1 Introduction

  Over the last two decades, there have been great efforts to identify what might be called the "missing link'' of astronomy: the protostar. Even the most optimistic agree that this work has given questionable results. The explanation may be either the brevity of the protostellar phase or the lack of an effective tool to sample the high density, optically thick, collapsing regions where stars form. Whatever the explanation, from an observational point of view, it is clear that the first step is to define the criteria which identify embedded young stars in their very earliest evolutionary stages.

The material surrounding such embedded young stars, presumably related to the dense condensation existing prior to star formation, is confined to a very small region, of the order of a few seconds of arc (0.1 pc typically), and its presence may be overshadowed by larger structures. The precise problem depends on the observing technique: (i) in the radio continuum by the extended HII regions generated by older generation massive stars; (ii) in molecular lines by larger scale low density (but optically thick) molecular clouds; and finally (iii) in the near and far IR by radiation of more luminous nearby stars in the same complex. Hence, high resolution and selective molecular tracers are needed to distinguish the emission from the environment of the youngest stars from that emanating from the surrounding medium.

We believe that H₂O masers (of star forming type, to distinguish them from those associated with late type stars) represent an excellent tracer of the earliest stages of star formation. The advantage of using H₂O masers as tracers is twofold: firstly their positions are known with very high accuracy ($\sim$0.1$^{\prime\prime}$) and hence one can limit the region to be studied to a small field around the masers. Secondly, several observations suggest that H₂O masers are closely related to the earliest high density phases of the formation of massive stars, prior to the development of an ultracompact (UC) H II region (Tofani et al. 1995 hereafter TFTH; Hofner & Churchwell 1996; Cesaroni et al. 1997a; Cesaroni et al. 1997b; Testi et al. 1998; Plume et al. 1992; Plume et al. 1997; Codella et al. 1997).

H₂O masers of star forming type have been found in the surroundings of HII regions (Wood & Churchwell 1989a; Churchwell et al. 1990). However, it is becoming more and more evident that they are not associated with the diffuse ionised gas of evolved HII regions, but with dense molecular clumps present in the same area, which in some cases contain embedded UC HII regions (e.g. Cesaroni et al. 1991). Recently, observations with high spatial resolution ($\ll 1\hbox{$^{\prime\prime}$}$) have shown that there are several cases where no UC HII region is associated with the H₂O masers (Hunter et al. 1994; TFTH). Considering that an H₂O maser requires a neighbouring stellar source in order to excite it (Elitzur et al. 1989), the lack of a compact radio continuum source suggests that the embedded massive star is not capable of exciting an observable UC HII region. This might be because the UC HII region is so optically thick and highly self-absorbed to be undetectable at radio wavelengths and thus very dense and very young.

The complex scenario that is emerging from more detailed studies of selected star forming regions suggests that different and probably independent episodes of star formation, each with its own HII region, may have occurred in the same complex and that H₂O masers are associated with the most recent episodes (Hunter et al. 1994; Felli et al. 1997; Schreyer et al. 1997).

At the same time, in several objects, the H₂O maser spots show an excellent positional agreement with molecular clumps revealed in several transitions and with far IR or sub-mm continuum peaks (i.e. dust), but not with radio continuum peaks. This situation is for instance seen in W3(OH) (Turner & Welch 1984; Wink et al. 1994; Wyrowski et al. 1997) where the molecular emission comes from the position of the H₂O maser, which is offset by $\sim 6\hbox{$^{\prime\prime}$}$ (0.06 pc) from a strong UC HII region. In Orion itself, the H₂O masers are not associated with the diffuse gas ionised by the Trapezium Cluster, but with the molecular peak in the BN/KL region and with a very weak unresolved radio source (component I: Churchwell et al. 1987; Felli et al. 1993; Gaume et al. 1998).

VLA observations in the (4, 4) inversion transition of ammonia towards other UC HII regions (Cesaroni et al. 1994) lead to similar conclusions: both H₂O masers and NH₃(4, 4) emission arise from the same position, well apart from continuum peaks. Cesaroni et al. (1994) conclude that the H₂O - NH₃ clumps are likely to be the site of massive star formation. Methyl cyanide seems to trace the same high density clumps (Olmi et al. 1993, 1996, 1996b) and one concludes that these are examples of the "hot core'' phenomenon, where the presence of a nearby luminous star causes the evaporation of dust grain ice mantles in the surrounding medium and consequently enhances the abundance of several molecular species (see Millar 1997 and Ohishi 1997).

  

Table 1: Coordinates of the field centres and fluxes of the associated IRAS sources

^a Maser component with higher peak flux from TFTH.
^b Coordinates of the maser peak from TFTH rounded to 1$^{\prime\prime}$.
^c Fluxes in Jy in the four IRAS bands (12, 25, 60 and 100 $\mu$m).

With this in mind, we have selected a sample of 12 H₂O maser sources that radio continuum observations with high spatial resolution and sensitivity (TFTH) have shown to be well separated ($\gt 10\hbox{$^{\prime\prime}$}$) from the closest HII region. A few of these objects have already been observed in some molecular lines (see Krügel et al. 1987; Serabyn et al. 1993) and most of them have been detected in the CS(7-6) and CO(3-2) transitions (Plume et al. 1992, 1997). Our goal is that of obtaining a full picture with high single dish resolution ($10\hbox{$^{\prime\prime}$}-30\hbox{$^{\prime\prime}$}$) of the molecular environment of the water masers. For this purpose, we have used IRAM 30-m observations of the C³⁴S(2-1), (3-2) and (5-4) transitions to estimate the size and density of the associated molecular clump (cf. Cesaroni et al. 1991). We have also observed the HCO⁺(1-0) and HCN(1-0) lines with the aim of studying the dynamics and ionisation degree of the molecular gas. Note that the HCO⁺(1-0) line may be a good tracer for infall (see e.g. Welch et al. 1987 and Rudolph et al. 1993), as well as of outflow (Cesaroni et al. 1997b). Observations of the CH₃OH(3-2) and (5-4) transitions and CH₃CN (8-7) and (12-11) transitions were also carried out with the aim of detecting the "hot core'' if present, as well as to determine the clump density and temperature. Methyl cyanide and methanol have advantages from this point of view as one can observe several transitions of differing excitation within one bandwidth. The relative intensities of these lines are thus essentially independent of calibration errors. Moreover, we have available LVG statistical equilibrium codes which permit us to interpret the results (see e.g. Bachiller et al. 1998; Walmsley 1987; Olmi et al. 1993) Finally, ¹³CO(2-1) and CS(3-2) were also mapped, in order to obtain information on the larger scale molecular gas distribution.

Our survey does indeed show that all our targets are associated with molecular clumps centred on the H₂O maser spots. In this respect, our results are analogous to those of Plume et al. (1992, 1997), who searched for high density molecular tracers in several transitions of CS and C³⁴S towards a large number of H₂O masers. The essential difference is that we have carried out a more detailed mapping of a restricted number of objects, whereas Plume et al. merely observed towards the water maser position.

Some of the results presented here have been covered in earlier publications but, for completeness, here we present the whole sample observed with the 30-m here. Partial results for S235A-B were reported in Felli et al. (1997). A detailed analysis of the survey observations for IRAS20126+4104 as well as comparison with Plateau de Bure interferometric observations and near IR observations were presented by Cesaroni et al. (1997a). These two objects are fairly typical of the complete sample.

In Sect. 2 of this article, we summarise the properties of the sample and give the water maser coordinates from TFTH. In Sect. 3, we explain the procedures used during the observations and data reduction. In Sect. 4, we present the data giving both line parameters towards the central positions and maps. We also discuss the properties of individual sources in Sect. 5. Finally, in Sect. 6, we summarise our conclusions.