Mon. Not. R. Astron. Soc. 301, 640-698 (1998)

Studies of ultracompact H II regions - II. High-resolution radio continuum
and methanol maser survey

A. J. Walsh,^1 M. G. Burton,^1 A. R. Hyland^2 and G. Robinson^3

^1 Department of Astrophysics and Optics, School of Physics,
   University of New South Wales, NSW 2052, Australia
^2 Southern Cross University, Lismore, NSW 2480, Australia
^3 School of Physics, University College, University of New South Wales,
   Canberra, ACT 2600, Australia

Accepted 1998 July 23. Received 1998 July 17; in original form 1998 April 15


 ABSTRACT

High spatial resolution radio continuum and 6.67-GHz methanol spectral
line data are presented for methanol masers previously detected by Walsh et
al. (1997). Methanol maser and/or radio continuum emission is found in 364
cases towards IRAS-selected regions. For those sources with methanol maser
emission, relative positions have been obtained to an accuracy of typically
0.05 arcsec, with absolute positions accurate to around 1 arcsec. Maps of
selected sources are provided. The intensity of the maser emission does not
seem to depend on the presence of a continuum source. The coincidence of
water and methanol maser positions in some regions suggests there is overlap
in the requirements for methanol and water maser emission to be observable.
However, there is a striking difference between the general proximity of
methanol and water masers to both cometary and irregularly shaped
ultracompact (UC) H II regions, indicating that, in other cases, there must
be differing environments conducive to stimulating their emission.
We show that the methanol maser is most likely present before an observable
UC H II region is formed around a massive star and is quickly destroyed as
the UC H II region evolves. There are 36 out of 97 maser sites that are
linearly extended. The hypothesis that the maser emission is found in
a circumstellar disc is not inconsistent with these 36 maser sites,
but is unlikely. It cannot, however, account for all other maser sites.
Key words: masers  stars: formation  ISM: general  H II regions radio continuum: ISM
	   radio lines: ISM

1 INTRODUCTION

Ultracompact (UC) H II regions are the small bubbles of ionized gas
surrounding a newly formed massive star, embedded in its natal
molecular cloud. They have received a considerable amount of
attention in recent years because of their effect on the dynamics and
energetics of molecular clouds, as well as by being fascinating
objects in their own right. They provide a major influence in shaping
molecular clouds by virtue of the energy they dump into the
medium, whilst still embedded.
  Two efficient methods for identifying such regions have
been proposed. The first (Wood & Churchwell 1989a,
hereafter WCa) involves the selection of candidates based on
their far-infrared colours, as detected by IRAS. The strong
emission of dust, radiating with a blackbody spectrum,
peaking around 100  m, gives the reddest colours of objects
listed in the IRAS PSC, in the bands of 12, 25 and 60 um. The
second method used to identify UC H II regions is to search for
masing activity and, in particular, 6.67-GHz methanol maser
emission, discovered by Menten (1991), as it is a very strong and
widespread transition, and is considered to be a signpost of a UC H II
region.
  Previously (Walsh et al. 1997, hereafter Paper 1) we chose 534
southern UC H II region candidates from the IRAS PSC, using the
selection criteria of WCa. Of those 534 candidates, we have reported
215 UC H II regions, identified either by association with methanol
maser emission at 6.67 GHz or by identification of a compact radio
source. Unfortunately, the original survey suffered from poor spatial
resolution because of a large primary beam of 3.3 arcmin, so that we
could not be sure of the association between methanol masers and UC
H II regions. UC H II regions are also gregarious and so a large beam
may also pick up emission from more than one UC H II region. It is
the purpose of this paper to present high (arcsec) resolution data for
both the methanol maser emission and any associated compact
continuum sources in order to determine the nature of the association
between UC H II regions and methanol masers.

2 OBSERVATIONS AND DATA REDUCTION

The observations were made using the Australia Telescope
Compact Array (ATCA) during 1994 July/August and 1995
January/February. The ATCA is an array of six 22-m antennas on
an east-west baseline with the widest separation being 6 km. Since
the ATCA is a synthesis telescope, observations of each object were
made by making a series of short integration cuts, of 2 min each, a
number of times, spaced over a period of 12 h. Typically five cuts
were taken on each object, but some have fewer because of bad
weather. The effect of this is to decrease the sensitivity of the final
image and to increase the relative inaccuracies of derived positions
of unresolved maser sources. Observations of program sources were
interspersed with calibrator sources every 30 min. The ATCA can
simultaneously observe two frequencies. The first selected frequency was
chosen to be the 8.64-GHz continuum over a bandwidth
of 128 MHz (with 32 channels). The second frequency was chosen
so as to observe the 6.67-GHz methanol maser emission. The
bandwidth at this frequency is 4 MHz, with 1024 channels, giving
a velocity resolution of 0.18 km s^-1 and a velocity coverage of 180
km s^-1. The observing frequency was varied between 6.667, 6.668,
6.669 and 6.670 GHz, as appropriate, to take into account the radial
velocities of the maser sources observed in Paper 1.
  The data were processed using the aips data reduction package.1
Each frequency was processed separately. The radio continuum
data at 8.64 GHz were averaged into a single frequency channel,
using avspc. Bad data were interactively edited with tvflg,
antenna gain solutions were found with atcalib and then calibrated
with atclcal. Single source files were created, and calibrations
were applied with split. The resulting source files were then
imaged and cleaned using mx, with typically 500 cleaning iterations.
The image maps were then visually inspected for evidence of
radio continuum emission.
  The spectral line data at 6.67 GHz were reduced in a similar
fashion, except that the data were not compressed into a single
frequency channel, to preserve the spectral information, and the
instrumental bandpass was removed with bandpass before single
source files were created, with split. After individual source files
were created, each UV source file was continuum fitted using uvlsf
so that separate images of the continuum emission at 6.67 GHz and
methanol emission could be made. The spectral line data were
corrected for the motion of the array with respect to the local
standard of rest, using cvel, and then both continuum and spectral
line images were made using mx. The final images were inspected
for continuum and maser emission in a fashion similar to the 8.64-GHz data.
  The positions of each methanol maser spot were determined to an
accuracy of 1.8 arcsec for each field, and then more accurate
positions were determined using the method of `super resolution'
used by Norris et al. (1993). For each maser spot, an image was
made using only the channels in which that maser spot was
dominant. The emission was then Gaussian fitted using jmfit to
accurately determine the position of the maser emission, assuming
it is unresolved. We estimate that the relative uncertainties of
individual maser spots in a single field is 50 mas for the majority
of cases. This increases for masers weaker than about 1 Jy, up to 0.5
arcsec for the weakest sources imaged, at 0.3 Jy. There is also
considerable uncertainty in the positional accuracy in declination
for any source closer than 10  to the celestial equator. We estimate
that a strong maser at a declination of -15  would have an associated
maser spot declination error of  0.1 arcsec and may increase to
larger than 2 arcsec for sources at a declination of -11. It is
estimated that the absolute positional uncertainty of any features
will be 1 arcsec. These estimates are made by correlating the
positions of the same maser spots that were obtained in both
observing runs. A total of 29 maser spots could be used to
experimentally determine the errors. We believe that the effect of
any proper motions of the maser spots over the short interval of 6
months between the two runs will be minimal.
  As a result of the nature of the short cut integrations used, only
the simplest morphologies of resolved continuum sources are easily
discernible from artefacts. Any interpretation of morphologies is
thus limited to the overall shapes of resolved regions. Furthermore,
radio continuum sources close to a declination of 0  are greatly
elongated in the declination axis. No interpretation is placed on the
morphologies of such continuum sources. The 1j sensitivity limits
are typically 1 mJy at 8.64 GHz and 10 mJy at 6.67 GHz, for the
continuum images. The flux limit on the methanol maser emission
is limited to detecting a single peak within the spectrum and is
typically 0.5 Jy, although it is possible to image some weaker
sources by using the known velocity coverage of a maser source
from Paper 1 and Caswell et al. (1995a), and imaging over that
velocity range.
  In Paper 1, it was noted that some of the listed maser flux
densities were underestimated, since the Parkes beam may not
have been centred on the maser emission. With the accurate
positions of the methanol maser and continuum emission obtained
in this paper, it is possible to correct for the profile of the ATCA
primary beam for most sources using pbcor. The HPBW of the
ATCA primary beam is approximately 8.6 arcmin, so that rescaling
of flux densities is only accurate out to about 5 arcmin from the
pointing centre. Beyond this the uncertainties in rescaling become
too large because of the low efficiency of the beam.

3 RESULTS

There are 276 observed fields, including 193 of the maser and UC
H II regions identified in Paper 1 and a selection of other IRAS
sources with no previously detected maser or radio continuum
emission. The remaining 22 IRAS sources with methanol and/or
radio continuum emission listed in Paper 1 were not observed
because of poor conditions. Each field chosen was centred on the
relevant IRAS position. Out of the 276 fields, a total of 364 sites
were found which had either methanol maser emission, a compact
radio continuum source, or both. There are 233 sites that exhibit
methanol emission; only 46 of these have associated radio continuum emission.
There are a further 131 sites that exhibit radio
continuum but no methanol emission, making a total of 177 sites
with continuum emission. There are 68 fields observed in which
there is no detectable methanol or continuum emission; these are
listed in Table 1. There are also three sources listed in Table 1 that
exhibit methanol maser emission, but could not be located in the
cleaned image. This may be a result of strong emission far from
the pointing centre, being picked up in the sidelobes.
  Details for those sources showing methanol and/or continuum
emission are given in Table 2. The first column lists the associated
IRAS name. Columns 2 and 3 give the Galactic coordinates for the
IRAS point source, which was used as the pointing centre. Columns
4-7 contain information on the methanol maser emission: columns
4 and 5 list the methanol maser offset position from the IRAS
position. Column 6 lists the peak flux density for the methanol
emission. Column 6 also has an identifying letter that corresponds
to the positions of selected methanol spots in the images of Figs 1
and 2. Column 7 lists the radial velocity for the methanol spot.
Columns 8-12 contain information on the radio continuum
emission: columns 8 and 9 list the RA and Dec. offset positions
for the radio continuum from the IRAS pointing centre, respectively.
Column 10 lists the 8.64-GHz continuum peak flux density, column
11 lists the 6.67-GHz continuum peak flux density. Integrated fluxes
have not been derived here, as the poorly sampled data makes it
difficult to obtain accurate flux measurements. The peak flux
densities are listed as a guide to the brightness of each region,
although this number does not take into account the extended
morphologies of some of the continuum sources. Column 12 lists
the proposed morphological type for the continuum emission,
where U means unresolved, I means irregularly shaped, C means
cometary shaped, D means double peaked and P means partially
extended (see Section 4.3 for definitions of the morphological types).

 )1 Provided by the ATNF.

   Table 1. The observed sources in our data set with no positive
   methanol or radio continuum identification.

   00445-1207 )a       07278-1826       16475-4609       18060-1816
   05173-0555          07333-1838       16557-4002       18067-1921
   05283-0412          07358-3243       17149-3916 )b    18072-1954 )d
   05304-0435          07395-1437       17178-3742       18092-1742
   05327-0529          07399-1435       17234-3405 )c    18159-1648 )b
   05331-0515          07502-2618       17242-3513 )c    18234-1444A
   05338-0624          08438-4340       17260-3445 )b    18246-1032
   05363-0702          08563-4711       17352-3153 )c    18263-1036
   05375-0731          09006-4830       17430-2822 )b    18308-0911
   05394-0151          09017-4814       17430-2900       18310-0806
   05400-0154          09018-4816 )b    17431-2846       18342-0655
   05396-0153          09230-5148       17432-2855       18360-0537
   06046-0603          09238-5153       17440-2823       18411-0338 )b
   06343-1036          10049-5657       17441-2910       18420-0512
   06361-0142          10309-5745       17443-2821       18451-0332
   06529-0755          11101-5829 )b    17474-2637       18491-0207
   06547-1012          16112-4943 )b    17488-1741       18595-3712
   07207-1435          16186-5044       17527-2439       19590-1249

 a Planetary nebula.
 b Methanol detected in Paper 1, but is not evident in these data,
 because of increased noise levels and/or variability of the maser source.
 c Methanol emission is present, but cannot be located in the CLEANed
 image.
 d Observed velocity range did not cover methanol emission.

  Sources listed in Table 2 and indicated with an asterisk (*) have
not been corrected for the profile of the primary beam. This is
because their offset positions are too great for an accurate rescaling
to be performed. Instead, the uncorrected flux densities are listed,
which will be considerably lower than the true values.
   CLEANED images of selected radio continuum sources are shown
as contour maps in Fig. 1. The criterion for inclusion of a continuum
source in this figure is that either the emission is extended in some
way, or that there is methanol maser emission close to the continuum
emission. Unresolved continuum sources with no associated
methanol emission are not shown as all the relevant information on
them is contained in Table 2. The positions of the methanol spots are
shown either by their associated letter from Column 6 of Table 2, or
by a `plus' (+) symbol. The letters were only used when there was
more than one maser spot, but not so many that the letters would
become unreadable. Fig. 2 shows offset positions for the methanol
maser spots from the IRAS pointing centre, as listed in Table 2. Only
those sites of methanol emission with four or more maser spots are
shown. In some cases the positions of the maser spots are in both
Figs 1 and 2. These cases are shown in Fig. 2 to indicate the letter for
the associated maser spot, since the continuum contour plot would
be too crowded to show the individual letters.

3.1 Comments on individual sources of interest

This section contains details of fields that are relevant to the results
presented in this paper. Fields which are illustrative of the major
characteristics of the data set are also described. Some other fields
of major importance have been detailed in Paper 1.

 IRAS 13471-6120. The methanol emission has also been
observed with the ATCA by Norris et al. (1993) and our relative
positions of the maser spots correlate well, although there is an
absolute offset of around 1 arcsec. We have detected a new
methanol spot, with a radial velocity of -54.5 km s-1 (A). The
maser emission is slightly offset from the centre of the unresolved
continuum emission, but is still within the continuum contours.
Caswell, Vaile & Forster (1995b) reported methanol and OH maser
emission from this source. The position of the OH emission appears
to be 0.5 arcsec offset from the methanol emission, but within the
continuum emission. This is not apparent from our image in Fig. 1,
as there is an offset of 1 arcsec in the absolute position with that of
Caswell et al., assuming the methanol emission reported here is
coincident with that of Caswell et al.

  IRAS 16484-4603. The relative positions of the methanol spots
agree with those of Norris et al. (1993). New features are seen in our
data, namely A and B. The position of the methanol emission with
respect to the continuum is similar to that of IRAS 13471-6120. The
continuum emission seems to be slightly extended to the west, with
the methanol also on the west side of the centre of the continuum
emission. The radio continuum image of Ellingsen, Norris &
McCulloch (1996) indicates that the emission is extended to the
north-east and that the angular position of the methanol maser spots
is perpendicular to this continuum emission. OH maser emission is
also reported by Caswell et al. (1995b) in three locations, two of
which are within the continuum emission contours, the third lying
about 4 arcsec to the north of the continuum emission. Water maser
emission is reported close to the continuum emission by Forster &
Caswell (1989). It is not clear whether the OH and water maser
emission sites are offset from the continuum or not, because of the
absolute positional uncertainty.

  IRAS 16533-4009. There are two distinct sites of methanol
emission, separated by 0.23 pc (assuming a near kinematic distance
of 2.5 kpc), in which the individual spots of both sites correspond
well with the relative positions published by Norris et al. (1993). An
unresolved continuum source is found close to the southerly site of
methanol emission, along with two OH maser emission sites. Both
sites of methanol emission seem to have a linear structure. The
southerly linear structure is extended radially with respect to the
continuum emission.

  IRAS 17016-4124. There are two separate sites of methanol
emission. The first is within the contours of continuum emission.
The second is offset by 40 mpc (assuming a distance of 2.7 kpc) to
the west of the centre of the continuum emission. The shape of the
continuum emission is cometary (about 80 mpc long), with the
methanol emission found towards the `head' of the continuum. OH
and water maser emission is reported by Forster & Caswell (1989),
both are found towards the `tail' of the continuum cometary region.
The methanol site overlying the continuum has weak evidence for a
linear structure, which is pointing perpendicular to the cometary
flow. The other offset methanol site has a similar linear structure,
pointing in a similar direction.


     Table 2. List of the 364 UC HII regions detected either by methanol
     maser emission or a compact radio continuum source. Each source
     denoted with an asterisk (*) indicates that the provided flux or flux
     density has not been corrected for the beam profile, and the numbers
     stated are most  certainly underestimates.

------------ file HII.tb2 --------------------------------------------------


  IRAS 17175-3544. This source is better known as NGC 6334F.
The continuum emission is of cometary shape, about 44 mpc across
(assuming a distance of 1.7 kpc). Methanol emission is found
scattered within the head of the cometary shape, along with OH
emission, detected by Gaume & Mutel (1987). OH and water maser
emission is also reported by Forster & Caswell (1989) within the tail
of the cometary continuum emission. We also find a separate site of
methanol emission 40 mpc to the side of the continuum emission.
Our data agree well with those of Ellingsen et al. (1996), although
the continuum emission they find extended to the south-east is
below our lowest contour and is not evident in Fig. 1. A third,
presumably unrelated, maser site is located approximately 2 arcmin
north of NGC 6334F.
  IRAS 17470-2853. The radio continuum emission consists of an
unresolved strong source, with weaker emission extending away
from the unresolved source, to the north-west. The extent of this
emission, including both sources, is 0.7 pc, assuming a distance of
9.1 kpc. It is not clear whether the extended emission is associated
with the unresolved source. Methanol maser emission is found at
the edge of the unresolved continuum source, at the point where the
two continuum features would meet. OH maser emission is also
located close to this point (Forster & Caswell 1989). A single spot of
methanol emission is found within the weak, extended continuum.
Water maser emission, detected by Forster & Caswell (1989), is
found between the two continuum features. The positions of the
maser sources do indicate that the two continuum features are linked.
  IRAS 18032-2032. Our continuum image shows a single unresolved source,
but the deeper image of Hofner & Churchwell (1996)
indicates a second continuum source 12 arcsec to the north-west. At
the position of this second continuum source, we find a site of maser
emission, comprising five maser spots. There are also two maser
spots associated with the first continuum source, overlying the
continuum emission contours, but offset to the south of the centre of
emission by 1 arcsec. Hofner & Churchwell also report a line of
water maser emission leading from one continuum emission source
to the other, and interpret this as a ridge of star forming activity.
  IRAS 18110-1854. The continuum emission is cometary in
shape, and methanol emission is found beyond the head of the
continuum contours. The position of the methanol emission also
agrees with that of water maser emission reported by Hofner &
Churchwell (1996).


4 DISCUSSION
4.1 IRAS selection criteria

It is found that the regions with maser and/or continuum emission
tend to have bright IRAS sources. In fact, 96 per cent of the IRAS
sources which have methanol or continuum emission within 1
arcmin of the IRAS position have 100  m flux densities greater
than 500 Jy, whereas 69 per cent of all of the 534 candidates have
100  m fluxes greater than 500 Jy. We cannot say whether the
weaker IRAS sources represent some contamination of the original
IRAS selection criteria (explained in Paper 1), since it may be that
the weaker sources are further away and therefore we are less likely
to detect methanol or continuum emission from them. As there is
no radial velocity information for those IRAS sources with no
maser emission, we cannot derive distance estimates and test this
hypothesis. However, van der Walt (1997) has shown that those
fainter IRAS sources are most likely contamination of embedded

non-ionizing stars, as a result of the broader distribution of them
about the Galactic plane. Similarly, Ramesh & Sridharan (1997)
have shown that the IRAS PSC selection criteria of WCa may
include up to 75 per cent of sources other than UC H II regions. They
suggest that these IRAS sources may be the result of embedded non-
ionizing stars, or of high mass equivalents of class 0 objects.


4.2 The association of methanol and continuum emission
Fig. 3 shows the distribution of the projected sizes of radio
continuum regions when maser emission is present (allowing us
to determine the distance). The top histogram includes all extended
continuum emission regions that do not have any methanol emission
projected on to the radio continuum contours. The lower
histogram shows the size of the extended continuum regions that
exhibit methanol maser emission projected on to the continuum
contours. It is clear that the continuum sources with maser emission
are generally considerably smaller than those without, and quantified
by a Kolmogorov-Smirnov (K-S) test, which shows the two
distributions have only a 0.7 per cent chance of coming from the
same population. Also, no continuum source is found with asso-
ciated methanol emission greater than 180 mpc in projected size.
This suggests that the maser emission is detectable before the
continuum emission and is associated with young, and therefore
small, UC H II regions, probably being destroyed as the UC H II
region expands. This agrees well with the hypothesis stated in Paper
1 that the masers are present before the UC H II region develops.
  The mechanism for pumping the methanol maser, although not
clearly understood, is generally considered to be linked to the radio
continuum emission (Menten 1991). We can now test this hypothesis,
by comparing the incidences of methanol and radio continuum
emission. As stated in Section 3, most sites of maser emission are
not associated with any detectable continuum emission. Furthermore,
we can place stringent limits on the association of methanol
and continuum emission by examining the relationship between
them. In Table 2, we have provided peak flux densities only as a
guide to the brightness of the continuum emission, so we cannot
directly compare maser and continuum fluxes. We can, however,
compare the distribution of the brightness of maser emission that is
and is not associated with detected radio continuum emission. If the
maser emission were closely associated with radio continuum
emission, we would expect the strength of the maser emission to
be dependent on the proximity of the continuum source. We have
tested this in the three histograms shown in Fig. 4. Here we plot the
distribution of the number of maser sources as a function of a
measure of the integrated maser luminosity. The integrated luminosity
measure is just the summation of the flux contributions of each
maser spot within a given site multiplied by the square of the
distance to the site. Three histograms are shown: the top one for all
maser sites that have no associated continuum emission within a
0.2-pc projected distance; the middle histogram for those maser
sites within a 0.2-pc projected distance of the peak position of a
radio continuum source, and are therefore likely to be associated;
the bottom histogram shows those maser sites that are actually
projected on top of the radio continuum contours, so that they lie
directly in front of the continuum source, along our line of sight. If
the maser emission is affected by the proximity of a radio continuum source,
then we would expect the distribution of the top and middle histograms
to be different. A K-S test indicates that they are
drawn from the same distribution, with 61 per cent probability.
Hence, we find little evidence to suggest that the intensity of the
maser emission is affected by radio continuum emission.


------------------------Fig.1-2-------------------------------
Figure 1. Radio continuum and metanol maser maps selected sources.
	  Not all radio continuum sources are shown here.
Figure 2. Relative positions of maser spots within an individual
	  maser site. The offsets are in arcsec from the IRAS source.
	  Only those maser sites with four or more spots are shown.
	  The typical size of the error bars on each maser spot are shown
	  by a representative error bar, usually in the bottom right corner
	  of each plot.

Also, if maser emission were enhanced by continuum photons, at
the masing frequency, along the line of sight, we would find an
enhancement in the brightness of maser emission when it overlaid
continuum contours. Using a K-S test, we are 51 per cent confident
that the bottom histogram in Fig. 4 for maser sites projected on to
continuum contours is no different than that for those masers with
no associated continuum source (the top histogram). It is also noted
that in the limited sample of maser sites that are projected on to
observed continuum emission (32 sites), there is no correlation
between the brightness of the maser emission and the brightness of
the continuum emission. A correlation would be expected if the maser
emission were significantly influenced by the continuum photons.
Thus, we can find no evidence that the maser emission is enhanced by
the presence of an observable background continuum source.
  It is possible that the methanol emission that is offset from the
observed continuum arises from radio continuum sources which are
too small and/or weak to be detected by our observations. The
methanol emission, in the majority of cases, may come from UC H II
regions that are very young, and not large or bright enough to show
up in our data as continuum sources. The observed continuum
sources, then, are UC H II regions that are generally older than the
regions responsible for pumping the maser transition.

4.3 Morphological type

As a result of the nature of the short cut integrations used,
interpretation of any morphologies is restricted to overall shapes
as most of the fine detail is hard to distinguish from artefacts. The
morphological types identified in Table 2 are cometary, irregular,
partially extended, double and unresolved. To compare this with
previous surveys, it is necessary to explain the criteria used to
classify the objects. We have chosen to use the classification criteria
of Wood & Churchwell (1989b, hereafter WCb) for cometary,
irregular and unresolved (spherical) morphologies. Other morphologies used
here are defined as follows: if the majority of emission
can be easily fitted with two two-dimensional Gaussians and is not
classified as one of the above, then the morphology is designated as
double. Sources that are listed as partially extended are essentially
unresolved, but some extended emission can be seen, although this
extended emission is not large enough to distinguish between any of
the above extended morphologies. Two morphologies used by WCb
are not listed here, namely the shell and core-halo morphologies.

None of our contour plots show conclusive evidence for such
morphologies. There are two reasons for this: first, as noted by
WCb, there are only a relatively small number of such regions,
complete ellipse towards the tail of the cometary region, although
the limited sensitivity does not show this explicitly. It may be
possible that shell and cometary morphologies arise from similar
circumstances. It has been suggested by Hofner & Churchwell
(1996) that the UC H II regions with a cometary morphology may
arise from a density gradient, with a high density at the cometary
head and low density at the tail. It is possible, then, that shell
morphologies occur if the density gradient is not as severe as that
for cometary regions. The relative number of morphologies found
in this survey is: 30 per cent unresolved, 30 per cent cometary, 28
per cent irregular, 7 per cent partially extended and 4 per cent
double peaked.
  How are cometary and irregular morphological types related?
Previously, a survey by Hofner & Churchwell (1996) of water maser
and continuum emission sites has indicated that there is a distinction
between the morphological types and the proximity of maser
emission. They show that the maser emission associated with noncometary
regions (including irregular morphologies) is often projected against
the continuum emission contours, whilst maser
emission is more likely to be offset from the continuum emission
of cometary regions. For the cometary UC H II regions, this was
interpreted as evidence for the maser emission arising from an
undeveloped and undetected UC H II region, associated with hot
compared to the other morphologies (16 per cent for core-halo and 4
per cent for shell). The second reason is that our data have lower
sensitivity, dynamic range and resolution than the data of WCb. The
weak extended emission may not be picked up from a core-halo
object in our survey, and may be classified as an irregular or
unresolved object, as is the case with IRAS 18006-2422. Some
shell morphologies identified by WCb are not evident in our data,
such as IRAS 17574-2403 and IRAS 18021-1950, where our larger
beam was not able to resolve the structure properly, particularly in
the central region with less emission (WCb's resolution is 0.4
arcsec). It should be noted that some of our images in Fig. 1 may
have some of the characteristics of both cometary and shell
morphology. Examples of this are IRAS 10303-5746, IRAS
14095-6102 IRAS 14567-5846, IRAS 14594-5824 and IRAS
16445-4516. Each of these sources has continuum emission that
extends in a parabolic shape, but is suggestive of forming a
ammonia clumps (Cesaroni et al. 1994). Thus, the dense core
associated with the water masers and ammonia emission creates a
density gradient that shapes the cometary UC H II region. We can
produce a histogram plot similar to fig. 23 of Hofner & Churchwell
for the methanol masers, as shown in Fig. 5. This figure shows the
projected distance of the sites of methanol emission from the radio
continuum source, for the cometary and irregularly shaped con-
tinuum morphologies. In comparison to the relative positions of
water masers, our data show a contrast for the different morphol-
ogies. The continuum sources with a cometary shape do show a
significant increase in the occurrence of methanol emission closer
to the continuum, whereas no such correlation is found for water
maser emission. Also, in contrast to the correlation found with
water masers and irregular UC H II regions, there is little correlation
between the positions of the methanol maser and UC H II regions of
irregular morphology. This difference between water and methanol
maser emission highlights the different environments that are
conducive to their formation. A K-S test of the distributions of
maser sites from the irregular and cometary-shaped continuum
sources indicates that there is only a 4 per cent probability that the
two are drawn from the same distribution. Thus, we believe that the
two morphologies are physically different from each other, and are
not the result of some projection effect.

Figure 3. Histograms of the projected sizes of extended radio continuum
     regions. The upper histogram shows all radio continuum sources without
     maser emission projected on the continuum contours. The lower histogram
     shows only those continuum regions where maser emission is projected on
     the continuum contours (and presumably directly associated with the
     continuum emission). It is obvious that the continuum regions with maser
     emission are generally smaller, implying that maser emission is
     associated  with the smallest, and therefore youngest, UC H II regions.

Figure 4. Three histograms are shown for the integrated flux of the
     maser emission. The top one is for all maser sites that have no
     associated radio continuum source. The middle one shows those maser
     sites that have a continuum source within 200 mpc projected distance, and
     are therefore possibly connected. The bottom one shows those maser sites
     that are projected on to the continuum contours such that the continuum
     photons may play a part in enhancing the maser emission along the line of
     sight. The similarity of the three distributions implies that there is no
     evidence the maser emission is enhanced by the presence of a continuum
     source.

Figure 5. Two histogram plots of the distribution of methanol maser sites
     from a nearby continuum source are shown for cometary (upper histogram)
     and irregularly (lower histogram) shaped UC H II regions. The masers are
     found preferentially closer to the cometary continuum regions, but not
     to the  irregularly shaped continuum sources.

  Even though we find that the distribution of methanol and water
masers about continuum sources is different, there are a number
of sources in which there is good positional coincidence of the
two types of masers, for example IRAS 18110-1854 and IRAS

18032-2032. It seems that, in these cases, they are associated with a
very young or undeveloped UC H II region that is not currently
observable as a continuum source, but may be visible as a hot
ammonia clump. This suggests that the masers may in some cases
arise in similar circumstances.

  It is possible that both maser species have more than one
environment in which they may be observed. Both methanol and
water maser emission are found associated with ammonia clumps.
However, methanol maser emission is also seen preferentially
closer to cometary radio continuum sources, where water maser
emission is not. Furthermore, water maser emission is also found
preferentially closer to irregularly shaped radio continuum sources,
where methanol maser emission is not.

4.4 The structure of methanol maser spots within a maser site

It has been previously reported (Norris et al. 1993), for a small
number of objects, that the relative positions of methanol maser
spots tend to lie along a line or arc, with a velocity gradient along the
axis. The interpretation given by Norris et al. (1993) is that the
masers lie within an edge-on protoplanetary disc. This data was for
the brightest 15 regions of methanol maser emission known. With
our considerably larger data base, we are able to further test this
hypothesis. Not all the 238 regions identified with maser emission
can be used, as many have few maser spots. Some 97 regions have
four or more maser spots, and were used to examine their relative

positions. These are shown in Fig. 2. As can be seen there are a
variety of shapes, from well defined linear/arc structures to apparently
random patterns. As it is hard to make a good quantitative
estimate of the proportion of masers lying in a linear/arc structure,
we have categorized the maser sites according to the ratio of major
to minor axes in the spread of maser spots. If the ratio is less than 3,
then the maser site is defined as having no evidence for linear
structure; between 3 and 5, some linear structure; and greater than 5
means a well-defined linear structure. Using these definitions we
find 61 that have no linear structure, 27 that have some linear
structure and 9 that have a well-defined linear structure.

  One way to test the hypothesis that the masers originate from a
circumstellar disc is by comparing the data we have to the expected
Keplerian orbits in a rotating disc. Our method estimates the
minimum central mass that will produce the observed radial velo-
cities. Fig. 6 shows P-V (position-velocity) diagrams for those 36
maser sources showing evidence for a linear structure. A P-V
diagram shows the relative position of a maser spot along the line
which the maser site is extended on the horizontal axis. The vertical
axis is the measured radial velocity. The P-V diagram can be used to
estimate the central mass in two ways. The positions of the spots are
either confined to two quadrants of the P-V diagram (with the origin
signifying the position and radial velocity of the central mass), or
they lie in a line. These two cases will be dealt with separately.

4.4.1 P-V diagrams with spots in two quadrants

The physical interpretation of maser spots lying within two quadrants
of the P-V plane is that they are orbiting around a central

mass, located at the origin, with all orbits on one side being towards
the observer and all orbits on the other side moving away, relative to
the rest frame of the central mass. Since there is currently no
velocity or precise positional information on the central stellar
object, the origin cannot be determined. However, all maser spots
must lie in one or two quadrants of the P-V diagram. For each
maser spot, the mass of the central star can be calculated using the
following equation, which describes Keplerian motion:

		       [ formula ]

where V is the radial velocity of the maser spot with respect to the
central mass (i.e. the origin), R is the distance of the maser spot from
the origin and   is the angle between our line of sight and the maser
velocity vector. As it is not known what part of the motion of each
spot is along the line of sight, the radial velocity, and hence the
calculated central mass, will only be a lower limit; we are determining
values for M cos2. Furthermore, our choice of the location of
the origin has been determined to find a minimum value for the
central mass. Each mass listed for those P-V diagrams in Fig. 6
with masers in two quadrants is the largest value of M cos2  found,
but is still a lower limit to the central mass. Most of these lower
limits do seem to fit in with what would be expected for an O or
early B-type star, but in several cases we note that the implied
masses are extremely large (> 100 M ), which we consider unlikely.

   Furthermore, the locations of the maser spots (in quadrant
diagrams) are not uniformly distributed, as would be expected if
observing random orbital positions. A random distribution of
spots should lead to an average of half the spots in one quadrant
of the P-V diagram and half in the opposite quadrant. In Fig. 6, the
distribution of spots is heavily weighted in favour of one quadrant in
at least 11 cases out of 36. This is necessary to achieve the lowest
possible derived central mass. If the central mass is located such
that approximately equal numbers of spots are found in each
quadrant, the derived central mass increases to unrealistic values
for O and B stars in these cases.


4.4.2 P-V diagrams with spots in a line

A slightly better constrained model is available, for example,
when the maser emission is found to be distributed along a line, in
the P-V diagrams. The model we adopt here is that the masers are
located on the edge of a thin ring, whose annular width is small
compared to its radius. The edge of the ring we see is where the
masing path length is longest. In this case the maser emission
kinematics will follow a straight line in the P-V diagram, with a
gradient given by
		       [ formula ]

where R is the radius of the ring. Since we do not know where the
central mass is located, again we can only derive lower limit
estimates of the mass of the central star. We have crudely derived
it by assuming that the radius of the ring is equal to the linear extent
of the maser spots (i.e. the annular thickness of the ring), which is
hardly consistent with the assumption that the ring is thin compared
to its radius. These are listed, with their corresponding P-V
diagrams, in Fig. 6. Again, it is apparent that the masses derived
do follow what would be expected of particularly massive stars.


Figure 6. Position-velocity diagrams of those 36 maser sites
     that show some evidence of a linear structure are shown.
     The horizontal axis is the offset in position
     along the major axis of the maser site, from the strongest maser spot.
     The vertical axis is the measured radial velocity.
     Each site has a lower limit central mass estimate (see text for details).
     Those sites where the central mass has been derived using a linear fit
     to the maser spots are shown with the linear fit drawn as the
     dotted line. Those sites where the mass has been determined by
     separating the maser spots into two quadrants are shown with the
     quadrant axes drawn on. The origin of these axes indicates the position
     and radial velocity of the central mass, chosen so that it takes the
     minimum possible value.
     Lower case letters (eg.`17016-4124a' and `17016-4124b') are used
     to refer to multiple maser sites detected within the same field.
     `a' refers to the maser site with the lowest maser spot
     velocity, `b' refers to the maser site with the second-lowest maser
     spot velocity, andso on.

  Since the mass is heavily dependent on the ring radius, the
derived masses will be too large to be realistic in all cases if the
ring radius is only 2.5 times the annular thickness.
  With the uncertainties in both these methods of mass estimation
through Keplerian motions, it is hard to categorically conclude that
the masers are not in Keplerian orbits about a central massive star.
Factors such as the position of the central mass need to be
determined before a rigorous test can be applied. Nevertheless,
the large values for some of the derived values of the mass make this
hypothesis seem unlikely.
  Our data do suggest that another interpretation is required to
explain the maser emission in many cases. We examine the case of
an expanding shock as an example in the next section.

4.4.3 Expanding shock model

There is no doubt that there are a number of maser sites that are
linear in extent which cannot be accounted for by chance align-
ments. A shock viewed from one side can produce such linear
structures readily. Norris et al. (1993) argued that a shock was an
unlikely explanation of their data, as it cannot produce the velocity
gradients seen in many of their maser sites. In our data set (see Fig.
2), it is found that most maser sites do not show a systematic
velocity gradient (only 12 out of 97 sites do have a velocity
gradient). Thus, we believe that a velocity gradient is not a general
feature of maser sites. This is partly the reason why the circum-
stellar disc hypothesis has difficulty in explaining all our data. It is
doubtful that velocity gradients have been smeared out in our short
cut integrations as we observe the same features as Norris et al.
(1993) in sites that were mapped in both programs (e.g. IRAS
13471-6120 and IRAS 14567-5846).
  The shock wave model naturally explains why many maser sites
are seen without associated radio continuum emission, as there is no
need for an ionizing source to be present. This implies that the
maser may also be associated with objects other than UC H II
regions.
  The shock wave hypothesis for spot locations does not require
them to be shock-excited, but rather that they are dense knots of
gas that have been compressed and accelerated by the passage of
the shock, with a sufficient column of material along our sight
line for masing to occur. The velocities of the masing spots
reflect the projection on our line of sight of the shock velocity.
For instance, for a spherically expanding 20 km s^-1 shock wave
1500 au from its origin, two spots 100 au apart on the plane of
the sky (0.5 arcsec at 2 kpc) would have a velocity difference of
5kms^-1, values quite typical of the data. Any spots in between
would show a smooth velocity gradient. In practice, of course,
the shock front is unlikely to expand uniformly unless it
traverses a homogeneous medium. Thus while velocity shifts
may reflect changing projection angle, they could also be
produced by varying shock speeds and a convoluted, or even
fractal, shock front. A closer analogy might be a wavy sheet.
Sight lines through the extrema of the sheet would be most
likely to exhibit maser emission due to increases both in column
of material and in velocity coherence.
   This is a general hypothesis and does not make specific predic-
tions for maser spots. Depending on local conditions, line, clusters

and isolated spots can be produced, and while velocities of adjacent
spots are likely to be correlated, they do not show systematic
   gradients. A test of this hypothesis would be to search for tracers
of shock waves associated with the masers, particularly through
near-infrared H2 vibrational-rotational emission, and high velocity
HCO mm line emission. In principle, the spatial and velocity
structure of the maser spots could then be used to model the
geometry of the expanding shock wave for individual sources,
and in particular to determine the degree of inhomogeneity of the
molecular cloud.


5 SUMMARY AND CONCLUSIONS

We have obtained high (arcsecond) resolution radio continuum (to a
limit of  1 mJy) and methanol maser (to a limit of  0.3 Jy) images
of a large number of UC H II candidates, using the ATCA. Contour
maps are provided for selected continuum sources, showing overall
morphologies of the regions, as well as relative position maps of
maser spots, with relative positional accuracy up to 0.05 arcsec. Our
major findings are as follows.

  (i) Most sites of methanol maser emission are not associated
with observable continuum emission, and most sites of continuum
emission show no signs of methanol maser emission.

  (ii) Continuum sources with associated maser emission tend to
be smaller than those which lack maser emission.
These first two points suggest that the methanol maser is observable
before the UC HII region phase and is probably destroyed as the UC
HII region develops. It is possible that a small fraction of these
maser sources are associated with embedded non-ionizing stars. We
predict that mid- to far-infrared sources should be associated with
these sites of maser emission as this radiation is required to explain
how the maser is pumped.

  (iii) The methanol maser flux is not dependent on the proximity
of a continuum source, whether the continuum lies offset from the
maser site or is projected behind it.

  (iv) A comparison of the positions of water and methanol maser
sites, with respect to differing morphologies of UC H II regions,
shows a marked difference. Methanol sites are clustered closer to
cometary-shaped continuum sources, whilst water masers are not.
Water masers, on the other hand, are found clustered towards
irregularly shaped UC H II regions whereas methanol masers are
not.

  (v) Despite differences in the association of methanol and water
masers with UC H II regions, there are some cases in which there is
good positional coincidence between the two maser types. Such
cases are associated with hot ammonia clumps, presumably a stage
of star formation before the UC H II region phase. Thus, the
environmental requirements for methanol and water maser
emission to be observable overlap, but are not the same.

  (vi) The hypothesis that methanol masers arise from circumstellar
discs is not inconsistent with some of our data. However,
in general it seems unlikely to account for the kinematics of
most of the masers that are found in a line or arc. The data is too
highly unconstrained to rigorously test the hypothesis, and
further information, such as the radial velocity and position of
the central stellar object, is required to further test this model.
Nevertheless, the minimum values derived for the central masses
in the most optimistic cases are very high, making the hypothesis
seem unlikely. We have provided an alternative model, suggesting that the
masers can be formed behind shocks, which adequately explains the
distribution of maser spots within an emission site.


ACKNOWLEDGMENTS

The authors would like to express gratitude to R. P. Norris for
helpful information on observing strategy and data reduction and
analysis, and to the ATNF for generous allocation of observing time
on the ATCA. This work is partially supported by an Australian
Research Council grant.


REFERENCES
Caswell J. L., Vaile R. A., Ellingsen S. P., Whiteoak J. B., Norris R. P.,
    1995a, MNRAS, 272, 96
Caswell J. L., Vaile R. A., Forster J. R., 1995b, MNRAS, 277, 210
Cesaroni R., Churchwell E., Hofner P., Walmsley C. M., Kurtz S., 1994,
    A&A, 288, 903
 Ellingsen S. P., Norris R. P., McCulloch P. M., 1996, MNRAS, 279, 101
 Forster J. R., Caswell J. L., 1989, A&A, 213, 339
 Gaume R. A., Mutel R. L., 1987, ApJS, 65, 193
 Ho P. T. P., Haschick A. D., Vogel S. N., Wright- M. C. H., 1983, ApJ, 265,
     295
 Hofner P., Churchwell E., 1996, A&AS, 120, 283
 Menten,K. M., 1991, ApJ, 380, L75
 Norris R. P., Whiteoak J. B., Caswell J. L., Wieringa M. H., Gough R. G.,
     1993, ApJ, 412, 222
 Ramesh B., Sridharan T. K., 1997, MNRAS, 284, 1001
 Sevenster M. N., Chapman J. M., Habing H. J., Killeen N. E. B., Lindqvist,
     M., 1997, A&AS, 122, 79
 van der Walt D. J., 1997, A&A, 322, 307
 Walsh A. J., Hyland A. R., Robinson G., Burton M. G., 1997, MNRAS, 291,
     261 (Paper 1)
 Wood D. O. S., Churchwell E., 1989a, ApJ, 340, 265 (WCa)
 Wood D. O. S., Churchwell E., 1989b, ApJS, 69, 831 (WCb)

 This paper has been typeset from
 LATEX file prepared by the author.

