Both in G353.1+0.6 and in G353.2+0.9 the molecular line profiles result from the superposition of the emission of different components with scale sizes from 1 to 0.5 pc, the lower value probably being set by the instrumental resolution. The HII masses, as given by Fea90, are only a small fraction of the total molecular masses observed in the two fields, at most of the order of few percent, although in G353.2+0.9 they are comparable to the fraction of mass of each of the small clouds which are lined along the southern front. Also the HII densities ( cm-3), derived from the high resolution VLA observations of Fea90, are lower than the H2 densities ( - cm-3), meaning either that the ionized gas has simply expanded or/and that the HII regions are very clumpy and the ionized gas has a volume filling factor .
The morphology of G353.1+0.6 is that of an evolved star forming region viewed edge-on, with stars/HII region/molecular cloud aligned in a south-north direction. Low resolution FIR (McBreen et al. 1983) and IRAS (Persi et al. 1986) maps closely resemble low-resolution radio-continuum maps (e.g. Schraml & Mezger 1969), indicating that the IR peak lies south of the molecular cloud and suggesting that the source of heating for the dust and the cloud is the radiation from the stellar cluster, of which N49 is the brightest star. The fragmentation of the cloud is also revealed by the structured and extended form of the ionization front, where Fea90 have found radio emission from ionized surface layers of small molecular blobs. These blobs must have densities of the same order as that of the parent cloud ( cm-3). The overall morphology is reminescent of a very late stage of a blister-type configuration and indicates that star formation has occurred at the edge of the molecular cloud. No traces of recent star formation were found close to the ionization front or within the molecular cloud (Fea90).
The geometry of G353.2+0.9 is rather different from what was expected on the basis of previously available data (see e.g. Fea90). In fact, only a weak "bar" of molecular gas was found to the south of the sharp ionization front observed in H and in the radio-continuum, whereas the majority of the molecular emission comes from behind the ionized gas and to the north of it. In this case, then, a late stage blister configuration is viewed face-on. The molecular components associated with the ionized gas (C, E, G, H) are just a small part of a much larger molecular complex. The low resolution IRAS maps (Persi et al. 1986) seem to indicate that the IR peak is located near the molecular peak, north of the ionization front. However, a high resolution map at 10 µ:m (Frogel & Persson 1974) shows an elongated resolved structure coinciding with the ionization front. Just like G353.1+0.6, G353.2+0.9 is a blister type HII region, but seen face-on, and with active star formation.
4.2. Molecular gas kinematics
Velocity dispersions can be assessed for each molecular component by averaging the 13 CO(1-0) gaussian widths (FWHM) over small areas around the peaks. In both G353.1+0.6 and G353.2+0.9 the average FWHM's are large ( ), and virial masses are systematically greater than LTE masses, at least for the smallest components. In G353.1+0.6 the clouds with larger dispersions are the two main subcomponents of A (FWHM ), north of the ionization front, whereas A, B and F in G353.2+0.9 have lower average line widths (FWHM ) and seem closer to virial equilibrium (see Tables 2 and 3). These results may be due to the different ages of the associated HII regions and/or to the different geometry with respect to the exciting stars. The components closer to the ionization front on the plane of the sky (C and E in G353.1+0.6 and C, E, G and H in G353.2+0.9) have similar dispersions ( 2-3 ) and in G353.2+0.9 their virial masses are much greater than their LTE masses. In both regions the corresponding 12 CO(1-0) average gaussian FWHM's are 0.5-1 larger than those of 13 CO(1-0); for O, and CO only few data are available, but their dispersions seem generally of the order of those of 13 CO(1-0).
The interaction between HII region and molecular cloud affects the kinematics of the gas, as indicated by the appearance of red-shifted and blue-shifted molecular emissions towards the ionization fronts. In G353.1+0.6 cloud A is composed of two main subcomponents at and and dominates the field north of the radio-continuum emission. Component E ( to -8 ) must lie in front of this emission, and is blue-shifted with respect to A, whereas component C ( ), which is red-shifted with respect to A, could be located behind the ionization front with respect to the observer. This suggests that the ionized gas is blowing a "bubble" of molecular matter. Rough calculations based on the "rocket effect" (see e.g. Spitzer 1978) indicate that, assuming a systemic velocity equal to (H109 ), molecular fragments surrounding the ionization front might be accelerated to the of C and E in a time yrs, comparable to the main-sequence lifetime of an O star. In G353.2+0.9 the situation is more complex. North of the HII region emission at two distinct velocities of about -6 and -2 are clearly visible (see Fig. 10); these velocity fields must have been pre-existing, since they are located far from the radio-continuum emission, and indicate the presence of two distinct clouds. In this case the velocity difference may originate from a larger scale pre-existing velocity pattern. However, around the ionization front the associated components show a large range in velocity. In fact, the molecular emission associated with the elephant trunk, C, has a , whereas that located south of it, E, has a . Components G and H have and , respectively, and lie west and east of E. Since C and E are almost symmetrical with respect to IRS 4, one of the possible ionizing sources (Fea90), and to its associated compact radio source, we can figure that, also in this case, the ionized gas is blowing a bubble of molecular matter, where C is facing the observer and E is on the opposite side. Components E, G and H may then be forming an expanding ring south of IRS 4. Again, calculations based on the rocket effect indicate that this velocity configuration might be reached in yrs. Thus, the correspondence between the radial velocities of A and C would be coincidental and star formation might have occurred preferentially in the molecular clouds B, C and F, rather than in A.
4.3. Molecular and ionized gas
We can estimate an age for the obscured ionized region in G353.1+0.6 by determining the rate of mass erosion of A, following the model of Spitzer (1978). We assume that the ionizing source is N49 and use the physical parameters given by Panagia (1973) for an O5 V star. Then we obtain yr-1 which, using the total HII mass estimate (81 ) of Fea90, yields yrs for the time needed to ionize that amount of gas. This time is less than the main-sequence lifetime of an O star. On the other hand, the lifetime of component A against photo-evaporation is yrs, of the order of or slightly greater than the main-sequence lifetime of an O star. Since we do not know the exact geometrical relationship between exciting stars and molecular clouds in G353.2+0.9, we cannot perform the same kind of calculation, although we might expect similar results. Thus, a considerable fraction of molecular gas may survive the erosion from UV radiation of the associated OB stars.
A reasonable evolutionary scenario for the HII region G353.1+0.6 can be constructed by trying to relate the original molecular gas between N49 and the present molecular boundary. If we assume that N49 formed at the very edge of a spherical cloud with a density of cm-3 and a diameter of 0.7 pc (the distance of the star to the ionization front) the mass of that cloud was . The ionization of the gas might have originated the present optical nebula (with an optical diameter of at least 3 pc) in yrs if the HII gas expands at the sound velocity ( ). This time is comparable with the lifetime of the ionized gas given above. Interestingly, the density in this gas after the expansion should be cm-3, similar to the upper limit of the electron density estimated by Fea90 (60 cm-3). Since the emission measure decreases with the fifth power of the expansion factor, this diffuse gas becomes undetectable in the radio-continuum. For the more compact structure of G353.2+0.9, its higher mean electron density and radio-continuum brightness temperature suggest that in this case the ionized gas is still close to the parent clouds and has not yet expanded significantly.
4.4. Effects of UV radiation
If we assume a distance between N49 and cloud A surface of 1 pc and use the physical parameters given by Panagia (1973) for an O5 V star, we get a FUV flux density which is a factor greater than the average interstellar flux density. The 100/60 µ:m colour temperature has been evaluated for both regions by Persi et al. (1986) and is 43 K towards G353.1+0.6 and 45 K towards G353.2+0.9. Hollenbach et al. (1991) have studied a PDR model which gives the 100/60 µ:m colour temperature versus the FUV flux density (see their Fig. 14): a colour temperature of 45 K corresponds to a FUV flux density of times the average interstellar flux density, which agrees with our estimate in G353.1+0.6. For G353.2+0.9 (see Sect. 3.2.7) we have obtained an upper limit of times the average interstellar flux density on the surface of the small clouds located south of the ionization front. The IRAS maps of the latter region by Persi et al. (1986) suggest that the far-infrared peak is close to the molecular peak, well north of the ionization front, where the FUV flux density might well be times the average interstellar flux density, in agreement with the model of Hollenbach et al. (1991). The extended 10 µ:m structure found by Frogel & Persson (1974), however, indicates the presence of hot dust close to the ionization front, but obviously this feature has been lost in lower resolution FIR observations. Thus, since in both regions 100/60 µ:m colour temperatures, 12 CO(1-0) peak temperatures and mean densities of the molecular gas are similar, the 12 CO(1-0) brightness temperature of the bulk of molecular gas seems unaffected by the age of the bordering HII region and is related only to the FUV flux density and to the gas density.
4.5. Heating of the molecular clouds
There are other indications that the heating model assumed for both regions, i.e. UV radiation from sources at the edge of the cloud, is correct. In G353.1+0.6 the excitation temperature of cloud A derived from O line ratios peaks at (0,0) and then suddenly decreases when moving northward, away from the ionization front. Although rather uncertain, the excitation temperature seems to change (at least for the subcomponent at ) from K at the peak to K towards the projected cloud center. Along the strip at , the 12 CO(1-0) excitation temperature peaks at and, with the exception of the (0,0) position, is some 5-10 K higher than the O excitation temperature. This clearly indicates that cloud A is colder inside and that the surface layer bordering the HII region is warmer than the more distant side. Then, the heating source is external and located south of the cloud. The same, although less clearly, is suggested by the excitation temperatures for clouds A and B in G353.2+0.9, that also seem warmer towards the HII region. The existence of a temperature gradient is also indicated in Fig. 15, which shows the [12 CO(1-0)] and the [13 CO(1-0)] versus the extinction (determined from Eq. 3) in both regions. All available data can be approximately fitted by a relation but with different T. This implies that each line samples an almost constant excitation temperature region, but the 12 CO(1-0) line is sensitive to a warmer envelope while the 13 CO(1-0) line traces an inner, cooler part of the molecular cloud. The dependence of brightness temperature on would then be mainly determined by the opacity rather than by differences. The 12 CO(1-0) data can be fitted by and , whereas, assuming LTE, K, a line width , the N (H2)/N (13 CO) ratio of Dickman (1978) and an isotopic ratio X (12 CO)/X (13 CO) , a theoretical value is obtained. This may be due to a residual 12 CO(1-0) beam dilution which lowers at cloud edges (i.e. at low to moderate locations), thus deforming the curve before it saturates. Instead, the 13 CO(1-0) points can be fitted by K (G353.1+0.6) and K (G353.2+0.9), and , which agrees quite well with the theoretical value for K and a line width . If there is a gradient in both regions, then column densities calculated with 's derived from [12 CO(1-0)] must be revised. But we find the error is at most %, and this certainly would not change the results of the previous sections.
4.6. Line ratios
The clearest difference between the two regions is in the ratios, which are in G353.1+0.6 (Sect. 3.1.7) and in G353.2+0.9 (Sect. 3.2.2). Because the latter values pose a few problems for their interpretation, e.g. by an LVG model, we have suggested the existence of a warmer and low density interclump medium affecting the 12 CO ratios in G353.9+0.2, as studied by Pagani et al. (1993) for the case of RCW 34. However, it is not clear whether the difference in line ratios between the two regions is due to geometrical effects (face-on versus edge-on structure) or to the different ages of the nebulae (e.g., after some time the interclump medium may be photoevaporated by the UV radiation), or both. Towards the S.E.C. line ratios are , i.e. "normal". On the other hand, 12 CO(1-0) self-absorption by cooler clumps with small velocity dispersion has been found in G353.1+0.6; this may be due to the exciting stars/cloud geometry, since molecular clumps not facing the HII region directly but still in front of the observer may not be efficiently heated by UV radiation. Being cooler, the line of sight components will appear as absorption dips.
© European Southern Observatory (ESO) 1997
Online publication: June 30, 1998