3.1. J-H vs H-K diagram
In an area of approximately 2 2 square arcmin centered around the OH/H2 O masers, we have detected 178 sources at K, of which 121 have been also observed in J and H filters. From their measured magnitudes, we have constructed the versus diagram reported in Fig.3.
According to the location of the sources in this plot, we can distinguish field stars from sources belonging to the stellar cluster. In fact, foreground and background stars are located along the main-sequence line (MS in Fig.3) or along the reddening line, while candidate young stellar objects, with an IR excess due to a circumstellar dust envelope, are those to the right of the reddening line. However, we note that as much as 50% of possible TTauri stars may still be located on the strip of the reddened main sequence (Aspin & Barsony 1994). In addition, several objects have colors of reddened early type stars with = 15-25, in agreement with the derived visual extinction. Therefore, the analysis of the color-color diagram, suggests the presence in G 35.20-1.74 of a conspicuous number of YSOs and early-type stars possibly related to the same star-forming region.
3.2. Infrared cluster and the K-luminosity function
The other tool to study the physical association of IR sources belonging to the same complex is to examine their spatial distribution. This is reported in Fig.4.
For comparison, in this figure we shown the spatial distributions of all the sources detected in K, of those detected only in K, of those with infrared excess, and those without IR excess as derived from the two colors plot of Fig.3. We have included in the IR excess sources also those with 2 but not detected in J. In Fig.4 the positions of the UC HII region (open triangle), of the OH (filled triangle) and of the two H2 O (filled square) masers are also illustrated. As shown in the figure, most of the sources with IR excess are concentrated in an area of approximately 20 in radius around the position of the UC HII region and inside the extended IR nebula, while the objects classified as field stars (no IR excess) lie mainly in the NW part of the region. Therefore, Fig.4 shows the presence of a young stellar cluster in G 35.20-1.74, composed by at least of 21 embedded objects with circumstellar dust, and other six reddened early type-stars, all within the boundary of the diffuse K emission. Figure 4 also shows that the extinction towards the stellar cluster is larger than in the surrounding regions, since in this area (as well as in that of the molecular cloud, see Sect.3.6) the density of background stars is lower. Similar embedded young stellar clusters have been found associated to others star forming complexes such as M17 (Lada et al. 1991), NGC 3567 (Persi et al. 1994) and LkH 101 (Aspin & Barsony 1994). Table I gives the position and the near-IR photometry of the objects with IR excess and with 2 but no detection in J found in G 35.20-1.74. In the same table two additional sources are given: that associated with HC1 (and MIR1), which is detected only at K, and that associated with HC2.
To study the nature of these sources, we have compared in Fig.5, the K magnitude distribution of all sources (K-luminosity function (KLF)) with that of the sources with and without IR excess, and sources detected only in K.
The total KLF peaks at K=15.5 mag, while the histogram of the cluster members (IR excess stars), is completely different from the other objects, and peaks at K=13.5 mag ( -1.4 if =23) suggesting that the young embedded cluster has a high percentage of bright early type stars (spectral types between B2-B3 if = 23).
3.3. Infrared energy distribution
All the mid-IR sources have been detected at K. MIR1 and MIR3 correspond to the H2 O maser HC1 and the UC HII region respectively. These sources will be discussed separately in the next sections. Combining the near-IR and the 11.2 flux densities, we have derived their infrared energy distribution illustrated in Fig.6.
The sources MIR2, 4 and 5 show a very steep energy distribution with a clear IR excess, also evident in the two colors plot of Fig.3, while the source MIR6 has a spectrum typical of a reddened early-type stars without IR excess. In addition MIR4, 5 are located within the diffuse nebulosity observed in the near and mid- infrared, and belong to the observed young embedded cluster discussed in Sect.3. 2.
The spectral slope between 2.2 and 11.2, defined as
is often used to classify the evolutionary state of a YSO (Lada 1987; Andrè & Montmerle 1994). Negative values of are those of class I sources and represent the earliest stages of evolution, when the protostar is still deeply surrounded by a thick dust envelope, class II objects are those with 0 2 and represent pre-main sequence stars surrounded by an optically thick disk, and class III objects () are the more evolved and surrounded by an optically thin disk.
In our case MIR1, MIR3, MIR4 and MIR5 have equal to -2.5, -2.6, -2.8 and -2.4, respectively, and can be classified as extreme class I objects. MIR2, with = -1.0 can still be considered as a class I object, while MIR6 with = 0.0 is a transition object between class I and class II.
3.4. The UC HII region
The morphology of the UC HII region is best described by the 2 and 6 cm radio continuum maps of WC. The structure is classified as "cometary", but the possibility of a blister type configuration has also been considered. This last hypothesis was reputed less appealing because the bright sharp edge of the radio emission is on the NE side, while the molecular cloud (see Sect.3.6 and Fig.9) extends to the SW of the UC HII region. However, none of the available molecular maps has the required resolution (2-3 ) necessary to study the molecular gas distribution (in high density tracers) on the NE side of the bright radio continuum edge, so the point cannot be settled for the moment.
Using the original (u,v) data of the VLA observations of WC, larger field maps at 2 and 6 cm were made. The overlay on the K image of the 6 cm map is shown in Fig.7. This shows that there is a very good agreement between the K and radio emission. A similar comparison with the 11.2 image also shows a very good coincidence between the two emissions.
The UC HII region radio continuum flux density at 5 GHz is 1.93 Jy. In the following we shall compare this value with the fluxes measured in the NIR and MIR, with the provision that this is the most difficult part of the spectrum, since the photosphere of an early type star, the ionized gas and the dust can give comparable contributions at K (see e.g the models of Natta & Panagia 1976). The expected K emission from the ionized gas (with no correction for extinction) can be obtained from the relation (Jy) = 0.26 S (Jy) (Howard et al. 1994), and is 500 mJy. The measured flux density at K in a 14 diaphragm is (see Table 1) 81 mJy. For = 23, the expected K emission becomes 75 mJy. As far as the photospheric stellar contribution is concerned, the spectral type of the early type star ionizing the UC HII region is (from radio data) O7.5 (WC). Using the stellar parameters tabulated by Panagia (1973) for ZAMS stars, the expected stellar photosphere contribution, after correction for extinction, is 23 mJy. Given the uncertainties, the sum of the two is in very good agreement with the observed value and shows that the dominant contribution comes from the free-free and free-bound emission of the ionized gas. The agreement between the extended radio and K emission also favors gas emission rather than stellar emission. There does not seem to be the need of reflected emission and intrinsic dust emission should be negligible, excluding the presence of very hot dust within the UC HII region.
Table 1. Coordinates and IR photometry of the sources associated with the molecular cloud.
At J the comparison changes in favor of a larger stellar contribution, since the expected stellar emission is 1.8 mJy, to be compared with an observed value (in a 14 aperture) of 2 mJy.
At 11.2 , the situation is reversed, the ionized gas and the star contribute at most few percent to the observed 27.8 Jy. The emission must come from warm dust. In fact, an increase of at least an order of magnitude from K to 11.2 due to dust emission is expected from model calculations (Natta & Panagia 1976). The good agreement between 11.2 and radio maps shows that dust is well mixed with the ionized gas inside the UC HII region. However, it is difficult with only the 11.2 flux to estimate the temperature of the dust within the UC HII region. In fact, given the already mentioned difference between the 12 IRAS flux and the 11.2 flux, no extrapolation to longer wavelengths using IRAS data is feasible for the UC HII region.
WC noticed that the total luminosity derived from the integral of the IRAS fluxes is larger (Log( / ) = 5.45) than that derived from the spectral type of the ionizing star, estimated from the radio continuum (for an O7.5 Log( / ) = 4.92). This effect was found to be consistently present over the large sample of UC HII regions examined by WC. Two possible solutions were examined by WC: 1) strong dust absorption of uv photons inside the UC HII region, which underestimates the spectral type of the exciting star and 2) presence of a stellar cluster with a large population of luminous, but non-ionizing stars, which may substantially contribute to the total luminosity. In view of our new observations we favor the second hypothesis. In fact, for the densities present in the UC HII region (the peak density is 8.6 104 cm-3), dust absorption of uv photons should be small (Felli, 1979), while a cluster of luminous stars is found in the NIR observations.
3.5. The H2 O masers
The new VLA radio maps from the WC observations were primarily made to see if there is any radio emission from the position of HC1 MIR1 (which is outside the boundary of the maps published by WC). No radio continuum emission was found at the two wavelengths, with upper limits (5 ) of 3 and 10 mJy at 6 and 2 cm, respectively (see also Fig.7).
In Fig.8 a blown-up image of the HC1 source at K is shown. The new more accurate positions of the H2 O maser (WC) and the present better astrometry at K now show that the maser is located almost on top of the NIR source. This last one is clearly non-stellar due to a tail extending to the south-west. The point source in Fig.8 to the north-east of HC1 is a foreground star, since it is detected in all three bands and does not have a color excess, while that at the south-west end of the tail has been detected only in K, similarly to HC1.
The observed color index (H - K) of HC1 is 4.7, i.e. the largest of all the sources in the field. The flux density at K is 5 mJy. The expected 5 GHz emission, in the case that the K emission comes from free-free and free-bound emission of ionized gas, is 20 mJy, if we do not correct for reddening, and 135 mJy if we use = 23, as for the neighboring UC HII region. In any case, the predicted flux is much larger the the VLA upper limit. This does not automatically excludes that there could be an HII region, but implies that, if present, it is strongly self absorbed in the radio continuum.
In the following we shall consider the possibility that the K and 11.2 emissions come entirely from a hot dust envelope around a YSO.
The color temperature between K and 11.2 is 500 K, without correcting for reddening, or 610 K, if we correct the K emission for = 23. A black-body that produces the observed flux densities should have a size of 9.7 - 7.3 AU and a luminosity of 240 - 300 , in the two cases. Dust temperatures of several hundreds degrees can be reached only in the immediate surrounding of a YSO, at a radius consistent with the previous size. The implied luminosities could be produced by a pre-main sequence star inside a dust envelope, but could also be obtained by accretion during the protostellar phase, without requiring the presence of an already formed star; in fact Palla & Stahler (1993) predict accretion luminosities up to 10 for mass accretions of 10 /year and stellar masses between 2-6 .
In summary, all the present evidences suggest that HC1 could be a YSO still in the accretion phase embedded in a hot dust envelope. The coincidence with the water maser confirms that the maser emission occurs in the earliest evolutionary phases of a YSO.
Other objects with characteristics similar to those of HC1 have already been found. The lack of a radio continuum counterpart associated with H2 O masers has been pointed out in many SFRs by Tofani et al. 1995. The prototype of this class of objects is the H2 O maser in W3OH (Turner & Welch 1984, Wink et al. 1994). An another example could be IRAS20126+4104 (Cesaroni et al. 1997) and component F in G9.62+0.19 (Hofner et al. 1996). In conclusion, as the resolution of the observation increases, there are more evidences that the maser emission occurs in the life of a YSO in phases preceding that of the UC HII region.
The distance of HC1 from the UC HII region (the projected value is 0.3 pc) seems to exclude any form of direct interaction between the two. Any agent traveling at the sound speed ( 1 km s-1) in the molecular cloud would take 3 105 years to cover that distance.
Not much can be said about this other maser. It is outside the field covered at 11.2 and also far from the field centre of the VLA observations. The closest NIR source (#1 in Table 1) is distant from the maser position by 2.6 . It is detected in all three bands and its color indices show an IR excess, even though not as large as that of HC1. It is located within the western boundary of the molecular cloud/sub-mm source (see Fig.9).
3.6. Overall morphology of the G 35.20-1.74 SFR
The peak of the 850 emission is located between the UC HII region and HC1. With the present 850 resolution it is not clear if this due to a true peak of cool dust emission between the two sources or, more simply, the sum of two independent contributions, one from the UC HII and one from HC1.
The diffuse HII region east of the UC one is outside both the 850 contours and the C18 O contours, indicating that dust and molecular gas have been destroyed by the radiation of the early type stars of the cluster.
The 850 and C18 O contours show a similar elongation in the east-west direction, and include the other maser HC2.
The morphology of the entire complex suggests that star formation has occurred independently in several places of the molecular cloud. The oldest event in time is that related to the stellar cluster and the diffuse HII region east of the molecular complex, the UC HII region may represent an younger event and the two masers the most recent ones.
© European Southern Observatory (ESO) 1997
Online publication: April 8, 1998