4. Interpretation and discussion
From Sect. 3 one can see that the molecular surroundings of IRAS 20126+4104 present intriguing kinematical characteristics, which change depending on the tracer observed. Now, we give an interpretation capable of accounting for all the features observed in a consistent manner.
In the following, we shall assume a distance to IRAS 20126+4104 of 1.7 kpc, as done by other authors (see WBM and references therein). This value might be in error by a factor 2.5, because the kinematic distance for a local standard of rest (LSR) velocity of km s-1 is about 4.2 kpc: however, this is highly uncertain as a peculiar velocity of only 5 km s-1 relative to the assumed rotation curve would imply a strong variation of the distance; moreover a distance of 4.2 kpc corresponds to a height of 265 pc above the galactic plane, which is very large for a young star forming region. We conclude that 4.2 kpc can be taken as an upper limit on the distance.
4.1. The HCO flow
As shown in Figs. 4 and 7, the molecular gas seen in the HCO (1-0) line presents a peculiar SE-NW bipolar structure with a velocity reversal at km s-1 and km s-1 respectively in the blue- and red-shifted gas. We believe that this can be explained if we are looking at an almost edge-on bipolar outflow, i.e. with , where i is the angle between the outflow axis and the line of sight. In fact, as discussed by Cabrit & Bertout (1986) (see their Fig. 1 and Table 1), if the opening angle, , of the cone representing the outflow is such that , then either lobe of the outflow is associated with both blue- and red-shifted gas. This is because the cone representing the outflow intersects the plane of the sky passing through the centre of the outflow itself. Then, if i is not exactly equal to , the velocity of the gas on one side of such plane (e.g. towards the observer) will be on the average higher than that on the opposite side (away from the observer): thus, one side would contribute to the inner wings, the other to the outer wings of the line. Examples of this are shown in Fig. 6 of Cabrit & Bertout (1986), where is assumed: their maps look qualitatively similar to what one might obtain by superimposing the maps in the top and bottom panels of our Fig. 7. Note that in this model the outflow semi-axis pointing away from the observer lies at SE, whereas that approaching the observer is at NW.
We conclude that the HCO (1-0) emission traces a bipolar molecular outflow with an axis slightly inclined with respect to the plane of the sky.
As noted in Sect. 3.1, the previous conclusions are inconsistent with the N-S velocity shift seen in the 13 CO(2-1) line by us and - on a larger scale - by WBM, who mapped the same region also in the 12 CO(2-1) line (see their Fig. 2). Such a disagreement can be explained by the fact that molecules like 12 CO or 13 CO sample more extended and lower density regions than CS or HCO and hence may also trace different velocity fields. Therefore, if two outflows do exist, then the CO lines give a better representation of the large scale outflow and of its mean physical parameters (temperature, from 12 CO, and mass, from 13 CO), whereas our maps in the CS and HCO transitions are much more suitable for studying the kinematics of the outflow on a smaller scale associated with IRAS 20126+4104 and the H2 O masers. Note that the existence of two outflows in the same region has been hypothesised also by other authors in analogous cases (see e.g. Ladd & Hodapp 1997).
4.1.1. Time scale and physical parameters of the flow
At the end of the previous section we have hypothesised the existence of two outflows seen on different scales in the same region. However, in the following we shall ignore the large scale flow seen by WBM and concentrate on the small scale flow, seen by us in HCO . We thus assume that also the 13 CO emission mapped by us with the 30-m telescope on a small region () traces the same outflow seen in HCO . Under such assumption, one can estimate the mass in the outflow by using the 13 CO(2-1) or HCO (1-0) line. The former is a better probe for mass estimates, because the value of the 13 CO abundance is known reasonably well. However, we have just pointed out that 13 CO is a low density probe suitable for sampling larger scales than those we wish to investigate in this paper; moreover, with the 30-m telescope we have mapped only part () of the 13 CO emitting region, as evidenced by the larger maps of WBM, which reveal emission on a scale of at least . In this respect, HCO is to be preferred because less extended in size, although, in turn, it is affected by great uncertainty on the value of its abundance, which can easily vary by an order of magnitude.
We made an estimate using both lines. For this purpose, we integrated our maps over the whole region mapped with the 30-m telescope, in the velocity ranges from -28 to -5 km s-1 and from -2 to 21 km s-1. The results obtained from the 13 CO(2-1) and HCO (1-0) maps are listed in Table 5. They have been derived under the hypothesis of optically thin emission, which seems reasonable for the line wings, and assuming [13 CO/H2 ]= , [HCO /H2 ] , and =30 K for both molecules; the latter corresponds to the peak of the 13 CO(2-1) line.
Table 5. Mass estimates for molecular outflow
The mass estimates derived from the two molecular species differ by about a factor 3-4, which is well within the uncertainties mentioned above. In the following we shall use the mass estimates from HCO , which seem more appropriate for deriving the physical parameters of the flow.
Estimates of the HCO outflow time scale, mass loss rate, momentum, and mechanical luminosity are given in Table 6.
Table 6. Physical parameters of HCO outflow
The time scale of the outflow can be estimated from the size measured from the PdBI maps (Fig. 7) and the maximum velocity at which emission is detected; the mass loss rate is derived from the values of Table 5 and the time scale; the momentum has been obtained by summing the contribution at each velocity in the flow (i.e. of each channel in our channel maps in the HCO line); similarly, the mechanical luminosity has been obtained by summing the kinetic energy at each velocity and dividing the result by the time scale. Note that the mass loss rate in Table 6 refers to the outflow, to be distinguished from the mass loss rate, , of the stellar object powering the outflow itself. The latter can be evaluated assuming momentum conservation and e.g. a stellar wind velocity of 200 km s-1: under this assumptions one obtains yrs-1.
We stress that the time scale in Table 6 is very likely an upper limit to the true age of the flow, because - as previously discussed - the flow axis lies almost in the plane of the sky: therefore, while the size of the lobes is a good estimate of the real size, the velocity measured underestimates the intrinsic expansion velocity. As a consequence, the mass loss rate, momentum, and mechanical luminosity in Table 6 are to be taken as lower limits. In particular, the mass loss rate looks unusually high: although this estimate is sensitive to the assumed HCO abundance, we believe that values such high cannot presently be excluded.
It is interesting to compare the energy in the outflow with that dissipated by the shocks emitting the H2 line. The luminosity of the H2 line is obtained from the integrated flux in the H2 image, which turns out to be about . This estimate has to be increased by a factor to take into account the global emission in all H2 lines (Davis & Eislöffel 1995) and is anyhow quite uncertain due to the unknown foreground extinction; nevertheless, as already noted, also the computed outflow luminosity is a lower limit, so that one can reasonably conclude that only of the flow energy is radiated in the shock. This is similar to values found for other flows (see Table 7 of Davis & Eislöffel 1995): the fact that such ratio is insensitive to the geometry of the different regions seems to indicate a unique acceleration mechanism for the outflow.
4.2. The ambient molecular gas
The outflow described in Sect. 4.1 arises from a molecular clump mapped in the 13 CO, CS, and HCO lines, as shown for example by the grey scale in Fig. 4. The densest part of this clump is seen in high density tracers such as those of Fig. 3 and in the PdBI maps of HCO (see grey scale in Fig. 7): clearly, the bulk gas surrounds the position of the H2 O masers and has an angular size of ( pc).
It is important to estimate the kinetic temperature, , and the mass of the ambient gas. This can be done using different tracers and the results are given in Table 7. In particular, can be taken equal to the peak of the 13 CO(2-1) line (see Fig. 1 of WBM), which is thermalised and optically thick at line centre, and appears resolved in our 30-m maps; alternatively, one can use the peak of the PdBI maps in the HCO (1-0) line. Another possibility is to assume that the CH3 OH molecule is populated according to LTE and derive the rotational temperature by means of Boltzmann plots of the CH3 OH(3-2) and (5-4) lines (see Fig. 14). These different approaches lead to kinetic temperatures in the range 30-50 K, which we consider to be encouraging agreement. One also sees that higher density tracers indicate slightly higher temperatures, which suggests that the temperature increases towards the centre of the clump; this indication will be confirmed by the even greater temperature of the core associated with the H2 O masers (see Sect. 4.3).
Table 7. Temperature and mass estimates for the bulk gas
The mass can be computed from 13 CO or HCO with the same method used for the values in Table 5: in this case the line emission has been integrated in the velocity interval from -5 to -2 km s-1 over the whole region mapped with the 30-m telescope. Otherwise, the C34 S lines can be used to obtain an H2 density of cm-3, by fitting the observed line intensities with a numerical model based on the large velocity gradient approximation (see e.g. Cesaroni et al. 1991): then the mass is computed once the clump radius is known ( pc). Finally, an additional estimate of the mass is obtained from the continuum flux measured at 1.3 mm by Walker et al. (1990): their beam should cover most of the clump and hence give a reliable value for the total mass of the ambient gas. Under the assumption that the 1.3 mm continuum is due to optically thin dust emission, we have used Eq. (A10a) of Mezger et al. (1990), with b =1.9, solar metallicity, and =40 K, to compute the gas mass. This is given in Table 7.
Note that the masses in Table 7 derived from 13 CO and HCO may be lower limits, since the emission is optically thick. In fact the optical depth at line centre can be easily derived from the ratio between 12 CO and 13 CO (WBM) and between HCO and H13 CO (Acord, priv. comm.) and turns out to be for 13 CO and for HCO . Note also that the mass derived from the 1.3 mm continuum differs from that computed by Walker et al. (1990) because of different values used for the distance, temperature, and dust emissivity.
The different estimates of the mass give consistent values, with the sole exception of that derived from 13 CO which seems sensitively larger than the others, although only by a factor . As already noted in Sect. 4.1, this is probably due to the fact that, given the large abundance of 13 CO, the 13 CO line is sensitive also to more extended, lower density/temperature regions than the other tracers. We conclude therefore that a mass of 100 is a reasonable estimate for the gas on size scales. This corresponds to an H2 density of cm-3.
Note that the existence of a large temperature increase towards the cloud core is very likely, as already noted above: here, we stress that such temperature gradient is not taken into account by the method used to derive a mass estimate from the millimeter continuum flux, which instead assumes the cloud to be isothermal. A better model capable of taking into account also temperature gradients would be necessary to compute the fraction of the mass in the ambient cloud (where K) and that in the core ( K), but this goes beyond our purposes. We shall come back to this point in Sect. 4.3.
4.3. The CH3 CN core
Both the single dish maps of the high density tracers (Fig. 3) and the HCO interferometer maps (Fig. 7) demonstrate that the emission in the range from -9 to 2 km s-1 is closely associated with the compact core detected in the CH3 CN and CH3 OH lines and in the 3.3 mm continuum (Fig. 9). The size of this core is 0.01 pc and hence we consider it likely that it is a "circumstellar" rather than "interstellar" in origin. The high temperature derived (see Table 8) confirms this. It coincides with the H2 O maser spots and with the nominal position of the IRAS source: while the former association is clearly established as the position of the masers is very accurate (the VLA beam is about 008), the latter is uncertain, due to the low angular resolution of IRAS. Nevertheless, in the following we shall use the name IRAS 20126+4104 to indicate the core seen with the PdBI: we shall justify this at the end of this section.
Table 8. Physical parameters of molecular core at IRAS 20126+4104 position
We now discuss the nature of such core.
From the dust continuum and the CH3 CN lines one can derive estimates of the diameter, mass and temperature of the core. These are given in Table 8.
The temperature has been derived from the CH3 CN lines measured towards the peak position by using two different methods. The first, called the rotation diagram (RD) method, assumes that all energy levels of CH3 CN are populated according to local thermodynamic equilibrium (LTE) with a single excitation temperature. The second, called rotational temperature equilibrium (RTE) method, uses two distinct excitation temperatures to describe the level populations (for further details see e.g. Olmi et al. 1993). Both methods assume optically thin emission in the lines.
We have applied the RTE method to the CH3 CN(5-4), (8-7), and (12-11) ground state lines and the RD method to the CH3 CN(5-4) ground state and =1 lines. In both cases a source diameter of has been assumed. The fits obtained with these techniques are shown in Figs. 15 and 16. In the latter we have also tried to correct for optical depth effects in the ground state CH3 CN(5-4) lines multiplying the corresponding column densities by a factor 5: in fact, this is a rough estimate of the CH3 CN(5-4) optical depth as derived from the ratio between the CH3 CN and C CN lines assuming [CH3 CN/H2 ]/[13 CH3 CN/H2 ]=70. The RTE technique is consistent with K, whereas the RD method implies K in the thin limit and K when the correction for the optical depth is applied.
We thus conclude that 200 K is the most likely value for in the core. Incidentally we note that neither the core nor the ambient gas temperature are sufficient to explain the observed line FWHM in terms of thermal broadening: in particular, as pointed out in Sect. 3.1, the FWHM increases from km s-1, in the lines tracing the ambient gas, to km s-1, in the transitions arising from the core. Such an enhancement resembles that seen in Mon R2 by Tafalla et al. (1996) and suggests that turbulence increases on small scales, close to the origin of the outflow.
The same methods used for computing give also an estimate of the source averaged CH3 CN column density, and hence of the gas mass of the core, assuming [CH3 CN/H2 ]= : this is indicated with in Table 8. Alternatively, the mass can be estimated, in the optically thin limit, from the continuum emission at 3.3 mm - which we assume to be mostly due to dust (see Sect. 3.3) - as already done in Sect. 4.2, assuming =200 K: this is indicated with . The two values agree very well, suggesting a core mass of .
As already discussed in Sect. 4.2, the assumption of a single temperature for deriving the mass from the continuum flux is not very realistic. However, we believe that can hardly be wrong by more than a factor , because the contribution to the millimeter continuum flux from the high-mass (), low-temperature ( K) ambient gas is comparable to that from the low-mass (), high-temperature ( K) core.
One can estimate the bolometric luminosity, , by integrating the four IRAS fluxes, shown in Fig. 11. Under the assumption that most of the emission seen by IRAS originates from the core, can be taken as an estimate of its luminosity. The result is given in column 7 of Table 8. In column 8 we show the spectral type of the star with luminosity , as derived from Panagia (1973). seems to indicate the existence of an early type (B0.5) star: this is perfectly consistent with the IRAS colours of IRAS 20126+4104, which satisfy the Wood & Churchwell (1989b) criteria for identifying UC HII regions. For an optically thin UC HII region fed by a B0.5 star one would expect a continuum flux at 3.6 cm of mJy; on the other hand, the continuum flux measured at 3.6 cm by Marti & Rodriguez (in prep) is much lower ( mJy), consistent with a B2.5 star. The discrepancy between the spectral types derived from and from the radio flux is well known (see e.g. Wood & Churchwell 1989a): in fact, the former tends to overestimate the spectral type if many stars fall in the IRAS beam, the latter underestimates it if dust is present in the UC HII region.
We conclude that there are clear indications for IRAS 20126+4104 being a young early type (B2.5-B0.5) (proto)star deeply embedded in a hot, dense, compact molecular core.
© European Southern Observatory (ESO) 1997
Online publication: April 28, 1998