4. Molecular gas and photodissociation regions
A photodissociation region (PDR) develops at the surface of a molecular medium illuminated by FUV radiation. In the PDR, molecular hydrogen is partly photodissociated and partly experiences fluorescent excitation, with consecutive emission of vibrational-rotational lines in the near-IR and of rotation lines in the mid-IR. H2 can also be excited collisionally if the density is sufficient, thus the physics of the infrared line emission is complex and diagnosis require spectroscopic observation of line strengths (see e.g. Sternberg & Neufeld 1999and references herein). The multi-line study of Pak et al. (1998) suggests that the H2 emission is UV-excited in several regions of the LMC whose properties are not too different from those of N 66. CO like H2 is photodissociated through absorption in UV lines and is self-shielded against photodissociation by the optical thickness in these lines. But CO is more easily photodissociated than H2 and exists only somewhat deeper into the cloud, typically in regions with a visual extinction larger than 1 magnitude compared to 0.1 mag. or less for the v=(1-0) S(1) line of H2. Their respective line emission comes mainly from these depths (see e.g. Lequeux et al. 1994for examples related to the SMC). The mid-IR Aromatic Infrared Bands (AIBs) and the continuum emission of the Very Small Grains discussed in Paper I are also strongly emitted in the PDR, simply because the matter density increases strongly when entering the front while there is still a large, relatively unabsorbed radiation flux to heat them through absorption of single photons. Thus we expect a strong correlation between H2 line, CO line and mid-IR emission. This is what is observed in N 66 as illustrated by Fig. 4, Fig. 5, Fig. 8 and Fig. 9. It is interesting to see that even in Peak C, which coincides with the main star cluster with its enormous and very hard ionizing flux, there is still some molecular hydrogen seen in that direction.
The detection of weak CO emission shows that CO has survived photodissociation in the whole region. A similar situation has been observed in the giant HII region 30 Doradus in the LMC (see Rubio et al. 1998, Rubio 1999). In this region, H2 knots and filaments have been detected associated to cold molecular gas as seen in CO(2-1) emission, surviving from the strong radiation flux of a dense cluster containing more than 60 O stars, as well the strong winds from these stars. Pak et al. (1998) has found a similar situation in several other regions of the Magellanic Clouds. However the high intensity of the [C II ] 158 µm line in the same regions and in N 66 (Israïl & Maloney 1993) shows that most of the CO has been photodissociated (unfortunately no [C I ] line observation seems to exist for the SMC). Pak et al. (1998) assume that the CO lines are optically thick, and explain the observations of the H2 v=(1-0) S(1), CO and [C II ] lines by the emission of uniform spherical clouds immersed in the UV field. There are many such clouds within the observing beam. Each of these clouds is stratified as explained before, with a CO core and a C II envelope. They adjust the cloud radius in order to obtain relative area coverages of the CO cores and C II envelopes such that the ratio between the average surface brightnesses of an ensemble of clouds in the CO and [C II ] lines matches the observed line ratio. In this model, an increased UV flux yields a decrease of the size of the CO cores and a corresponding decrease of the CO line intensity. However there is no guarantee that the CO lines are optically thick. Even if they are optically thick, one can imagine many variants such as that proposed by Lequeux et al. (1994) in which the medium is inhomogeneous, with the densest clumps emitting the CO lines and the lower-density regions emitting the [C II ] line and a part of the H2 line. Another possibility is the development of pillars and elephant trunks in the parental molecular cloud by the effect of both the UV radiation flux and the strong stellar winds. The existing observations do not permit a clear choice between these possibilities. Consequently, in this paper, we will limit ourselves to qualitative considerations, and defer to a further paper a quantitative modeling.
It is unlikely that the CO lines are optically thin because it is difficult to understand how the column density of CO could be adjusted to the small value needed for optical thinness over the large observed spatial extent. Moreover, limits can be set to the density of the CO-emitting medium from the observed intensity ratio of the CO(2-1) and CO(1-0) lines. The CO(2-1)/CO(1-0) line intensity ratio (in units of ) is 1.3 for the emission in the spur, which is adequately sampled in the observations of Paper I. For the rest of the region around N 66 proper, the CO(2-1) observations are not well sampled but it is still possible to have an idea of the global CO(2-1)/CO(1-0) line intensity ratio by integrating the respective emissions over all the map. We find a slightly higher ratio of 1.4. This ratio depends on the kinetic temperature, on the density and on the optical depth. Examination of the CO excitation diagrams calculated in the large velocity gradient approximation by Castets et al. (1990, Fig. 16 and Fig. 17) shows that in the optically thin case CO(2-1)/CO(1-0) line ratios of 1.3-1.4 require a density larger than 103 molecules cm-3 even for kinetic temperatures as large as 100 K in the emitting region. The densities should be even higher in the optically thick case. In the PDRs close to bright H II regions like the Orion ridge, temperature can rise to 100 K and this is probably also the case for a part of the CO-emitting regions in N 66. Even in this case, the density must be larger than 103 mol. cm-3.
Presumably the medium is clumpy, and we observe the faint average CO brightness of optically-thick hot clumps with a very small surface filling factor. Given the observed Tmb 1 K, the surface filling factor of these clumps is less than 1% instead of 10% found in less extreme situations in the SMC, as for example in the N 66 spur (Rubio et al. 1993; Lequeux et al. 1994). In the large UV field around the H II regions of N 66, the CO-emitting parts of the clumps have shrunk and the smallest clumps have been completely photodissociated. This reduces the average CO line brightness with respect to the "normal" situation (before the appearance of the hot stars) in spite of an increase of temperature. The contrast between the strong CO emission in the NE spur and the weaker emission in the direction of the H II region can easily be explained in this way, the UV radiation field being considerably smaller around the spur.
One can also explain qualitatively in the same way the observations of the [C II ] 158 µm line of Israïl & Maloney (1993). This publication contains a contour map of the integrated [C II ] line flux obtained with the Kuiper Airborne Observatory with an angular resolution of 55" and a sampling of 40". The reference position of this map appears to be the CO peak. One can recognize the general morphology of the region of N 66, with the bar and the spur. An interesting thing is that the emission of the spur is almost as strong as that of the bar. This confirms that the PDRs in the bar have a small surface filling factor (they are just the surfaces of the molecular clumps), while C II is photoionized into C III as soon as it enters into the H II region. On the other hand, the clumps have a bigger surface coverage in the spur and C II can survive. This compensates for the higher excitation of the line near the H II regions.
We now concentrate on the morphology of the N 66 region. Consider Fig. 5 which displays the CO maps at various velocities superimposed over the smoothed LW2 map. Most of the mid-IR or H2 line emission peaks have a CO counterpart, even Peak C which corresponds to the main ionizing star cluster. Each peak has its own radial velocity, with a possible velocity gradient along the bar. There is also a molecular shell around the main PDR, associated to the velocity range 150-155 km s-1, better seen in the total integrated CO emission (140-165 km s-1) at the lower right panel. Clearly the bar molecular structure existed previously to massive star formation and only a small amount of the initial molecular gas has not been ionized by the central cluster. Part of the remaining gas is probably pushed by the stellar winds and supernova explosions to form the molecular shell, as the inner side of which we see the PDR tangentially. It is difficult to know if the remaining molecular gas in the bar is located in front or behind the stars formed recently. According to Massey et al. 1989, extinction is small for their stars towards the core (E(B-V)=0.14 on the average), but the interstellar matter is very clumpy and is not likely to produce much extinction even if it lies in front of the stars. In Paper I, we suggested that the formation of the stars which are presently seen optically in N 66 is not coeval. Peak C contains only unreddened stars according to Massey et al. (1989) but Peaks E, H and I for example contain reddened stars suggesting that the interstellar material has not been spread out as much as in Peak C. We proposed in Paper I that star formation along the bar of N 66 has taken place in a sequential way, starting with stars in the more evolved Peak C. This is in agreement with the model and numerical simulations of Elmegreen et al. (1995). The discovery of embedded stars reported in Sect. 3 sheds a new light on this picture. These stars are located in Peaks B, D, E, F, H and I, representing a very recent stellar generation, and there is also an embedded star seen towards Peak C. Whether this star lies within the main stellar cluster NGC 346 or is located behind or in front is not known. It could be that the stars in the core have blown up a cavity in the parental molecular cloud and we see the borders of this cavity delineated by the H2 emission. Current star formation is thus taking place in the interface region between the ionized and the molecular gas. There is also an embedded massive star (or group of massive stars) at the northern tip of the spur, with associated AIB and H2 line emission. It indicates a new site of star formation. Curiously, Peak I while quite strong in CO and AIBs is relatively weak in the H2 line and is not visible in the [C II ] line map of Israïl & Maloney (1993). Presumably the far-UV flux is still weak in this region.
© European Southern Observatory (ESO) 2000
Online publication: July 13, 2000