4 Discussion

4.1 Origin of the HI expanding shell

The HI structure found in this paper resembles the expanding shells often associated with O stars (Cappa & Benaglia 1998), WR stars (Arnal 1992; Arnal et al. 1999) and OB associations (Rizzo & Bajaja 1994; Rizzo & Arnal 1998). In the following we shall explore if the HI shell has been created by the stellar wind from HD56925 or from its O-progenitor. Let us estimate, with a few assumptions, some important parameters to distinguish between the two possibilities. We shall consider a spherical expanding bubble, with a radius of 50 pc, an expansion velocity of 12 km s^-1, and a mass of 1500 $M_\odot$. With these values, we estimate a dynamical time for the shell of 2.3 Myr and a kinetic energy of $2.2\; 10^{48}$ erg. The dynamical time was computed as a mean value between the adiabatic and the radiative case (McCray 1983). This time is in fact an upper limit, but a good approximation, to the age of the HI structure (Dyson 1989). This age and the kinetic energy associated to the HI shell clearly indicate that its origin is due to the main-sequence phase of HD56925. Indeed, the most massive stars, greater than 25 $M_\odot$, remains a few million years at the main sequence (Maeder & Meynet 1994) and creates interstellar bubbles of 30-60 pc in this stage (Weaver et al. 1977; García-Segura & Mac Low 1995). It is thought that WR's are descendant of these massive stars. On the other hand, the RSG, LBV or WR phases are short-lived, probably 10⁴-10⁵ yr (Maeder & Meynet 1994; Langer et al. 1994). We therefore conclude that only the O progenitor of HD56925 had sufficient time to develop the HI expanding shell reported in this paper. A similar conclusion was reached by Marston (1996) concerning to the large IRAS shell found around several WR stars, including the one linked to HD56925.

By adopting typical values for the mass-loss rate of 10^-6 $M_\odot$ yr^-1 and a wind velocity of 1000 km s^-1 for the O-progenitor star, we estimate that this star deposited into the ISM during 2.3 Myr a total energy of $\sim$ $2.3 \; 10^{49}$ erg in the form of stellar wind. If this star has blown up the shell which we are observed in HI, it implies a kinematical efficiency of nearly 10%. Although the uncertainties in these computations are large and hard to be estimated, this value for the efficiency is in good agreement with the models that predict this type of structures (Weaver et al. 1977; Van Buren 1986) and reinforces our hypothesis of a main sequence origin of the HI shell.

 

Table 2: Physical parameters of the CO emiting regions

Component		$\frac{2\rightarrow 1}{1\rightarrow 0}$	$\frac{^{12}{\rm CO}}{^{13}{\rm CO}}$	$T_{\rm ex}$	n(H2)	N(CO)	$\tau_{10}$	$\tau_{21}$	$\tau(^{13}{\rm CO}$)	m(H2)
				K	103 cm-3	1016 cm-2				$M_\odot$
A1		0.5	--	5	0.6	1.0	0.6	0.6	--	96
A2		0.5	--	6	1.0	0.4	0.8	0.8	--	28
$B_{\rm all}$		0.8	--	8	1.8	0.5	0.3	0.4	--	160
$B_{\max}^{\rm blue}$		1.0	30	10	5.0	1.6	0.6	0.8	<0.1	10
$B_{\max}^{\rm red}$		1.0	5	10	58	21.	12	--	$\sim$ 0.1	120
$B_{\rm HII}$		2.0	30	80	<1.6	0.1	<0.1	0.2	$\leq$0.1	8

The LVG approximation was used in all cases but $B_{\max}^{\rm red}$. For this particular region, we have estimated
the size of the emiting region and taken into account the beam filling factor. All masses were computed by assuming a distance of 5 kpc.

4.2 Physical parameters of the molecular gas

We have estimated the physical parameters of the molecular gas from the CO line emission by modelling the excitation of the CO lines using the Large Velocity Gradient (LVG) approximation. For comparison with the model, we have smoothed the J=  $2 \rightarrow 1$ line data to the J=  $1 \rightarrow 0$ resolution. For component B, the ¹³CO data was also taken into account in order to estimate the opacity of the ¹²CO lines. Six different components have been defined. The first three are the ambient components A1 and A2 and the most extended emission of the component B (hereafter $B_{\rm all}$). The fourth component considered is the one projected onto the southern HII region, named as $B_{\rm HII}$. Finally, we also considered the region around the CO maxima in component B, where the ¹³CO behaviour is striking. We have subdivided this component in the two velocity ranges having the greatest difference in the ¹²CO/¹³CO ratio (see Fig. 5b), named as $B_{\max}^{\rm blue}$(for the velocity interval 50-54 km s^-1) and $B_{\max}^{\rm red}$ (for the velocity interval 54-58 km s^-1). The results derived from the LVG analysis are shown in Table 2. This table gives estimates for the excitation temperature ( $T_{\rm ex}$), H₂ density (n(H₂)), CO column density (N(CO)), the opacities of the CO J=  $1 \rightarrow 0$, CO J=  $2 \rightarrow 1$ and ¹³CO J=  $1 \rightarrow 0$ lines, and the total mass of the regions (m(H₂)), derived from the two line ratios (second and third columns) and the ¹³CO intensities. A kinetic temperature ($T_{\rm K}$) of 10K was assumed for all the regions but $B_{\rm HII}$. The physical conditions in $B_{\rm HII}$ seem to be very different from those of the other regions. The large ratio between the J=  $2 \rightarrow 1$ and the J=  $1 \rightarrow 0$ lines is explained by optically thin emission with relatively high $T_{\rm K}$. The most likely $T_{\rm K}$ is about 80K and we present in Table 2 the results derived for this $T_{\rm K}$. Components A1 and A2 are subtermally excited ( $T_{\rm ex}$ < $T_{\rm K}$), with typical $T_{\rm ex}$ of 5-6K. We obtain higher excitation temperatures in component B. For $B_{\max}^{\rm blue}$ and $B_{\max}^{\rm red}$, we derived $T_{\rm ex}$ similar to $T_{\rm K}$, indicating that the lines are thermalized, and densities significantly higher than for A1 and A2.

For $B_{\max}^{\rm red}$ we did not apply the LVG method since this region has a very low value of the ¹²CO/¹³CO ratio. Such low value for this ratio usually indicates the presence of a region of very high opactiy in the ¹²CO lines. Since we have found this low ratio toward only one observed position, we conclude that the emiting region corresponding to $B_{\max}^{\rm red}$ is unresolved by the J=  $1 \rightarrow 0$ beam. To derive the physical properties in this component, we have assumed an optically thin emission for the ¹³CO and optically thick emission of the ¹²CO. For a ¹²CO/¹³CO isotopic ratio of 60 (Wilson & Matteucci 1992), we derive an optical depth of 12 for the J=  $1 \rightarrow 0$ line of ¹²CO. From the ¹²CO J=  $1 \rightarrow 0$ line intensity and assuming that the emission in this line is thermalized to the $T_{\rm K}$ of 10K, we derive a beam filling factor for $B_{\max}^{\rm red}$ between 0.1 and 0.2. With the optical depth we have estimated N(CO). Adopting a beam filling factor of 0.15 and a fractional abundance for CO of 10^-4 we have also estimated n(H₂). Although our estimates of the H₂ density can be uncertain because of the unknown $T_{\rm K}$, the large differences in the H₂ density between $B_{\max}^{\rm red}$ and the other components indicate that a rather small region has a density larger than other components.

4.3 Molecular gas interacting with the nebula and the WR star

In Sect. 3.2 we have discussed the morphological and kinematical arguments for the relationship between the different CO components and NGC2359. In the previous section, we have derived their physical conditions. In this section, we shall consider our observational findings together with those already published, in order to address the nature and the origin of the interaction of the molecular gas with the nebula and its exciting WR star. The component A1 is located mainly to the NE of the map, while the component A2 is present to the SE of the nebula. Both components are narrow (2 km s^-1) and do not show any morphological or kinematical structures that indicate any disturbance produced by the nebula or the WR star. In contrast, the component B has some kind of ubiquity in this field, in agreement with optical emission lines like the [O3]. It appears as a broad component (4.5-5 km s^-1) following the east and the south border of the HII region which surrounds the nearly-spherical bubble. We noted that the CO emission is stronger outside the optical border of the HII region, but significant emission is also present in projection onto the HII region. We think that the component B is the only component which undoubtely interacts with NGC2359, because of its morphology, the large width of the lines, the higher temperature and the striking kinematics.

As already mentioned, the eastern part and the southern border of the HII region are clearly bounded by the most intense component B. Furthermore, the presence of faint CO emission perfectly correlated with the HII region it is also striking. These results are also in agreement with the vibrationally excited H₂ emission observed at 2.2 $\mu$m (St-Louis et al. 1998). The CO emission is dense and more opaque outside the HII region, while onto the HII region it is less dense and has a higher temperature (up to 80K). This temperature is compatible with the dust temperature inferred by Marston (1991) in the same region.

At this point we wish to consider the ¹³CO behaviour in the southern part of the component B. We have only detected this molecule near the peak of the CO emission. This indicates that this is a unique region where the ¹²CO/¹³CO changes appreciably, most likely due to an increase of the opacity in the ¹²CO lines. Only 1$^\prime$ to the east of this position (essentially one beam) there is no significant emission in ¹²CO from the component B, just at the western border of the component A2. This spatial anti-correlation between components A2 and B is also present toward the southern interface. Furthermore, there is a bridge in the CO emission which seems to connect both components within a small area (see slice 3 of Fig. 4), suggesting that both components are related. The CO in the HII region might be heated by the WR radiation field (St-Louis et al. 1998), but the interface between the components A2 and B may not be excited by the same mechanism.

All these observational facts can be explained in a scenario in which a shock of 10-14 km s^-1, driven by the expanding hot bubble, impacts on the component A2. As a consequence of the shock, the gas was accelerated up to the velocities of the component B. The large width of the component B when compared with the component A2 can be understood by means of an increase of the turbulent motion due to the energy injected by the shock. Furthermore, the systemic velocity of the HI shell found at larger scales (presumably near the rest velocity of the gas) is also similar to the velocity of the component A2, supporting this scenario. The shock heats the gas and forms a thin layer of compressed material with large H₂ density and with high opacity in the ¹²CO. This thin shocked layer would remain unresolved by the beam. The kinematics shown in Fig. 4 and the low ¹²CO/¹³CO ratio at 53-56 km s^-1 towards the CO peak are very suggestive of this thin shocked layer. Indeed, the large optical depth of the CO lines can only be explained if this emission arises from a very small region unresolved by the beam, probably less than 0.2 pc wide. Furthermore, the large CO column density of $B_{\max}^{\rm red}$ as compared with $B_{\max}^{\rm blue}$ would indicate large densities for these radial velocities (up to several 10⁴ cm^-3). Obviously, in the scenario of the shocked gas responsible for the component B, $B_{\max}^{\rm red}$ would represent the material which has been recently shocked. High angular resolution observations are needed to disclose this particular region and to check our interpretation.

Although our hypothesis seems reasonable, it should be pointed out that this shocked region is small and it cannot be spatially resolved in our data. Another crucial point is the weakness of the "bridge'' which connects the components A2 and B. Higher angular resolution and more sensitive observations are needed to confirm the presence of the shocked layer and to fully study its properties. If this is the case, NGC2359 would have the first direct evidence of shocked molecular gas in a WR environment. The O-stage of HD56925 have created in the ISM the HI shell reported in this paper and the IRAS shell reported by Marston (1996). In consequence, the optical nebula and its surroundings, where we find most of the CO emission, have a more recent origin than the large scale HI shell. The HII region has a size that indicate a dynamical age of less than 10⁵ yr. This age is comparable with the WR or a previous phase, such as LBV or RSG (Maeder & Meynet 1994). On the other hand, many optical shells of enriched material are found close to the WR star (Esteban et al. 1990). This leads us to think that the material now interacting with the HII region was ejected in previous episodes of RSG or LBV phase. Chemical studies using other molecular line transitions, especially those capable of tracing the molecular gas enriched by these phases, will definitively confirm or discard these ideas.