4 Variations between target regions

In general, the most striking regional variation is between the sources in Cha I and those in the other three regions. All of the sources in Cha I, except for ISO-ChaI 192 (Cha I 2), show little or no absorption due to volatile ices and show silicates either in emission or an emission/absorption composite. As it lay outside the CVF1 images, due to rotation of the spacecraft between the two sets of observations, we only have a 9.3-16.3 $\mu$m spectrum for ISO-ChaI 192. However it shows CO₂ ice absorption and also appears to show deep silicate absorption ( $\tau_{\rm Si} \sim 2$-4), so it is probably an embedded object. Persi et al. (1999) found that ISO-ChaI 192 lies in a dense core and that it is probably the engine for the bipolar outflow detected by Mattila et al. (1989). Jones et al. (1985) found that several other YSOs have formed around the edge of this core, rather than at its centre: these were apparently formed during a burst of star formation triggered by a wind from the nearby star HD97300. Our results are consistent with this: ISO-ChaI 192 appears to be heavily embedded, and has a very red SED. The rest of the surrounding objects show similar spectral characteristics, and appear to be surrounded by far less circumstellar/foreground material than ISO-ChaI 192.

12 of the 20 objects in Serpens show deep absorption features, 7 show composite silicate profiles, and one (SVS2) shows silicate emission. It is clear that Serpens contains a wide range of YSOs, ranging from the most heavily embedded object observed (Ser B 11) to a source showing silicate emission. By contrast almost all of the sources in both RCrA and $\rho$ Oph show deep absorption features, with only 1 of the 7 sources in RCrA and only 2 of the 10 sources in $\rho$ Oph showing composite silicate profiles. Given that YSOs are generally expected to "sweep out'' their circumstellar material as they evolve (Shu et al. 1987), this seems to imply that the observed star formation is at a similar evolutionary stage in both RCrA and $\rho$ Oph, probably an earlier stage than that in Serpens. However, given the broad variation in the sources observed in Serpens, this conclusion remains somewhat tentative. It should also be noted that the observed fields (of $2\hbox{$^\prime$ }\times2\hbox{$^\prime$ }$) are much smaller than the star-forming clouds, which typically extend over several square degrees of the sky, so the populations observed here may not be representative of the entire clouds.

  
4.1 Silicate depth and H₂O bending mode

Figure 8: The strength of the bending mode of water ice plotted against the silicate optical depth. Different regions are distinguished by different symbols, as shown in the legend.

Figure 8 shows the measured equivalent widths of the 6 $\mu$m water ice feature plotted against the optical depth of the silicate feature: clear regional variations are apparent. As discussed above, all but one of the sources in Cha I are characterised by little or no ice absorption and by silicate emission; the other 3 regions show absorption from both silicates and ices with only a few examples of weak silicate emission. The sources in RCrA are characterised by similar absorption strengths throughout, whereas far deeper absorption and variation is observed in both $\rho$ Oph and Serpens. It is clear from Fig. 8, however, that the sources in $\rho$ Oph show markedly less water ice absorption relative to silicates than those in Serpens. This could be due to sublimation effects, indicating a difference in the thermal structure and history of the two clouds, or could represent a real difference in the chemical composition of the two clouds; without measuring the corresponding gas-phase lines we cannot say which. Sub-millimetre observations of gas-phase H₂O (Ashby et al. 2000) do show an under-abundance of gaseous H₂O in $\rho$ Oph A relative to S140, but to the best of our knowledge no similar direct comparisons between the sources observed here exist in the literature. It is also worth noting that the silicate:water ice ratios appear to be approximately constant for all the different embedded sources in both $\rho$ Oph and Serpens. This indicates that in both clouds the composition of the absorbing intracloud material is roughly constant along different lines of sight, a result consistent with the previous Serpens work of Eiroa & Hodapp (1989).

  
4.2 H₂O and CO₂ bending modes

Figure 9: The ratio of the CO2:H2O ice equivalent widths plotted against silicate optical depth. This ratio is directly proportional to the ratio of ice column densities: adopting the band strengths from Gerakines et al. (1995) results in an equivalent width ratio of 1 corresponding to a column density ratio of $\simeq$0.17.

Figure 9 shows the ratio of the CO₂:H₂O ice equivalent widths (which is proportional to the corresponding ratio of column densities) plotted against the silicate optical depth. The results from Serpens and RCrA are fairly well correlated, with CO₂:H₂O ice column density ratios ranging from 0 to 0.16 (adopting the band strengths from Gerakines et al. 1995). There is a possible trend showing the CO₂:H₂O increasing with decreasing silicate optical depth, but this is somewhat unclear in these data.

The data from $\rho$ Oph, however, show a far greater scatter with CO₂:H₂O ice column density ratios ranging from 0 to 0.4. This is consistent with previous observations: for example Chiar et al. (1994, 1995) found a greater scatter in column densities of CO and H₂O ices against A_V in $\rho$ Oph than in Serpens, RCrA or Taurus. The reasons for this are unclear but it seems likely that, as suggested by Chiar et al. (1995), local conditions around the young stars in $\rho$ Oph play a role. Detailed study of the gas-phase abundances of CO₂ and H₂O towards these objects will further constrain this problem, but without such observations or further knowledge of the envelope structures we merely note the presence of this discrepancy here.