1000 cm-3)
freeze-out of gas phase species on to grains must take place in some
degree. For each scenario, we find at least one model able to
reproduce the majority of the observations; however, this is true
only for a limited period of time, implying the clump may be
transient. Its lifetime is model dependent. This result can be
explained with the aid of a "cyclic model'' for molecular clouds where
star formation is occurring: some regions within the molecular cloud
collapse into dense small clumps and form stars; once the star is
born, the clumps are easily destroyed due to a total gas depletion
and/or external activities such as shocks and outflows. The material
they return to the ISM is then re-processed to form new clumps. Star-less
cores such as TMC-1 however might not undergo a collapse
phase. In this case, for the cyclic model to be valid, depletion must
be the main cause of the loss of H2O and O2.
The period for which the H2O and O2 abundances determined by SWAS and the molecules observed by other telescopes, are matched reasonably well by the models, is - in general - rather brief. The consequences of this brevity on the cycle-time required in molecular clouds would be severe. However, this period can be considerably extended if it is assumed that CO and N2 are not retained on grain surfaces during the freeze-out process, but are, in fact, returned promptly to the gas where they initiate a late-phase chemistry. This chemistry persists for an appreciable fraction of the likely age of the molecular clouds; it also reflects a C:O ratio of close to unity. A similar approach has been suggested by Bergin et al. (2000) where CO, N2 and CH4 are allowed to evaporate because the temperature of the dust grains is maintained at 30 K. They, too, find that this approach reproduces the low abundances of H2O and O2. The main difference in our approach is that, by retaining CH4 on the grains, we do not increase the abundances of hydrocarbons and cyanopolyynes in the gas phase (Nejad & Millar 1988). While no attempt can be made to produce a precise fit for any particular object, given the nature of the SWAS observations, it is clear that the overall results of the SWAS mission can be interpreted in terms of time-dependent chemistry, and that this interpretation is strengthened if selective freeze-out occurs.
We have also shown that the range of physical conditions considered is appropriate to the occurrence of bistable steady state solutions. The so-called HIP phase offers a natural explanation for low abundances of O2 and H2O, even at steady state. Also, the intrinsic uncertainties in the chemical network, even for the simple molecular species discussed here, are compatible with a bistable behaviour. The possibility that the two chemical phases contribute to the observed column density is most likely and a combination of different chemical phases still may explain both the large abundance of water towards Sgr B2 and the ubiquitous low abundance of molecular oxygen. The upper limit found for gas phase H2O in starless cores is much more striking. However, the SWAS results refer only to the ortho form and a definitive conclusion should wait for a comparable study of the para form of water. This is still for the future.
Acknowledgements
We thank the referee for constructive comments that helped to improve on an earlier version of this paper. SV and DAW thanks PPARC for financial support. ER, DAW and GPdF acknowledge the TMR "Astrophysical Chemistry" #ERBFMRXCT970132.
Copyright ESO 2001