4.1. Ammonia abundance for the regions I(N)w and I(N)e
The study of ammonia abundance is important because it provides observational constrains for the chemical models that try to explain how this molecule is formed in the interstellar medium, and it may be used as an age indicator of the molecular cloud (e.g., Suzuki et al. 1992). Although the chemical models calculate the evolution of the ammonia abundance as a function of time with a limited confidence, due mainly to the uncertainties concerning the formation of this molecule in the interstellar medium, they agree that the time necessary to reach the equilibrium abundance is approximately 106 years. If depletion effects are not consider in the model, the equilibrium abundance for the ammonia molecule is 10-7. If there is depletion, this value might be 10-8 (Bergin et al. 1995) or 10-9 (Hasegawa & Herbst 1993), even though in this last case the abundance reaches equilibrium after 108 years.
Comparing the ammonia abundances listed in Table 4 with the equilibrium abundance, we conclude that either the regions I(N)w and I(N)e have an age lower than 106 years, or that ammonia depletion is important in both sources. This result was also obtained by Kuiper et al. (1995), assuming LTE conditions. Another relevant aspect that must be taken into account is the dependence of the equilibrium abundance on the nitrogen abundance in the interstellar medium, and therefore the gradient of nitrogen with galactocentric distance must be also considered (Vilas-Boas & Abraham 2000).
4.2. Comparison with LTE models
In this sub-section, we will derive, under assumptions of LTE, the physical parameters of the source I(N)w and compare them with the parameters derived in Sect. 3 under non-LTE conditions. We calculate these parameters at the (0´,0´) position because it is the only point in which the NH3(J,K) = (2,2) transition was observed. In the LTE calculation we considered a three-level system, in order to obtain the rotational and kinetic temperatures ( and respectively) through the equations (Ho & Townes 1983; Walmsley & Ungerechts 1983):
where and are respectively the optical depths in the transitions NH3(J,K) = (1,1) and (2,2), C(22 21) and C(22 11) are the collision rate coefficients given by Danby et al. (1988).
To determine the optical depth, we assumed that the cloud has a spherical shape, that the excitation temperature () is the same in all hyperfine lines of the NH3(J,K) = (1,1) spectrum (Ho & Townes 1983). Thus, using the ratio between the (1,1) main line and its inner hyperfine satellites, can be calculated from:
where and are, respectively, the optical depth for the main line and the inner satellites, with = 0.28 (Ho & Townes 1983) and (SW85). To obtain the optical depth for the NH3(J,K) = (2,2) transition, it is necessary to substitute in (12) by , the antenna temperature of the (2,2) main line.
As we have less equations than unknown parameters, it is necessary to make some assumptions to solve the system. There are at least two clear alternatives: we may assume that either the excitation and rotational temperatures have the same value, or that the source fills completely the antenna beam. Here, we adopted the first alternative since high resolution observations indicate filling factors smaller than one for these sources (Kuiper et al. 1995; Kraemer & Jackson 1999; Megeath & Tieftrunk 1999).
We present in Table 5 the physical parameters obtained under LTE and non-LTE conditions. The NH3 column density and the ammonia abundance are not sensitive to LTE departures. The largest discrepancies are found in the kinetic and rotational temperatures. In fact, the kinetic temperature estimated under LTE conditions is about twice the value obtained in non-LTE. The optical depth of I(N)w, in both transitions, is slightly smaller than what it was found in previous studies (e.g., Forster et al. 1987; Vilas-Boas et al. 1988; Kuiper et al. 1995). We believe that the differences can be attributed to the lower signal-to-noise ratio in previous observations. The beam filling factor is larger under LTE conditions. This was already expected because under non-LTE conditions the gas is confined in clumps and not spread in a homogeneous cloud.
Table 5. Comparison between the physical parameters for I(N)w, calculated in LTE and non-LTE conditions.
4.3. Evolutionary state of the regions
We will try to determine now the evolutionary state of the three regions identified in our ammonia observations. A possible way to explore this question is comparing the distribution of indicators of early stages of star formation. In Fig. 3, we show the distribution of infrared sources, detected in the JHK bands by Tapia et al. (1996) in a small region of the mapped area. We can see a decrease in the number of infrared sources towards the region I(N)w and also the existence of a cluster of sources around NGC 6334 I, indicating that it is more evolved than I(N)w. Other observational results favorable to this scenario are the detection of an UCHII region and the identification of a molecular bipolar outflow associated to this source (Rodríguez et al. 1982; Bachiller & Cernicharo 1990). There are no infrared observations towards I(N)e, to indicate its evolutionary stage.
Our observations, together with the non-LTE model (SW85), show that the physical properties of the clumps in I(N)w and I(N)e are similar. However, the volume density of clumps in I(N)w ( 6000 pc-3) is at least a hundred times higher than in I(N)e, resulting in a collision rate which is hundred times higher in I(N)w than I(N)e. Thus, assuming that collisions result in coagulation of the fragments, which finally collapse to form stars after several collision times (SW85; Scalo & Pumphrey 1982), we expect that I(N)e is in a earlier star formation stage than I(N)w.
The evolution scenario proposed by SW85 and the results presented in this work indicate that the regions I(N)w and I(N)e could form in the future star clusters, similar to those detected in NGC 6334 I.
© European Southern Observatory (ESO) 2000
Online publication: October 2, 2000