4 The Hourglass Region

Our EPIC data also reveal X-ray emission from the Hourglass Region in M 8. Beside the ionizing star Herschel 36, the HG also harbours the ultra-compact H II region G 5.97-1.17 as well as a number of infrared sources some of which are believed to correspond to embedded massive stars. In fact, Woodward et al. (1990) performed near infrared imaging of the HG, unveiling a Trapezium-like stellar cluster of very young hot stars at the core of the HG. G 5.97-1.17 could be a YSO (probably a late-B star) surrounded by a circumstellar disk that suffers photoevaporation by the radiation from Herschel 36, similar to the so-called proplyds in the Orion Nebula (Stecklum et al. 1998). Herschel 36 is probably a massive YSO that is more evolved than the exciting stars of ultra-compact H II regions. In fact, Stecklum et al. (1995) reported infrared and HST H$\alpha$ images that reveal an elongated jet-like structure associated with H 36 oriented roughly perpendicular to the elongated dust structure around the star that could be indicative of a circumstellar disk.

Woodward et al. (1986) proposed a "blister'' scenario of the star formation in the HG where the massive star H 36 formed near the edge of the molecular cloud. This picture is also supported by the results of Chakraborty & Anandarao (1997, 1999). These authors studied the kinematics of the HG and found the H  II bubble around H 36 to be expanding which suggests that the ionization front is still destroying the surrounding molecular cloud and pressure equilibrium has not yet been reached. In addition, these authors reported high velocity flows to the south of H 36 that indicate a Champagne flow.

Figure 9: EPIC X-ray contours superimposed on a HST WFPC2 H$\alpha$ image of the HG region in the Lagoon Nebula. The X-ray contours reveal an elongated X-ray source at the location of the O7 V star Herschel 36 and of the UC H II region G 5.97 -1.17 as well as an emission lobe extending over the southern part of the optical HG nebula. The EPIC contours were smoothed by convolution with a Gaussian of $\sigma = 2$ pixels (1 pixel = 1 arcsec2) and the contour levels correspond to 0.3, 0.4, 0.55, 0.81 and 1 EPIC (MOS1 + MOS2 + pn) count per arcsec2 in the 0.5-1.2 keV band. The X-ray source below the HG is associated with SCB 142 and 146 (see Table 2).

The EPIC soft-band image of the HG region (Fig. 9) reveals an extended X-ray emission probably due to the contributions of the young stellar object H 36, the UC H II region and possibly a diffuse emission from a hot bubble created by the wind of Herschel 36. In particular, we find no correlation between the soft X-ray emission and the embedded IR sources discussed by Woodward et al. (1990). While the X-ray emission peaks around the position of H 36, it presents also an enhancement towards the south-east of H 36 (roughly coincident with the southern lobe of the optical HG nebula).

Woodward et al. (1986) found that the electron density (4000 cm^-3) on the eastern side of the ionized cavity is about ten times larger than the electron density on the western side (see their Fig. 15). Therefore, we find that the X-ray emission is enhanced over the higher density part of the HG region.

We have extracted EPIC-MOS spectra of the Hourglass in the energy range 0.2-1.2 keV and fitted them simultaneously with the ROSAT-PSPC spectrum extracted from observation rp900374n00. The source counts were extracted over an elliptical area designed to avoid contamination from neighbouring sources. The background-corrected spectra can be fitted ( $\chi^2_{\nu} = 1.25$, 55 d.o.f.) with an absorbed mekal model with $N_{\rm H} = (1.11^{+.15}_{-.17}) \times 10^{22}$ cm^-2, kT = 0.63^+.07_-.05 keV. The observed and unabsorbed fluxes in the energy band 0.2-2.0 keV are respectively $1.1 \times 10^{-13}$ and $1.7 \times 10^{-12}$ erg cm^-2 s^-1. The latter value corresponds to an integrated intrinsic luminosity of $6.6 \times 10^{32}$ erg s^-1. Assuming that the X-ray emission from H 36 roughly follows the $L_{\rm X}/L_{\rm bol}$ relation proposed by Berghöfer et al. (1997), we expect H 36 to contribute $1.5 \times 10^{32}$ erg s^-1, i.e. about one fourth of the total X-ray luminosity of the HG region.

Although we cannot completely rule out that the shape of the contours in Fig. 9 might result from the superposition of the PSFs of several faint point sources, it seems very tempting to associate the soft X-ray emission of the HG with a diffuse emission from a hot plasma around H 36. Diffuse X-ray emission has been observed in a few H  II regions only. For instance, Seward & Chlebowski (1982) found an extended diffuse X-ray emission in the Carina Nebula. They attributed the heating of the emitting gas to the interaction of the stellar winds of the rich population of early-type stars with a dense cooler cloud. Wang (1999) reported diffuse X-ray emission arising in blister shaped regions of the 30 Dor complex outlined by loops of H  II gas. Such a description could also suit the X-ray emission from the Hourglass Region. On the other hand, recent Chandra observations of the massive galactic starburst NGC 3603 revealed an extended diffuse X-ray emission in the immediate surroundings of the cluster core (Moffat et al. 2002). Moffat et al. suggest that the diffuse X-ray emission may arise from multiple merging/colliding hot stellar winds, with a rather low contribution of unresolved PMS stars. Since the HG region is a much less extreme star formation region, a superwind scenario (as in NGC 3603) that would require a large concentration of hot massive stars, is less likely to account for the X-ray emission seen in the HG.

As pointed out above, the density distribution in the vicinity of the Hourglass is believed to be highly non-uniform (Woodward et al. 1986) and a sophisticated numerical model such as the one developed by Comerón (1997) is therefore needed for a detailed comparison. In fact, simulated X-ray maps of a blister model presented by Comerón show that the X-ray emission consists of two components: an extended one from the bubble expanding into the intercloud (low-density) medium and a compact one in the vicinity of the star (i.e. in the high density part still contained in the molecular cloud). The simulations of Comerón (1997) indicate that the intrinsic soft X-ray emission from the compact component should dominate the extended emission component. The observed enhancement of the extended emission over the high density part of the HG is at least in qualitative agreement with this description.

Could the stellar wind of Herschel 36 (O7 V) account for the X-ray luminosity of a wind-blown bubble? The theory of these bubbles has been considered by Castor et al. (1975) and Weaver et al. (1977), with further additions by Mac Low & McCray (1988) and Chu et al. (1995) amongst others. The stellar wind sweeps up a dense shell of material, but within the shell there will be a region of hot shocked gas (see e.g. Strickland & Stevens 1999 for a schematic view of the structure of a wind-blown bubble). Within the framework of this theory, the parameters of the bubble depend upon the wind luminosity of the star, the density of the ambient medium and the age of the bubble (see e.g. Chu et al. 1995).

To derive an order of magnitude estimate, we adopt $T_{\rm eff} = 39~000$ K and $\log{(L_{\rm bol}/L}_{\odot}) = 5.31$ for H 36. From these parameters, we expect a mass-loss rate $\dot{M} = 4 \times 10^{-7}$ $M_{\odot }$ yr^-1 and a terminal wind velocity $v_{\infty} = 2300$ km s^-1 (Howarth & Prinja 1989; Prinja et al. 1990). This yields a kinetic luminosity of the stellar wind ( $L_{\rm wind} = 1/2~\dot{M}~v_\infty^2$) of $6.7 \times 10^{35}$ erg s^-1. For simplicity, we assume that the ambient density is constant at the higher value derived by Woodward et al. (1986) with $n_{\rm e} = 4000$ cm^-3. Adopting an expansion velocity of the HG of $\sim$10 km s^-1 (Chakraborty & Anandarao 1997, 1999) and an angular radius of the cavity in the molecular cloud of about 25 $^{\prime\prime}$ (corresponding to 0.2 pc at a distance of 1.78 kpc), we infer a dynamical age of about 20 000 yrs.

From these values, we derive the theoretical properties of the bubble surrounding H 36. The central temperature of the bubble should be $\sim$ $5 \times 10^6$ K (i.e. $kT \sim 0.44$ keV), the predicted radius is 0.4 pc and the expected X-ray luminosity amounts to $L_{\rm X} \sim 10^{35}$ erg s^-1.

We note that the predicted radius is larger than the observed radius. This discrepancy is reminiscent of similar problems reported in the literature when the wind luminosity - age - radius relation from the bubble theory is compared to actual observations (see e.g. Oey 1996; Nazé et al. 2001 and references therein). The most outstanding discrepancy between the model and our observations of the HG region concerns the expected X-ray luminosity which is more than a factor 175 larger than the observed one. Note that mass-loading of the bubble cannot explain the discrepancy since one expects a mass-loaded bubble to be even more X-ray bright (Arthur et al. 1996). Part of this discrepancy may instead be due to our estimate of the kinetic luminosity of the wind of H 36. For instance, Stecklum et al. (1998) noted that the IUE spectra of H 36 are similar to those of $\theta^1$ Ori C (HD 37022) and suggested hence that H 36 is close to the ZAMS and has a less developed wind than a typical O7 main sequence star. Assuming a mass-loss rate four times lower and a wind velocity half the values used above, the discrepancy would be reduced to a factor $\sim$12. Note however that Leitherer (1988) derived $\dot{M} = 7.6 \times 10^{-7}$ $M_{\odot }$ yr^-1 and $v_{\infty} = 1650$ km s^-1 for $\theta^1$ Ori C which yields about the same value of the kinetic luminosity than derived hereabove. Another parameter that might account for some of the discrepancy is the density of the ambient medium. We have adopted a rather large value which should be appropriate for the molecular cloud. However, in the blister scenario, part of the bubble expands into the much lower density intercloud medium and the X-ray luminosity of the bubble around H 36 may be set by the density of the latter.

Although the theoretical values are much higher than the observed X-ray luminosity, we conclude that the kinetic luminosity of the stellar wind of Herschel 36 is most probably sufficient to account for the existence of a wind-blown bubble.