Therefore, we also estimate X-ray luminosities with three different power-law distributions $S_{\rm {x}}\propto E^{-\alpha}$ ( $\alpha=0.5,1,2$) to derive an estimate for the influence of $\alpha$ on the total X-ray luminosity (using the web-based PIMMS tool). The derived luminosities vary by less than a factor 1.5 for the different $\alpha$. As hot plasmas are dominated by bremsstrahlung above $kT\sim 1$ keV (Fink & Trümper 1982), which can also be approximated by non-thermal power-law distributions, the X-ray luminosities derived this way are close to the luminosities calculated using the Raymond-Smith plasma (they differ only a factor 2 at maximum).

We calculated $L_{\rm {x}}$ only for X1 to X8 (Table 1), because the other sources could be foreground or background objects, and thus no reliable distance is known. We find X-ray luminosities in the 10³¹ erg s^-1 range. These values are high, and above or at the upper end of classical T Tauri X-ray luminosities, i.e., in the regime of younger Class I sources (Feigelson & Montmerle 1999) and/or more massive Herbig Ae/Be type objects (Zinnecker & Preibisch 1994; Gregorio-Hetem et al. 1998; Hamaguchi et al. 2000). Unresolved binary systems could alter the results slightly, but not by orders of magnitude.

3.2.2 Variability

We do not see strong intensity flaring in our observations (Fig. 3), but sources X2, X3 and X4 show absence of detected photons in the time interval between $6\times 10^3$ and 10⁴ s, which indicates possible variability in these sources. The other sources show a rather constant count rate within the 20 ksec of our observation. To quantify the possible variability, we performed a statistical test of the photon arrival times following the approach outlined by Preibisch & Zinnecker (2002): given the number of detected photons in one source in the total integration time, we compute the mean count rate and specify the time period in which two photons are expected assuming a constant count rate. Then we determine the maximum number of photons $N_{\rm {max}}$ detected in this source within any time interval as long as the one derived for the two photons. The Poisson probability to find $N_{\rm {max}}$ or more counts due to statistical fluctuations in a period for which two photons are expected, is given by

$$\begin{eqnarray*}P=1-\sum_{k=0}^{N_{\rm {max}}-1}~{\rm e}^{-2} \frac{2^k}{k!}\cdot \end{eqnarray*}$$

The probability of variability (pov) is given by pov=1-P (for details see, e.g., Bevington & Robinson 1992). The pov values for X3 and X4 are only $68\%$ and $85\%$, giving just marginal evidence for variability within our total observing time. Contrasting to that, pov equals $98.3\%$ for X2, therefore this source is consistent with X-ray variability.

   
3.3 Mass estimates from the near-infrared data

The 2MASS near-infrared K-band data provide a tool to estimate the masses of the infrared sources. Taking the observed K-band brightnesses m_K at the given K-band extinction ( $A_{K}\sim 0.11\times A_{\rm {v}}$) we get an estimate of the intrinsic brightness M_K of the source assuming the distance d=2 kpc:

$$\begin{eqnarray*}m_{{K}}-M_{{K}}=5~ {\rm {log}} (d)-5+A_{{K}}. \end{eqnarray*}$$

Different pre-main-sequence tracks are discussed in the literature (e.g., D'Antona & Mazzitelli 1994; Baraffe et al. 1998; Palla & Stahler 1999; Siess et al. 2000; D'Antona et al. 2000). While some of them differ significantly at the low-mass end, they are better comparable going to masses > $1~M_{\odot}$ (Siess et al. 2000). Most of the calculated tracks do not cover stars more massive than $2.5~M_{\odot}$, but Palla & Stahler (1999) calculate the pre-main-sequence evolution to $6~M_{\odot}$ and Siess et al. (2000) to $7~M_{\odot}$. At the high-mass end, which we are particularly interested in, both calculations agree well. The tracks of intermediate-mass stars (> $2~M_{\odot}$) are following almost horizontal tracks of equal luminosity in the Hertzsprung-Russel diagram, and the luminosity is almost independent of the temperature. Thus, it is possible to estimate luminosities and by that masses from our K-band observations, but the data do not give any information about the age of the objects. Siess et al. (2000) stress that any age determination below 10⁶ yr is highly uncertain. In the following, we use the tracks compiled by Palla & Stahler (1999), but the results are similar when using the tracks of Siess et al. (2000).

Comparing the intrinsic K-band brightnesses M_K with the theoretical tracks by Palla & Stahler (1999), we can derive mass estimates for the near-infrared objects. We apply this procedure for the near-infrared sources corresponding to X1, X6, X7 and X8, because only those are 2MASS detections and likely belong to the forming massive clusters (Sect. 3.2). The pre-main-sequence tracks provided by Palla & Stahler (1999) for stars up to $6~M_{\odot}$ do not comprise sources intrinsically brighter than -1.6 mag in the K-band. Sources more massive than $8~M_{\odot}$ do not have an optically visible pre-main-sequence phase, because they start nuclear burning already being deeply embedded in their natal cores (Palla & Stahler 1993). Therefore, X6 and X7 could be more massive than $6~M_{\odot}$, but it has to be noted that the derived model brightnesses M_K neglect luminosity due to the disk, i.e., heated dust in the disk, and accretion luminosity within the disk and between the disk and the protostar. Palla & Stahler (1999) stress that the understanding of these processes is still inadequate for quantitative predictions, but as the near-infrared excess in X6, X7 and X8 is not very high (Fig. 2), to our judgment the additional disk luminosity in these sources is not the dominant effect. This is different for source X1 which has considerable infrared excess and makes K-band brightnesses difficult to be compared with theoretical predictions. But as stars below $2~M_{\odot}$ are at least 2 mag fainter in the theoretical predictions than the observed brightness of X1, $2~M_{\odot}$ can still be regarded as a tentative lower mass limit for this source. Real upper mass limits are given by the total bolometric luminosity of the whole cluster of $\sim$ $10^4~L_\odot$(Sridharan et al. 2002). A luminosity of $10^4~L_\odot$ corresponds in the case of main sequence stars to a B1 star ( $13~M_{\odot}$). As a cluster of stars is forming in IRAS 19410+2336, the mass distribution of the cluster members is likely to follow an initial mass function. In spite of their larger number, the luminosity-mass relation for low-mass stars is lower ( $L\propto m^{2.8}$ for $m<1~M_{\odot}$) than for high-mass stars ( $L\propto m^{4}$ for $1<m<30~M_{\odot}$, Schatzman & Praderie 1993), and still the major contribution of the total luminosity stems from the massive cluster members. For a detailed IMF discussion of the initial source sample see Sridharan et al. (2002). As the most massive object is likely the central mm source, which is not detected in X-ray emission, the X-ray detected sources are $\leq$ $10~M_{\odot}$. Additionally, the models give the approximate luminosity of the sources and we get an estimate of the $L_{\rm {X}}/L_{\rm {bol}}$ ratio ( $L_{\rm {X}}$ is taken from Sect. 3.2 and Table 1). The parameters and results are listed in Table 2.

 

 

Table 2: Near-infrared parameters of most likely cluster sources with 2MASS counterparts. Given are masses ($\leq$ $10~M_{\odot}$), bolometric luminosities $L_{\rm {bol}}$, and the ratio $L_{\rm {X}}/L_{\rm {bol}}$ .

src	mK	$A_{\rm {v}}$	AK	MK	mass	$L_{\rm {bol}}$	$L_{\rm {X}}/L_{\rm {bol}}$
	[mag]			[mag]	[$M_{\odot}$]	[$L_{\odot}$]	[10-5]
X1	11.7	17	1.7	-1.5	>2	>10	<70
X6	11.0	18	1.8	-2.3	>6	-	-
X7	8.8	12	1.2	-3.9	>6	-	-
X8	12.8	17	1.7	-0.4	>3	>30	<2

The derived masses are in the intermediate-mass regime of Herbig Ae/Be stars, clearly more massive than usual T Tauri or Class I sources. Zinnecker & Preibisch (1994) observed X-ray emission from Herbig Ae/Be stars, and those sources showed rather soft spectra. Preibisch & Zinnecker (1995) speculate that the observations could be explained by coronal emission due to a shear dynamo. As the spectral signatures of the sources we are studying here are significantly harder than those found in their sample (Sect. 3.2), we suggest that the X-ray emission could be due to magnetic star-disk interactions as proposed for very young Class I sources (Montmerle et al. 2000). In that framework, it is possible that the sources observed in IRAS 19410+2336 are precursors of the better studied Herbig Ae/Be stars. The possible $L_{\rm {X}}/L_{\rm {bol}}$ ratio-regime (Table 2) for two sources is at the upper boundary consistent with ratios found in very young low-mass objects ($\sim$ $2\times 10^{-4}$, e.g., Feigelson & Montmerle 1999; Preibisch & Zinnecker 2002), but at the lower end it is also consistent with the $L_{\rm {X}}/L_{\rm {bol}}$ ratios found in samples of Herbig Ae/Be stars (Zinnecker & Preibisch 1994; Gregorio-Hetem et al. 1998). It could be argued that the X-ray emission is due to low-mass counterparts, but as the derived X-ray luminosities are significantly larger than typical values in the low-mass regime, most of the X-ray emission stems likely from the intermediate-mass sources.

   
3.4 The X-ray sources in detail

Source X1 and the main mm core:

the most remarkable X-ray source is X1, which does not only show strong reddening but also the largest infrared excess indicative embedded protostellar objects (Feigelson & Montmerle 1999). As the infrared derived mass is $\geq$ $2~M_{\odot}$ but the X-ray spectrum harder than usual Herbig Ae/Be stars (Zinnecker & Preibisch 1994), X1 is probably not a normal Herbig Ae/Be star but a precursor of such an object.

Figure 4: Zoom into the center of the field presented in Fig. 1. Contours and symbols are the same as in Fig. 1. Additionally, the black contours on white ground right at the center show the cm-peak, the triangle pinpoints the H2O maser position and the square the CH3OH maser positions (Minier et al. 2001; Beuther et al. 2002c).

Source X1 is the X-ray source nearest to the core center but not coincident with a mm core. Figure 4 shows that the main mm core coincides with a compact cm source as well as with H₂O and Class  II CH₃OH maser emission (Minier et al. 2001; Beuther et al. 2002c), while the X-ray source is clearly offset and coincides with a near-infrared source $\sim$6'' to the west. Thus, the main mm core is likely to contain the youngest and most massive object of the cluster, and we do not detect X-ray emission from this source, but only from another object nearby that might be slightly more evolved. Therefore, we do detect X-ray emission from pre-main-sequence sources but not from the youngest and massive center of the cluster. Assuming optically thin dust emission, we estimate the core mass of the central source from the 2.6 mm continuum data to about $80~M_{\odot}$ and the H₂ column density to $\approx$ $2\times 10^{24}$ cm^-2, corresponding to a visual extinction $A_{\rm {v}}$ of approximately 2000 (for details on mm dust calculations see Beuther et al. 2002a). Taking into account this visual extinction and assuming different Raymond-Smith plasma temperatures for the X-ray emission, we can calculate a range of upper limits for the X-ray luminosity of this source: a 3 keV plasma temperature corresponds to an upper limit of $L_{\rm {X}}<9\times 10^{34}$ erg s^-1, while a temperature of 10 keV implies $L_{\rm {X}}<4\times 10^{34}$ erg s^-1. A lower plasma temperature of 1 keV results in an upper limit of $L_{\rm {X}}<14\times 10^{36}$ erg s^-1. However, it has to be noted that the possibility remains that there is no significant X-ray emission in the earliest stages of massive star formation.

Sources X2 and X5:

source X2 is at the edge of the southern mm emission and associated with some faint K-band emission that might be a reflection nebula from the main sources to the south. As the K-band emission is weak, it is not detected by 2MASS. Being in the near vicinity of a mm core and showing a very hard X-ray spectrum, it is possible that X2 might be powered by an embedded protostellar object. It has to be noted that this source is the only one with observed X-ray variability. Source X5 is also too faint in the near-infrared to be detected by 2MASS, and it is difficult to decide whether the nearby faint K-band emission is a stellar feature or a reflection nebulae. But as X5 has the hardest X-ray spectrum of our sample, it is very likely a young protostellar object within the cluster of IRAS 19410+2336.

Sources X3 and X4:

these sources are both close to the same near-infrared source, but accurate astrometry indicates that X4 rather than X3 is associated with it (Fig. 4). Additionally, we find an optical counterpart for X4 in the Digitized Sky Survey provided by the Space Telescope Science Institute. Based on the X-ray spectrum, which is not as hard as the spectra of the other cluster members, and the optical and infrared data, we cannot distinguish whether X4 belongs to the cluster or is an unrelated foreground object. In contrast, X3 might be associated with more diffuse K-band emission (Fig. 4). As the hard X-ray spectrum of X3 is indicating an early evolutionary stage, this source most likely is a very young pre-main-sequence object belonging to the cluster under investigation.

Sources X6 and X7:

these sources also show hard X-ray spectra, and both are within the large-scale region of the southern massive core as seen in Fig. 1 (left), but no mm core corresponds to any of them. This is reflected in the color-color diagram, where they inhabit a reddened region without strong infrared excess. The infrared-derived mass estimates are not conclusive, but the data - X-ray and near-infrared - are consistent with precursors of Herbig Ae/Be stars of masses between $5~M_{\odot}$and $10~M_{\odot}$.

Source X8:

source X8 is associated with the strongest mm peak of the northern cluster as well as with a near-infrared and even an optical counterpart (as found in the Digitized Sky Survey). The 2MASS data show that the infrared source is reddened and has infrared excess, and we derive a lower mass limit of approximately $3.5~M_{\odot}$. The X-ray spectrum is softer than for the southern cluster members, thus it is possible that X8 is in a slightly more evolved state of evolution.

Sources X9 and X10:

the X-ray spectrum of these sources is softer, but they are still reddened in the color-color diagram and X9 even shows infrared excess. It is likely that they are also young stellar objects, but we are not able to determine whether they belong to the cluster or whether they are foreground objects.

Sources X11, X12 and X13:

these sources also have a rather soft spectrum, and the color-color diagram locates them on or near the unreddened main sequence (X11 is too weak to be a 2MASS detection). As they are also spatially offset from the large-scale mm core, we regard them as not associated with the massive star-forming region.