4.1. NH3 in the direction of IRAS sources
Twelve of the 108 HII regions observed have associated IRAS sources with color indices of ultra compact HII regions. We consider that an IRAS source is associated with the HII region when it is displaced by less than the telescope half power beam width from the observed position. These sources are listed in Table 2 where Columns (1) and (2) give the names of the ammonia and IRAS sources, (3) and (4) give the equatorial coordinates of the IRAS sources, (5) the distance in arc minutes between the observed positions and the IRAS sources, Column (6) through (9) give the IRAS fluxes, and finally Column (10) the luminosity obtained integrating the four uncorrected IRAS color bands. Attached to the flux density for each IRAS band is given the flux quality. It is remarkable that 7 of these positions have positive ammonia detection, while in the other 5 there is evidence of weak emissions. These results show that there is a high detection rate of NH3 when the observed position has an associated IRAS point source with color index of ultra-compact HII region. This detection rate is compatible with the rate of 70 % observed by Churchwell et al. (1990) toward northern hemisphere ultra-compact HII regions using the Effelsberg radio telescope.
Table 2. Compact IRAS Sources with Color Index of Ultra-Compact HII Regions.
4.2. Distribution of NH3 sources in Galactic longitude
The distribution in Galactic longitude of the HII regions and the detected NH3 sources are similar, as can be seen in Fig. 1. However, the distribution of HII regions presents two peaks, one between - with 36 sources and the other between - with 13 sources. The NH3 distribution also has a peak at -, but does not present any source at the position of the second peak. Moreover, the region between - has only two ammonia sources detected, compared with the 20 observed HII regions. Kinematic distances indicate that the two sources are in the solar neighborhood. To test the statistical significance of these results, we applied the Kolmogorov-Smirnov test to the samples. Using the complete distribution, we found an 88 % probability of both samples having the same distribution, while the probability increased to 99.9 % when only longitudes larger than were considered. This result confirms the importance of the absence of NH3 sources in the interval -. This range of longitudes is located between the local Cygnus arm and the Perseus arm, at a distance from the Sun of more than 3 kpc. The absence of ammonia in this direction could be explained if the observed HII regions were located at larger distances from the Sun than others, but this in not the case. In fact, the sources G287.3-0.7 and 287.9-0.8, which are associated with the Carina Nebula, are located at about 2.7 kpc from the Sun and G291.3-0.7 has a kinetic distance between 2.2 and 3.5 kpc. These are typical distances of sources seen in other directions.
The absence of NH3 sources could also be explained if the HII regions were in a more advanced evolutionary state or if they were mass limited, with the molecular clouds surrounding them already dissociated. However, the existence of dense molecular gas observed in this longitude interval through formaldehyde absorption, and CO(1-0), HCO+(1-0) and CS(2-1) emission lines, seem to invalidate this argument (Whiteoak & Gardner 1974, Wouterloot & Brand 1989, Batchelor et al. 1981, Bronfman et al. 1996). Another possibility is that the physical conditions for the excitation of the ammonia molecule are no longer present, even if the molecule does exist. This argument can be discarded when we look at other molecules, like HCO+ and CS which are excited under physical conditions similar to the metastable ammonia lines.
4.3. Distribution of NH3 sources with distance to the Galactic center (Fig. 3)
The parameter which differentiates the samples of HII and NH3 sources at galactic longitudes between - and - is the distance to the Galactic Center. The region with no NH3 emission, between and , is located far from the center, near or outside the solar orbit. Histograms of the distributions of the selected HII regions and detected ammonia sources as a function of distance to the Galactic Center are shown in Fig. 2. The graph of the observed HII regions shows an almost symmetric distribution with a peak at 6.5 kpc. The ammonia distribution has a peak at the same position but presents a strong asymmetry with a small number of ammonia sources at distances larger than 6.5 kpc. To obtain the statistical significance of this asymmetry we plotted the ratio of the number of ammonia sources to HII regions as a function of the distance to the Galactic center. A very well defined linear trend is found, with a slope of -0.08 kpc-1 and a correlation coefficient . If, as discussed above, there is dense molecular gas associated with these HII regions and the physical conditions are favorable to NH3 emission, then the decrease in the number of detected ammonia sources could be due to a decrease in their brightness temperature, which puts them below the detectability limit of our radiotelescope. The decrease in brightness temperature could be due to a decrease in the ammonia abundance with the distance to the galactic center, as a consequence of the gradient in the nitrogen abundance. In fact, N and O abundance gradients of about -0.08 dex kpc-1 have been identified in HII regions (Shaver et al. 1983; Simpson et al. 1995; Afflerbach et al. 1997), and in type II planetary nebulae (Maciel & Chiappini 1994).
4.4. The distribution of NH3 sources with brightness temperatures
The gradient of nitrogen abundance, the detection rates and the distribution of NH3 sources with brightness temperature were used to determine the distribution of sources with H2 column density. Several assumptions were made: (a) the brightness temperature of the source is proportional to the NH3 abundance. This is valid for low optical depth, which is a good assumption since for more than 60 ammonia sources observed toward HII regions (MacDonald et al. 1981, Vilas-Boas et al. 1988, Churchwell et al. 1990) the average value of is around unity, (b) the NH3 abundance is proportional to the nitrogen abundance, true if the regions have similar H2 and metal abundances (Graedel et al. 1982). The effects of other parameters, as cloud age, time to reach equilibrium, UV illumination, shocks and dust mantle destruction are supposed to be averaged, since we observe molecular clouds associated with HII regions in different stages of evolution. (c) the angular size of the molecular clouds where the ammonia sources are located is large compared with the radiotelescope beam, the consequences of this assumption will be discussed later, (d) the filling factor (fraction of the beam covered by the NH3 sources) is the same for the whole cloud. This is true if the ammonia sources are formed by small high density clumps, uniformly distributed in the molecular cloud. Evidence in favor of this assumption are given by Stuzki & Winnewisser (1985), who showed that in order to explain the anomalies in the intensity of the hyperfine satellite inversion lines, it is necessary to assume that the NH3 sources are formed by a large number of independent clumps with sizes of the order of 10-2 pc and masses of about a solar mass, (e) the kinetic temperature of each clump (or its distribution) is independent of the galactocentric distance. Although a gradient in kinetic temperature is observed in HII regions, where the main coolant is ionized oxygen, it is not expected in molecular clouds with temperatures smaller than 50 K, where the coolant is CO, which is optically thick in these dense regions. Assumptions (c) and (d) justify the subsequent use of measured antenna temperature T instead of brightness temperature.
Let us define as the fraction of the detected sources with temperature between T and at distance R from the Galactic Center. The total fraction of sources detected at distance R will be:
where is given by the detection limit of the radiotelescope and is the maximum temperature of the sources (the excitation temperature in the limit of an optically thick source). Since we assumed that the brightness temperature is proportional to the nitrogen abundance, we can write:
where is some reference distance at which the source temperature is and dex kpc-1 is the N gradient in the Galaxy (Maciel & Chiappini 1994).
Also, since we have assumed that the only difference between the molecular clouds at different distances is their NH3 content, the sources at a distance R and temperatures between T and will have temperatures between and at the distance . Therefore we can write:
Eq. (1) becomes:
where now the minimum temperature is a function of R, given by:
The slope of will be:
From (4), (5) and (6) we obtain:
Comparing this expression with the best fit to our data in Fig. 3, , we conclude that should vary as , with . If the angular sizes of the molecular clouds are smaller than the radiotelescope beam size, the filling factors will decrease with distance to the observer. In this case, the real number of ammonia sources should be larger than the detected number at larger distances to the observer. The mean distance of the sources to the observer, on the other hand, increases as the galactocentric distance decreases, as can be seen in Fig. 4, for this reason we expect .
Under the assumptions listed at the beginning of this section, the brightness temperature reflects the NH3 column density and, after corrections for the gradient in the abundance of this molecule with the galactocentric distance, it also represents the H2 column density. Therefore, the distribution indicates that the fraction of molecular clouds associated to HII regions decrease as the H2 column density increases. Since the actual sizes of the molecular clouds are not known, we cannot convert H2 column density to mass, but it is possible that the distribution of with T is a consequence of some relation , where M is the mass of the cloud. Relations of this type are found in the distribution of clumps in Oph (Motte et al. 1998) and also in other molecular clouds (Loren 1989, Blitz 1993), with varying between 1 and 2.5.
Another quantity which can be calculated once the distribution of sources with T is known is the mean brightness temperature:
where we have assumed . Using Eq. (2) we obtain:
The maximum observed temperature is 1.5 K for NGC6334, the minimum temperature can be taken as 0.1 K, three times the rms of the observations, using these values and in Eq. (9) we obtain
In Fig. 4 we present the calculated mean temperature for four interval bins in galactocentric distance. The three points at the largest distance to the Galactic Center define a line with slope 0.26, in agreement with the value derived in Eq. (9), however, the point closest to the Center falls well below this line. This behavior would be expected if the assumption that the sources cover completely the antenna beam is valid up to distances to the observer of about 3 kpc, as can also be seen from the figure. At this distance the antenna beam will correspond to a linear size of about 3.5 pc. Actual sizes of molecular clouds associated with HII regions, obtained by mapping, are available for only a few sources. For other ammonia sources, specially those observed towards compact HII regions, only one position was studied and the size of the region calculated from the filling factor, under the assumption of LTE. However, as mentioned before, Stutzki and Winnewisser (1985) showed that a large fraction of warm ammonia sources present anomalous intensities in the satellite lines, probably caused by the superposition of small dense clouds in non LTE. The filling factor, in this case, is interpreted as the ratio between the solid angle occupied by all the clumps and the solid angle of the antenna beam. This interpretation is different from the usual assumption that the size of the molecular cloud is the product of the antenna beam size and the filling factor. Therefore, molecular clouds can be larger than the beam size and still have filling factors smaller than one.
© European Southern Observatory (ESO) 2000
Online publication: March 21, 2000