Astron. Astrophys. 358, 257-275 (2000)

4. Discussion

A lot of SO/CS observations have been presented as well as some exploratory chemical modelling. Large variations in the SO/CS abundance ratio can be explained by differences in "chemical age", density, initial oxygen abundance, as well as varying ionization levels caused by cosmic ray flux or X-ray sources.

4.1. The observed variations of the SO/CS abundance ratio in the light of chemical modelling

As evident from our ratio maps (Figs. 1a-1v) and the more accurate determinations from ³⁴SO and C³⁴S observations (Table 3) the SO/CS abundance ratio exhibits pronounced variations, well outside the relevant error bars, between the nineteen cloud cores of our sample as well as within some of the clouds. This is indeed neither unexpected nor astonishing in the light of the exploratory chemical modelling presented in Sect. 3.

We propose that the main cause of the variation of the SO/CS abundance ratio between clouds is the very high sensitivity to a variation of the initial O/C⁺ elemental abundance ratio (see Fig. 6). Relatively small variations of the elemental abundances/depletions of oxygen and carbon among the clouds would be required. Some influences of variations of the cosmic ray and/or X-ray flux may also be present, since the propagation of cosmic ray particles would be sensitive to e.g. variations of the Galactic magnetic field, while the X-rays originate in young stellar objects. Likewise, some influences of cloud age and more or less rapid chemical evolution due to somewhat different average cloud densities cannot be excluded. Moreover, variations of the UV light flux as well as lower visual extinction (than the 25 mag assumed in the modelling) may be important in lower density clouds (cf. Nilsson 1999). Since the abundances of SO and CS are almost linearly dependent upon the available amount of free sulphur, the SO/CS abundance ratio is essentially insensitive to the initial S abundance (as is apparent from Fig. 12). We will, however, return to all these questions in a subsequent Paper III (Olofsson et al., in prep.), where the very abundances of SO and CS, as well as the H₂ cloud and column densities, will be mapped across the clouds in the present sample.

We now turn to a general discussion of the observed large variations of the SO/CS abundance ratios within some clouds and the apparent lack of such variations within other clouds. It is again obvious from the chemical modelling results (Fig. 6) that relatively small variations of the initial oxygen and carbon abundances could cause large changes of the SO/CS ratio, but such local variations of the gas-phase elemental abundances may be less likely. They could though originate from different local depletions onto dust grains. Instead we propose that where large local variations of [SO]/[CS] are observed this is more likely a consequence of the combined effects of density changes (Fig. 10) and local variations of the cosmic ray and X-ray ionization levels (Fig. 11). In all these cases the local chemical evolution is accelerated, resulting in a change of the time needed to reach a chemical steady state and a large SO abundance. In this respect we may say that the SO/CS abundance ratio is a diagnostic of the "cloud age". We here also should note that an increasing local cosmic ray or X-ray ionization level leads to a somewhat reduced late time SO/CS abundance ratio, while, on the contrary, a density increase raises this ratio (cf. Figs. 11 and 10). Furthermore, since the propagation of (charged) cosmic ray particles would be sensitive to magnetic fields, we may indeed expect some local variations of the ionization in denser regions, where the magnetic field strength also is expected to increase due to "flux freezing". Large local increases of the X-ray ionization would be expected near YSO's, as has been so clearly demonstrated for the Oph molecular cloud by Casanova et al. (1995).

In this rather pronounced framework of expectations the apparent lack of SO/CS variations in many sources may be more surprising than our discovery of large variations within a few sources. However, to some extent this may be an effect of spatial resolution, since large variations have been seen only in nearby clouds (cf. Table 5).

Table 5. O₂ late time abundance predictions, maximum values, resulting from [SO]/[CS] maximum.
Notes:
a) The adopted ratios of Table 3, if not otherwise stated
b) Initial [O]/[C⁺] required to get the observed [SO]/[CS] at late times for the highest observed [SO]/[CS], cf. Figs. 5 and 22, for
c) Maximum value at late times, for the [O]/[C⁺] value in previous column, [O₂]/[H₂] is 10^-4 times lower
d) maximum value using SO()/(C³⁴S(2-1))
e) maximum value from map using SO()/CS(2-1)
f) possible influence of unresolved outflow, see text
g) Using the SO data of Gottlieb et al. (1978), and CS data from Liseau et al. (1995), and unpublished SEST data
h) N(SO)/(22N(C³⁴S)) for the NW position of Nummelin et al. (2000)
i) N(SO)/(22N(C³⁴S)) for = (0´,0´) and (), Pratap et al. (1997)
j) Swade (1989a, 1989b)
k) Bergin et al. (1997)

We have until now avoided a comparison of the chemical evolution time scale with other relevant time scales. One such time scale is the free-fall time - the characteristic collapse time of a pressure free gravitationally unstable gas core - which has been estimated to be yr, where n is the H₂ volume density (particles per cm³) (Spitzer 1978). We note that the free-fall time scale is of the same order of magnitude as the time needed to reach late time chemical equilibrium. Another time scale of concern in connection with chemical modelling is that relevant for molecular depletion onto cold dust grains. This molecular sticking time scale has been estimated to be yr, where m is the mass (in amu) of the molecule in question, is the sticking efficiency and n is the H₂ volume density (cf. Williams 1993; Dzegilenko & Herbst 1995). The sticking efficiency is believed to fall in the range , so that for SO and CS the "sticking time" becomes yr, which is also of the same order of magnitude as the time required to approach ("late time") chemical equlibrium and hence could strongly influence the late time chemical abundances. However, since interstellar molecules are observed at abundance levels close to those predicted by pure gas phase models (without sticking), it appears that efficient desorption mechanisms must also exist. But their efficiency is rather uncertain, as are the sticking efficiencies of the various molecules (cf. Williams 1993; Bergin et al. 1995; Dzegilenko & Herbst 1995).

In our astrochemistry calculations we have not included molecular depletion onto/desorption from grain surfaces. In their extensive chemical modelling efforts Bergin et al. (1995) also investigated such processes. They suggest that if the interstellar grains have an outer layer of CO ice then the binding energies to the grain mantle may be considerably lower than commonly assumed for many species, and a significant amount of the molecules will remain in gas phase.

Therefore we have to stress that our exploratory chemical modelling (assuming no net adsorption onto grain surfaces) is indeed just a first step in considering what may cause variations in the SO/CS abundance ratio.

4.1.1. NGC 1333

The NGC 1333 molecular cloud, at a distance of only 350 pc, is the most nearby source in our sample. It is also unique in that the SO/CS abundance ratio rises from very low values near the driving source SVS 13 of a prominent outflow (Liseau et al. 1988) to very high values along an extended north-south elongated ridge at the end of the outflow. We have here mapped the variations of the SO/CS abundance ratio by means of the optically thin ³⁴SO() and C³⁴S(2-1) transitions. The SO/CS abundance ratio is estimated to be as low as 0.5 near SVS 13, which is the (0,0) offset position in our maps (Figs. 1d-1e). In the north-south elongated ridge region the SO/CS abundance ratio falls in the range 2-7, with the highest ratio near the far infrared source IRAS 4 which is located about from SVS 13. It is instructive to point out, when comparing to more distant sources, that if NGC 1333 would have been located at a distance further away than about 2-3 kpc it had not been possible to clearly discern the large variations with our beam.

The same elongated ridge of gas is apparent in the ¹³CO and C¹⁸O 1-0 and 2-1 observations by Warin et al. (1996). These authors argue that this is a compressed "shell" of gas formed as the winds/outflows from SVS 13 (dynamical age yr) and several other nearby sources have swept up the material in the parent molecular cloud. The cloud density in this north-south elongated ridge was estimated to vary from to cm^-3 and the H₂ column density is cm^-2, corresponding to a visual extinction , from the cloud core to the surface, of 11-16 mag (Bohlin et al. 1978). From their follow-up CS(2-1, 3-2, 5-4) mapping study of the NGC 1333 region Langer et al. (1996) derive an average density of cm^-3 in the compressed ridge. Warin et al. (1996) identified several massive ( 20-40 ) condensations in the compressed shell, which were proposed to be potential sites for the formation of the next generation of stars. While the average temperature in the region is below 20 K, the temperature rises to 33-42 K in the cluster of 18-46 IRAS sources which is contained in the molecular cloud ridge (Jennings et al. 1987).

The extensive molecular line observations of one of these sources, IRAS 4, by Blake et al. (1995), have revealed a very dense ( cm^-3) core at a kinetic temperature in the range 20-40 K. A young (dynamical age of a few thousand years) outflow also originates in this region.

This information about the physical conditions in the NGC 1333 core regions may clarify the large SO/CS abundance ratio variations across this region. According to our chemical modelling an initial O/C⁺ abundance ratio of 1.8 to 2.6 is required to achieve SO/CS abundance ratios as high as 2-7, at late time chemical equilibrium. In the dense cores present in the compressed ridge this chemical equilibrium would appear much earlier in time than in the region near SVS 13, which has been evacuated by the outflowing gas according to Warin et al. (1996). In the less dense regions the chemistry may not yet have reached a late time equilibrium state and hence the SO/CS abundance ratio still remains below 1. In this scenario we have only used the accelerated chemistry and larger SO/CS abundance ratios appearing at increasing densities (Fig. 10). Higher ionization levels, as expected near YSO's, would speed up the chemical evolution, but also could decrease the late time SO/CS abundance ratio (Fig. 11). There is in fact X-ray emission associated with the infrared source SVS 16 (Preibisch et al. 1998), in our map close to the "edge" of the dense region. As discussed in Sect. 4.1 there are also other processes which could modify the chemical evolution such as increased ionization by the UV radiation from newly born stars.

4.1.2. Orion A

This well-studied molecular cloud exhibits the lowest SO/CS abundance ratio in our sample, only 0.1-0.5. We could propose that the Orion A region is much younger (less chemically evolved) than most molecular clouds in our sample, but this seems impossible in view of the rich chemistry of the Orion A core regions (cf. Irvine et al. 1987; Sutton et al. 1995). The very low SO/CS abundance ratio appears to be explainable if the initial O/C⁺ abundance ratio is low (; i.e. C/O ) i.e. a comparatively C rich environment. This is indeed the current case. The CI abundance in Orion A has been observed to be large ([C]/[CO] : Phillips & Huggins 1981; Tauber et al. 1995; White & Sandell 1995). However, this is not necessarily a confirmation of the high initial C abundance. A high C concentration "today" probably requires a two-phase cloud model with denser clumps embedded in an interclump medium. The remaining free carbon then would mainly reside in the interclump medium, which is more easily penetrated by ionizing UV radiation, while the more evolved chemistry would dominate in the dark, denser cores.

4.2. The SO enhancement in the outflows of Orion A, NGC 2071 and W 49N

Although outflows are present in many of our observed sources, the kinematical evidence thereof is rarely apparent in our SO and CS spectra (at the current sensitivity level). However, e.g. Chernin et al. (1994) and Lefloch et al. (1998) have demonstrated that the outflow contribution dominates in higher excitation lines of SO and CS. The SO/CS abundance ratios are strongly enhanced in the Orion A and NGC 2071 outflows where we estimate the [SO]/[CS] ratios to be about 24 and 2.2, respectively. This should be compared to the ambient cloud abundance ratios of 0.2 for both sources. The very strong SO abundance enhancement in the Orion A outflow was studied in some detail already by Friberg (1984) (cf. Irvine et al. 1987; Sutton et al. 1995). A similarly strong SO abundance increase recently has been observed in the Sgr B2(M) and (N) cores by Nummelin et al. (2000). Such high SO abundance enhancements would be expected in molecular shocks (e.g. Mitchell 1984; Pineau des Forêts et al. 1993). The rather high SO/CS abundance ratio in the W 49N core (about 2.2) also may be due in part to the existing massive molecular outflow source (e.g. Downes et al. 1982; Scoville et al. 1986), although this very distant outflow would be unresolved in our 40" beam and no clear kinematical evidence is apparent in our spectra (see Paper I).

4.3. Predictions of O₂/CO abundance ratios

This question has been adressed in some detail in our chemical modelling, presented in Sect. 3. The initial O/C⁺ abundance ratio required to produce the observed SO/CS abundance ratio at steady state may also be used to predict the O₂/CO abundance ratios for each source. Such model predictions (for the standard cloud parameters given in Sect. 3) are presented in Fig. 22 for H₂ densities of , , and . We note that the O₂/CO abundance ratio is essentially independent of the cloud density (for a fixed O/C⁺ ratio) while the SO/CS ratio increases markedly with increasing density. This steady state behaviour can also be seen in Fig. 10 and Fig. 13. In Table 5 we tabulate the range of observed SO/CS abundance ratios in our sources (see Table 3) together with recently published results for Sgr B2,  Oph, TMC-1, L134N, and M17 (the references are given in the table). For a few sources, with limited information, only a single SO/CS abundance ratio is given. Since our SO and CS observations mainly probe higher density regions (see Sect. 2.3) we use the model in Fig. 22 together with the highest observed SO/CS abundance ratio to estimate the O/C⁺ abundance ratio initially required in the various clouds (listed in the fourth column of Table 5). We assume that the initial O/C⁺ abundance ratio was the same within a cloud (cf. Sect. 4.1). Finally, in the last column of Table 5, the predicted steady state O₂/CO abundance ratio is given. Although the tabulated O₂/CO abundance ratio was calculated using it is also approximately applicable for densities as low as (when the initial [O]/[C⁺] ratio is known, cf. Fig. 22) if (and only if) the steady state chemistry is reached. However, regions with such low densities are not probed very well by our SO and CS observations.

Fig. 22. O2/CO abundance ratio vs SO/CS at late times for three different H2 densities. The initial O/C+ ratio varies between 1.0 and 4.0. Note that the O2/CO ratio is insensitive to variations in H2 density for O/C+ . The diagram may be used to predict the O2/CO abundance ratio from the observed SO/CS abundance ratio

The estimated initial O/C⁺ abundance ratios vary from 1.3 (M17) to 2.5 (NGC1333, NGC2071, and L134N). The variation of the predicted O₂/CO abundance ratio is larger: 0.05-0.4. The sources in Table 5 with high [O]/[C⁺] ratios, and hence high [O₂]/[CO] ratios, should be the most promising candidates for O₂ detection.

4.4. On the detectability of O₂ by the SWAS and Odin satellites

The best limits on the O₂/CO abundance ratio, after 100 hours of SWAS O₂() integration at 487 GHz in several molecular clouds, are reported to be below 0.005 (Goldsmith et al. 1999) which is almost two orders of magnitude lower than our highest predicted ratio (see Table 5 and previous section). Although low beam filling due to the large SWAS antenna beam (HPBW of ) may provide a partial explanation, the very low abundance limits observed seem to require an enhanced carbon abundance in the cloud interiors causing efficient destruction of O₂ [cf. Eq (2)]. Turbulent mixing between the outer diffuse regions and the cloud cores has been investigated as a possible cause of C enrichment (Chièze & Pineau des Forêts 1989; Xie et al. 1995). However, such an enrichment simultaneously would reduce the SO abundance, contrary to our observational results (for the high density gas). This failure to simultaneously accomodate a high SO/CS abundance ratio and a low O₂ abundance appears to indicate a fundamental problem in current chemical models.

To increase the search sensitivity the Odin satellite has been equipped with a low noise HEMT preamplifier for the O₂ transition at 119 GHz (HPBW of ). Based on the receiver noise temperatures and O₂ excitation calculations (Bergman 1995; Maréchal et al. 1997b) we estimate, using the appropriate beam sizes, that the Odin sensitivity at 119 GHz will be about an order of magnitude higher than the SWAS sensitivity at 487 GHz. This is valid for small and dense clouds (size SWAS beam) as well as for more extended (size SWAS beam) and less dense regions.

© European Southern Observatory (ESO) 2000

Online publication: June 26, 2000
helpdesk.link@springer.de