H2O masers were found in SFRs since the earliest spectral line
surveys (Genzel & Downes 1977) and were recognized to be
closely related with young massive stars. More
recently Palla et al. (1993) and Codella et al. (1994) suggested that water
maser activity is indeed
present in the earliest evolutionary phases of a high mass (proto-)star,
even before the onset of an UCHII region. Nevertheless, it was clear that
the comparison of the low resolution data (e.g. single dish maser
observations, radio continuum observations of the free-free emission from
HII regions and IRAS observations of cool dust clouds around luminous stars)
could give but a general
indication. Only higher resolution (
)
observations can enable one to disentangle the complexity of
high mass SFRs and allow one to search for the stellar
source directly connected to the maser. In fact, according
to the shock excitation model of Elitzur et al. (1989), the
exciting stellar source should be located very close to the maser (
104 AU or 10
at 1 kpc). Consequently, arcmin
coincidences have very little significance.
On the radio side of the spectrum the NRAO Very Large Array (VLA) offers the possibility of obtaining sub-arcsec resolutions for the water masers and for the radio continuum emission from UCHII regions (Forster & Caswell 1989, hereafter FC89; Tofani et al. 1995; Jenness et al. 1995; Hofner & Churchwell 1996). One of the most important results of these studies has been that H2O\ masers and UCHII regions, although generally found in the same SFR, are not necessarely closely related to each other, and that the powering sources of the free-free continuum and of the maser radiation have to be searched in distinct objects, most probably YSOs belonging to the same SFR but in different evolutionary phases.
Consequently, to identify the stellar source powering the maser, arcsecond resolution observations are required, in particular in the far infrared and submillimeter regions of the spectrum. In fact, it is in these wavebands that the embedded YSOs are expected to emit most of their energy. Although some attempts in this direction have been carried out (Jenness et al. 1995), the current instrumentation does not allow one to attain the required resolution and sensitivity in these spectral ranges.
Another powerful probe that became available in the past few years is molecular line interferometry at centimeter and millimeter wavelengths (Cesaroni et al. 1994; Cesaroni et al. 1997; Codella et al. 1997; Hofner et al. 1996; Olmi et al. 1996) which enables arcsecond resolution of hot molecular clumps and cold dust clouds surrounding the water masers. These studies, focused on a few selected sources, confirm that the water masers are indeed associated with hot molecular clumps and millimeter continuum sources, and that the expected luminosities of the objects embedded inside these cores is that of high mass stars. However, this method is not feasible for large surveys.
In order to search for the stellar sources physically associated with the masers we decided to undertake a systematic NIR imaging survey of a sample of SFRs containing masers. This is motivated by the fact that the modern NIR array detectors offer the required sensitivity and resolution even with small telescopes and with short integration times. We were also guided by the expectation that in its early evolutionary stages high mass stars could be detectable at NIR wavelengths even if still deeply embedded in a dust cloud, due to the emission of hot dust around the star (Testi et al. 1997). Previous attempts to look for near infrared sources close to H2O masers had been carried out with single element photometers and large diaphragms (Evans et al. 1979; Moorwood & Salinari 1981a,b; Epchtein & Lépine 1981; Braz & Epchtein 1982) and could not reach the sensitivities and resolutions needed to separate the various YSOs in each region.
In this paper we present NIR observations with arcsec resolution of a sample of H2O/OH masers in SFRs taken from the list of FC89. The complete radio data were kindly provided by R. Forster (1992 private communication).
The SFR type of the masers in our sample (to distinguish them from
those around late type stars) relies on the selection criteria used
by FC89, basically the presence of a thermal continum radio source
(an HII region) close to the maser. As an additional check,
we have derived for those fields that contain an IRAS point source
(27 out of 31)
the (25 
m/12 m) and the (60 m/12 m) colours.
All the points fall in the box of the (25
m/12 m)- (60 m/12 m) diagram that defines the locus
of UCHII regions (Wood & Churchwell 1989).
This is a further confirmation of the star forming nature of the
selected fields, even though, strictly speaking, it is not directly
related to the nature of the star associated with the maser itself.
In Testi et al. (1994) (hereafter Paper I) we
have presented the results obtained toward a subsample of 17 SFRs,
which is in fact the first attempt to compare maser and NIR observations at
arcsec resolution.
Assuming that a NIR source is physically
associated with the maser if it lies within
from it, we were
able to identify a NIR source in 
of the cases.
In 
the NIR source was closer than 
. In this
paper we present new higher resolution and higher sensitivity observations of
31 SFRs from the list of FC89 (5 regions overlap with that presented in
Paper I and are indicated with an 
in Table 1 (click here)).
Three of the regions in the present sample (G351.41+0.64; G009.62+0.19;
G035.20-1.73) have been studied also with
high resolution observations at other frequencies
(Persi et al. 1996; Testi et al. 1997
and Persi et al. 1997).
In all these cases the physical association of the selected NIR source
with the maser is amply confirmed by independent criteria.
In Sect. 2 (click here) the observational parameters and the data
reduction procedures are summarized. In Sect. 3 (click here) the results are
shown. In Sect. 4 (click here) we show that close positional coincidence alone
is not sufficient to establish a physical association, and that
the NIR colour is the complementary information. In fact, the NIR sources
associated with the masers all show a strong NIR excess,
while the background NIR sources are distributed along the
reddening line.
In Sect. 5 (click here) the main conclusions are summarized.