The comparison of the observed intensity of the H2 1-0S(1) line in the central torus of NGC 2346 to the NH98 model predictions shows that the emission can be produced in the hot PDR generated by the radiation of the central star, once the effect of X-ray heating and time-dependent (advecting) chemistry is taken into account. The best fit is obtained by models with relatively low density of the neutral gas (in agreement with the low density of the ionized material inferred by several authors). However, these PDR models require the luminosity of the hot central star to be significantly higher (L250 L) than current estimates.
If the H2 lines are emitted in the PDR, we expect to observe a similar morphology in the ionized and H2 emitting gas. In NGC 2346, the PDR origin of the H2 lines is supported by the fact that the same morphology is seen in H2 and in H (see Walsh 1983). Also, we detect Br emission in the two H2 peaks, with a N-S profile that follows that of the 1-0S(1) line (Fig. 3). The intensity of Br predicted by the models is very low, of the order of erg cm-2 s-1 sr-1, comparable to the observed values (4-6 erg cm-2 s-1 sr-1). This supports our estimate of L.
In principle, the observed intensity of the H2 1-0S(1) line can also be accounted for by the emission of the shocked gas produced by the expansion of the torus inside a precursor red-giant wind. However, we estimate (following NH98) that one needs a rather high value of the mass-loss rate in the red-giant wind ( M yr-1). This, in turn, implies a high density of the pre-shock gas ( cm-3), which is not supported by any existing observation. In fact, as discussed by Zuckerman and Gatley (1988), the main difficulty in ascribing the H2 vibrationally excited emission to shocks comes from the high momentum rate these models require. Many authors (see the review by Kwok 1993) have proposed that the formation and expansion of the PN shell (or torus) is related to the action of the fast wind from the central star. In this case, the ambient gas (pre-shock red giant wind) gains momentum approximately at the rate at which momentum is delivered to the torus by the fast wind. Since the fast wind is radiation driven, this rate () must be L/c. In NGC 2346, we estimate that is at least erg cm-1 s- 2, i.e., more than 600 times the present value of L/c (for L=250 L) and 12 times higher than the maximum L/c reached by the star in its earlier evolution, according to the evolutionary tracks of Blöcker (1995).
The interpretation of the H2 emission in terms of shocks is often justified in the literature by the low measured ratio of the 2-1S(1) to the 1-0S(1) intensity. This argument, however, is not very strong, since in dense PDRs the low vibrational H2 levels are thermalized. The PDR models discussed in Sect. 4.1 predict a ratio 2-1S(1)/1-0S(1)0.15 (not very different from the 0.10-0.18 range predicted by shock models; see Sect. 4.2), independently of the density and stellar luminosity. These values are somewhat higher than the observed ratio (0.07). It is possible, and worth further investigations, that models tend to overestimate the fluorescent component of high vibrational lines, possibly because of uncertainties in the collisional deexcitation rates.
An interesting result of our observations, and one that we cannot account for with our simple models, is the variation of the 2-1S(1)/1-0S(1) ratio with position along the slits, ranging from about 0.08 to 0.15 along the W slit, and from 0.08 to 0.23 along the E slit (see Fig. 3). These variations are not monotonic with the distance from the peaks, but show evidence of structures, especially along the W slit. It is possible that this is due to density variations, which affect the fluorescent contribution to the 2-1S(1) line. If the emission is due to shocks, this could trace variations in the propagation velocity of the shock in an inhomogeneous medium.
Further support to the PDR origin of the H2 emission can be obtained by observing lines from higher vibrational states. We show in Table 3 PDR model-predicted values for lines not detected so far in NGC 2346, which, however, are accessible from space. The lines are very weak with respect to the 1-0S(1), with ratios that do not depend significantly on the density. We expect that they will be substantially weaker in shocks. To complete the discussion of the H2 spectrum, we show in Table 4 model-predicted values of the intensity of mid-infrared lines in the v=0-0 band for four of the PDR models described in Sect. 4.1. These lines have been observed by ISO in a number of PNe (among them NGC 2346; Barlow et al. in preparation), and may be useful diagnostic of the physical conditions (see NH98).
Table 3. High-v H2 PDR predicted line ratios
Table 4. Mid-Infrared H2 PDR predicted line intensities
© European Southern Observatory (ESO) 1999
Online publication: February 23, 1999