It is of interest to study the detection rate of methanol masers in the whole Galaxy. As a first step, we made a histogram of the detection rate versus galactic longitude (Fig. 2). The detection rate was calculated as the ratio of the number of detections to the total number of reported observations per galactic longitude bin. The histogram shows that the probability to detect the 6.7 GHz methanol masers is higher in the inner Galaxy. It does not change very much if one uses only IRAS sources selected by Wood & Churchwell colour criteria.
The distribution of 6.7 GHz masers on the Galactic plane is shown in
Fig. 3. This diagram is based on our data and the results of
Menten et al. (1991a);
Schutte et al. (1993);
MacLeod & Gaylard (1992);
Caswell et al. (1995a)
and van der Walt et al. (1995). In total, 270 sources were
taken into consideration. Only near kinematic distances were used in
cases of ambiguity. No kinematic distances were calculated in the zone,
restricted by two dashed lines, i.e., closer than 10 to the
center-anticenter line, where kinematic distances are highly unreliable.
The cross point of the dashed lines represents the position of the Sun;
the Galactic center is located at the (0,0) position. The arc
in the left part of the diagram is the Sagittarius arm.
The 6.7 GHz methanol masers typically show a significant velocity offset from the radial velocity of host molecular clouds. Figure 4 shows the histogram of the absolute difference between the radial velocity of the strongest methanol feature and the radial velocity of thermal CS lines (in some sources it was CO) for 157 methanol masers taken from this survey and surveys by Menten (1991a); Shutte et al. (1993); van der Walt et al. (1995), and Caswell et al. (1995a) and for which the cloud velocity was available.
The average difference is 5.5 0.7
km s-1, which is a factor of two to three larger than the thermal
velocity dispersion, even in hot molecular clouds. In some sources this
difference is larger than 10 km s-1. The velocity shift can be seen in
the spectra in Fig. 1 where the CS velocity is shown by dashed lines. The
methanol spectra themselves show a large velocity dispersion, the difference
between extreme spectral features is typically 5 to 15 km s-1, with 25
to 30 km s-1 in some sources. Figure 5 shows a histogram of the velocity range
for 157 6.7 GHz methanol masers taken from all available surveys
(this paper, Menten 1991a;
Shutte et al. 1993;
van der Walt et al. 1995,
and Caswell et al. 1995a).
Is is evident that the velocity range less than
2 km s-1 is rare, the mean velocity range is 8.0 0.5 km s-1,
and exceeds velocity dispersion of host molecular clouds.
In this respect the 6.7 GHz methanol masers are similar to H2O masers
with high velocity spectral features, although to a smaller scale.
One possible explanation of the large
velocity dispersion in the spectra of 6.7 GHz methanol masers could be
related to the excitation mechanism. A large velocity gradient can provide
easier escape of
photons,
a condition required in some maser models to maintain the population
inversion.
Another possible explanation of the large velocity dispersion is that
the maser condensations are gravitationally bound to massive stars,
and are circling around them with Keplerian velocities. If the mass
of the star is (O-star), then the Keplerian velocity of
5 km s-1 corresponds to a distance from the star of 1080 A.U. The observed
linear separation between 6.7 GHz maser spots is of this order of
magnitude
(Norris et al. 1993). Moreover, in many cases the maser
spots lay along straight lines or arcs, and it was suggested
(Norris et al. 1993) that they originate in circumstellar disks. The large
range of methanol maser velocity features is consistent with their
circumstellar origin. The massive O-stars associated with 6.7 GHz
methanol masers could be very young, newly born stars embedded in a dense
dust molecular core. The star itself is invisible because of a large
extinction in the core, but could ionize an ultracompact HII region,
which can be detected as a continuum radio source.
The association of 6.7 GHz methanol masers with ultracompact HII regions
established in this study (see Sect. 5.3) corroborates this conclusion.
As noted in the introduction, our observing sample contained 326 IRAS sources which we selected with the colour criteria for ultracompact HII regions by Wood and Churchwell (1989). For this sample the methanol maser detection rate was 11 per cent. Of the total of 326 IRAS sources 76 coincided with ultracompact HII regions detected as compact continuum radio sources at 5 GHz by Becker et al. (1994) in a deep survey with the VLA. In 19 of them 6.7 GHz methanol masers were detected. The corresponding detection rate for the compact radio sources is 25 per cent which is considerably higher than the mean rate 11 per cent for the whole IRAS sample. Therefore the compact thermal continuum radio sources are much better candidates for detection of 6.7 GHz methanol masers than IRAS sources selected solely with Wood and Churchwell criteria.
It is well established, that the 6.7 GHz masers are variable (see,
e.g., Caswell et al. 1995b). The study of maser variability,
which requires more systematic and long term observations, was beyond the
scope of this work; however, a comparison of our spectra with those
published in the literature shows variations larger than the estimated
errors for some sources. In those cases when the spectral resolution
is different, as in
van der Walt et al. (1995), or when the observed positions differ by a sgnificant
fraction of the telescope beamwidths, such
as in the case of 02455+6034 and 18324-0820, the variability of the source
cannot be asserted.
MacLeod &
Gaylard (1992);
Gaylard & MacLeod (1993);
Schutte et al. (1993);
van der
Walt et al. (1995) made their surveys with
the 26-m Hartebeesthoek radio telescope whose
beamwidth is 7 arcmin, close to the
Medicina value, and their observed positions are either the same as ours
or the difference is not larger than 20-30 arcsec. Therefore when the
spectra, obtained by these authors, differ from ours (in 174.20-0.08,
17589-2312, 18310-0825, 18379-0546, 78.12+3.63) these differences most
likely may be attributed to the source variability.
This conclusion is particularly true in the case of 174.20-0.08 where
the number of the spectral components and the relative intensities are
different.
In the case of 17599-2148 we have a non-detection while it was observed by
Schutte et al. (1993) with Jy well over our sensitivity limit.
Other examples of possible
variability are our detection of weak masers towards 18064-2008,
18236-1241, and 19220+1432, where no 6.7 GHz masers were found by van der
Walt et al. (1995).
Thus, our results, like those by
Caswell et al. (1995b), show that the 6.7 GHz
masers are often variable.
Copyright The European Southern Observatory (ESO)