Please Note: the e-mail address(es) and any external links in this paper were correct when it was written in 1995, but may no longer be valid.
Dipartimento di Astronomia e Scienza dello Spazio,
Università degli Studi di Firenze, Largo E. Fermi 5, I-50125, Firenze,
Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, I-50125, Firenze, ITALY
It is well established that 22.2 GHz water masers are associated with star forming regions (SFRs). masers are considered good tracers of optically obscured molecular cores in which a star forming process is occurring. Particularly, maser emission has been detected where other phenomena typical of high-mass star formation are observed: ultracompact (UC) HII regions, molecular outflows, far-infrared (FIR) sources. This indicates that water masers are primarily associated with high-mass star forming regions.
Wood & Churchwell (1989) (hereafter WC) have proposed criteria to identify UC HII regions based on FIR flux density distribution (and hence the colours) of sources extracted from the IRAS Point Source Catalogue (IRAS (1985)). More recently, several works have pointed out that the population of IRAS PS selected through the WC criteria may contain not only massive O-type stars, but also lower mass protostars (White et al. (1991),Hughes & MacLeod (1994),Kurtz et al. (1994)). The possibility of answering this question comes from the investigation of the occurrence of water masers in the WC sample. Palla et al. (1991) and Palla et al. (1993) (hereafter P1 and P2 respectively) have performed a survey of masers towards a sample of IRAS PS which satisfy both the WC and the Richards et al. (1987) criteria, which select compact molecular clouds with high densities (). P1 and P2 found that the IRAS PS with 60 micron flux density greater than 100 Jy are UC HII region candidates, while fainter sources may be connected with lower mass B-type stars. The aim of this work is to verify if the maser properties of the IRAS PS located inside the WC colour box have any dependence on the flux density distribution. In order to do this, we have performed a survey of maser emission towards the IRAS PS located inside the WC box, but which do not satisfy the Richards criteria. The survey has been carried out with the 32-m Medicina (Bologna, Italy) radiotelescope, used also by P1 and P2. Therefore, our results are directly comparable with the conclusions of Palla and his collaborators.
Figure 1: colour distribution of the IRAS PS observable at Medicina which satisfy the WC criteria (continuous line). The crosses represent the sources with , while the open squares are for sources with . The dashed line bounds the region given by the Richards criterion on colour, while the dot-dashed line stands for .
In order to choose the sources for our sample, as a first step all the IRAS PS (1846 sources) which satisfy the WC colour criteria were selected. These criteria are: (i) (using the notation that ) and (ii) . The IRAS PS without a detection at 25 and/or 60 micron bands have been rejected. On the other hand, the IRAS PS with an upper limit at 12 micron have been accepted, because they satisfy the WC limits automatically. The aim of our observations is to investigate the nature of the IRAS PS located inside the WC box, but outside the region delimited by the Richards criteria. These involve IRAS colours different from those used by WC: (i) and (ii) . We have projected the Richards criterion on colour into the plane creating a region (hereafter called the ``Richards strip'') that overlaps in a diagonal fashion the WC box. A sample of 170 IRAS PS located inside the WC box, but outside the Richards strip, have been selected. For this selection we have also considered only the sky region observable with the Medicina radiotelescope with ). (No bias has been introduced by considering only these sources.) Finally, 10 IRAS PS with have been rejected, according to the classical IRAS spectrum of a SFR, rising towards longer wavelengths. Thus, our final sample contains 160 IRAS PS (hereafter the ``NR'' sample). Figure 1 shows the distribution in the colour plot of the IRAS PS (observable at Medicina) which satisfy the WC criteria (continuous line). The dashed lines bound the Richards strip, while the dot-dashed line stands for the points for which ; below this line are the rejected IRAS PS. In Figure 1 the IRAS PS are plotted with a cross if (Sample A) or with an open square otherwise (Sample B). The limit of 100 Jy at 60 micron have been chosen because:
The samples A and B have a different distribution in the colour plot of Figure 1. The bulk (459 out 501 sources; 92%) of the objects from the sample A are inside the Richards strip and the sources which extend out of this region are located very near its boundary. On the other hand, sample B extends out of the strip delimited by the Richards criteria: only the 75% (363/481) is inside this region, a smaller fraction than for sample A. This is an indication that the A and B samples contain IRAS PS associated with objects of different nature.
Figure 1 also shows the colour plot distribution of the NR sample. We have:
Thus, the colour region investigated with our observations is primarily populated by faint sources from sample B.
The observations were made with the 32-m radiotelescope at Medicina during several runs between 1990 and 1994. The half-power beam width at the frequency of the water maser line (; 22235.07985 MHz) is . The antenna efficiency of the radiotelescope was 38% with a maximum gain of . The system temperature in good weather conditions was 120 K. The calibration of the obtained spectra was made using the continuum source DR21; the uncertainty of calibration is 20%. The spectra are corrected for telescope gain changes with elevation. The pointing accuracy is (rms).
The observations were made in beam switching mode; for each source the integration time was 5 minutes (on and off source). The backend was an autocorrelation spectrometer with 1024 channels. A 25-MHz bandwidth was used, corresponding to a spectral resolution of and a total velocity coverage of . The average detection level () is 6 Jy.
Out of the 160 IRAS PS of the NR sample, only 11 show maser emission; 2 are new detections (IRAS and IRAS ). Therefore, the detection rate is low: 7%. Considering the detection rate relative to the two samples NRA and NRB, it can be noted that there is a strong dependence on the value. The detection rate for the NRA sample is 24% (10 detections out 42 sources). On the other hand, out of the 118 sources of the NRB sample, only one has been detected (1%).
Considering also the maser sources not detected with the Medicina radiotelescope due to either variability or insufficient sensitivity, but known from the literature, we add 11 other masers (7 of the NRA sample and 4 of the NRB sample). Therefore, the total number of known maser sources in the NR sample is 22. Table 1 compares the detection rates of this work with those given by P1 and P2. Since all these results have been obtained with the same detection limit, they are directly comparable. The comparison shown in Table 1 indicates that if one considers only the bright sources (), then the detection rate does not depend on the IRAS colours (24% vs. 26%). This means that the NRA sources are of the same nature of the bright ones located in the Richards strip: UC HII regions, as confirmed by the high detection rates. Therefore, it can be concluded that all the bright IRAS PS of the sample A are UC HII regions candidates. As a consequence, the lower limit of the Richards criteria on colour () does not select all high-mass protostars. In order to select these objects from the sample of IRAS PS located in the WC box, a less restrictive limit on should be used. Figure 1 shows also the distribution of the IRAS PS associated with known maser sources. As already reported, among the 22 masers of the NR sample, the majority (17 sources; 77%) belong to the NRA sample. It can be noted that maser sources do not populate in a homogeneous fashion the investigated colour region: they are located near the boundary of the Richards strip. Even if this is primarily due to the fact that the majority have high 60-micron flux densities and therefore reflect the distribution in the colour plot of the NRA sample, it is also worth noting that the 5 faint IRAS PS in Figure 1 are located near the Richards strip. The limit can be used for the selection of high-mass protostellar candidates, since it bounds almost all (except one) of the maser sources.
Table 1: Results of water maser surveys
Figure 2: Galactic latitude distribution of the observed samples.
The nature of the IRAS PS of the NRB sample remains an open question. Table 1 shows that there is a large difference between the two detection rates of the two faint samples (). This indicates that the lack of detections of the NRB sample primarily reflects the different nature of this sample and not a sensitivity effect. Since it is well known that maser emission is primarily associated with high-mass SFRs, the lack of masers indicates that the NRB sample does not contain objects of this kind. This is confirmed by the comparison between the galactic latitude distributions of the observed samples (NRA and NRB) shown in Figure 1. The distribution of the NRB sample is really larger than that of NRA sample, suggesting that the faint sources are relatively nearby and not connected with high-mass stars. Therefore, it is confirmed that the population of O-type stars estimated from the number of IRAS PS located inside the WC box may be overestimated. In order to obtain an estimate of the fraction of IRAS PS which, even if located inside the WC box, are not connected with an O-type star, we have to extrapolate our result to the whole sample of 1846 IRAS PS (without the instrumental restriction on declination). The estimated fraction is 11% (195 out 1846 sources). This value is a lower limit because it takes into account only the faint sources outside the Richards strip. In fact, as already reported in Sect. 2, P1 and P2 have concluded that also the faint IRAS PS of their sample (inside the Richards strip) may be not contain only O-type stars.
We have studied the frequency of occurrence of 22.2 GHz maser emission in a sample of 160 IRAS sources selected using the WC colour criteria to identify high-mass star forming regions. The aim was to verify if the maser properties of the IRAS PS located inside the WC region have any dependence on the IRAS spectrum and, consequently, to investigate the nature of the objects of the observed sample.
Out of the 160 IRAS PS, only 11 have been detected, 2 for the first time. Therefore, the overall detection rate is low: 7%. There is a strong dependence of the detection rate on the 60-micron flux. For the sources brighter than 100 Jy the detection rate is 24%, while for the weaker sources it decreases to 1%. Considering our results and those given by previous comparable surveys, it can be noted that the detection rate for the bright IRAS PS does not depend on the IRAS colours. This indicates that all the bright IRAS PS located inside the WC box are UC HII regions candidates.
There is a net distinction for the weak sources (): the detection rate is 11% for sources with greater than 0.61 and 1% for the rest. The lack of masers indicates that the NRB sample does not contain high-mass star forming regions, as confirmed by its galactic latitude distribution, much broader than that of UC HII region candidates. As a consequence, the population of O-type stars estimated from the number of IRAS PS located inside the WC colour box may be overestimated. The expected fraction of the WC sample not connected with O-type stars is at least 11%.