Astron. Astrophys. 346, 243-259 (1999)

4. Discussion

In this section, we are going to build a model of G5.89 and put it into the context of the formation of massive stars and UCHII s. First we will deal with the foreground cloud of dust which is thought to be responsible for the appearance of G5.89. The presence of this foreground cloud together with the half-spherical appearance of G5.89 at almost all observed wavelengths leads us to build a model of the source which assumes a spherical shell of dust that is half blocked from view. We then argue that we also found evidence for the long-known outflow from G5.89, but not for a circumstellar disk which could serve as a driving mechanism for this outflow. Combining these findings with a look at the surroundings of G5.89, we argue that the object contains one single young massive star while at the same time it is obviously a part of a cluster of such stars.

4.1. The surroundings of G5.89

4.1.1. Clouds of dust - the foreground extinction

The measured flux at 1.3 mm indicates that substantial amounts of cold dust are present at the location of G5.89. In Fig. 6, the 90% contour line of the dust contribution to the 1.3 mm flux shows that this emission originates clearly offset from G5.89 proper. After subtracting the free-free contribution, the remaining flux of 8.5 Jy should entirely arise from cold dust. For the following estimates, we assume that the dust emission is optically thin at 1.3 mm. Using the mass absorption coefficient at 1.3 mm given by Ossenkopf & Henning (1994) and assuming a temperature of 30 K, we derive a total mass of 7.5 . Assuming that this mass is concentrated in a core of 16" diameter, the corresponding dust opacity from Ossenkopf & Henning for 21 µm yields an extinction of about 58 mag at that wavelength. Thus, even if only a fraction of this measured column density is situated in front of G5.89, it explains easily why even at 21 µm we cannot see through the foreground dust.

The similarity of the shape of G5.89 over the wavelength range between 1.6 and 21 µm means that the rim of this cloud has to be rather sharp. The column density needs to "jump" exactly across G5.89. The beam size of the 1.3 mm measurement does not allow us to measure the spatial distribution of the cold foreground dust accurately. However, Fig. 6 clearly shows the dark cloud south and west of G5.89 and its brownish rim stretching across the source at least at 2.2 µm . Furthermore, the extinction towards G5.89 was derived with high spatial resolution in Sect. 3.5. The result in Fig. 9 shows clearly that the extinction is rising in the expected direction until the NIR emission drops off completely. The cloud is also obvious by its blacking out of background stars, as seen in Fig. 6. We note that the orientation of the cloud's rim is consistent with the orientation of the source in the 1.3 mm map.

The rims of this cloud appear relatively bright in all three narrow-band filters, but interestingly are weakest in Br . Although the widths of the filters and the calibration uncertainties of the images do not allow a quantitative analysis of such weak structures, this seems to indicate that the UV field or shocks by winds from nearby stars are sufficient to excite the H₂(1-0)S1 line, but not to ionize the cloud material to a large extent. If the jump in density is as large as speculated above, UV emission cannot penetrate as deeply into the cloud as a shock wave triggered by winds might. Thus, a larger emitting region would be provided in than in Br and the former line would consequently appear brighter.

4.1.2. Star formation in G5.89's neighbourhood

In Paper I, we argued that G45.45+0.06 is a young cluster of massive stars still in its formation phase. From a comparison with the Orion BN/KL region (named after Kleinmann & Low 1967 and Becklin & Neugebauer 1967) we inferred that formation in clusters might be the rule rather than the exception for massive stars. The high-resolution images of G5.89 again reveal a number of unresolved sources embedded in extended emission. However, the situation is different: G5.89 is only 2.6 kpc away compared to 6.3 kpc for G45.45+0.06. Thus, we reach 2.5 times the linear resolution for G5.89 (1000 AU in ) compared to G45.45+0.06. If the two objects are indeed similar, as we will argue below and in Sect. 4.4.1, G5.89 can for example correspond to one of the embedded radio sources in G45.45+0.06.

In Paper I, we identified the stars by means of two-colour photometry. This method was again applied to G5.89 and the result can be seen in Fig. 12. Some sources are evidently stars such as source 1 (moderately reddened) and 12. From a comparison of Fig. 11 with Fig. 6, it becomes clear that the chain of sources 6-9-7-10-11-13-15 belongs to the suspected rim of the foreground cloud. Thus, outstanding sources among them may just represent irregularities in the surface of this cloud. This is probably the case for sources 7 and 9 which have the same colour as 1 and are thought to be reflected light from that star. On the other hand, sources 6 and 0 get quite close to the ZAMS when a de-reddening is applied as was derived for the rim of the cloud across the radio shell (see the typical de-reddening vectors in the figure). The latter two sources are thus probably stars. From their locations in the map and in the colour-magnitude diagram, sources 3 and 5 are candidates for high-mass stars.

Some sources could be de-reddened directly because they are inside the area where Br and radio data exist. These sources have de-reddening vectors attached to them in the figure. It appears that sources 15, 16, and 20 are non-stellar objects, while source 19 (and probably also 18) is obviously a foreground star.

Thus, we have indeed a few candidate young stars close to G5.89, which are summarized in Table 3. More facts can be extracted from Fig. 6: An area of extended Br emission is seen around position (+17",-20"), which is identical with the area of weak emission in Gomez et al. (1991). The distance of that region to G5.89 is roughly 25" so the whole complex has about the same linear size of 0.33 pc as G45.45+0.06. In Sect. 4.3, we will explicitly compare G5.89 to the embedded sources in G45.45+0.06.

Table 3. Presumed candidates for young stellar objects close to G5.89

The presence of the more evolved HII Region G5.90-0.43 in a projected distance of 120" or 1.5 pc also represents a similarity to the situation for G45.45+0.06. In Paper I, we followed speculations by Wilner et al. (1996) that star formation spread from G45.45+0.06 to its two neighbouring UCHIIs at distances of about 5.8 pc. In that latter case however, G45.45+0.06 is thought to be the originator of star formation in the region, while in the case of G5.89, its neighbour is obviously more evolved and thus G5.89 itself would be the object where triggered formation of stars is taking place.

4.2. The morphology of G5.89

4.2.1. The hollow sphere

CWW modelled the emission from the dust around G5.89 to fit the observed spectrum of the source. Their "model 5" became the de facto standard model for comparison with observational data. It was essentially recalculated in Faison et al. (1998), who claim their model differs from CWW in density and outer radius. However, the numbers given in their paper are the same as those in CWW. The model consists of a spherically symmetric dust distribution around a central star of spectral type O6 ZAMS. A fairly large cavity at the sphere's centre is depleted of all dust in order to fit the near-infrared part of the spectrum. Now that we have high-resolution information, we can compare not only spectral but also spatial information with these model calculations.

The code we used for the radiative transfer (RT) calculations was developed by Manske et al. (1998) and is based on a method given by and described at length in Men'shchikov & Henning (1997). The code was used in one dimensional mode as the spherically symmetric dust distribution requires no further refinements. To compare the results of the calculations to the observational data, we calculated simulated maps of the source at 1.6 µm , 2.2 µm , 3.5 µm , 10.6 µm , 11.7 µm , 12.8 µm and 21.0 µm . One half of the output maps were arbitrarily set to zero to account for the assumed foreground extinction which also covers the central object. The last step was to convolve the resulting maps with a PSF of 1" FWHM as a representative observational beam size. As an example, the calculated map at 11.7 µm is shown in Fig. 13a.

Fig. 13a-f. Comparison between model calculations and observational data. a  Simulated map at 11.7 µm . The lines show the extraction profiles that were used on the observed maps to extract the profiles shown in C-F. Also shown is the aperture used for photometry on both the real and simulated maps. b  SED. The solid line is the result of our calculations, The diamonds denote our observational data, the crosses are observations on the simulated maps. The squares are IRAS fluxes. The crosses denote fluxes measured on the simulated maps. All our own fluxes were measured inside the aperture marked in part A, except the 1.3 mm flux, which is the total measured dust contribution. c  Radial distribution of the observed mass column density. The solid line is derived from the observations as described in Sect. 3.6. The dashed line again marks "observations" on the simulated maps. The dotted line denotes the "real" column density as integrated from the input dust distribution for the RT model. d  Radial temperature distribution. The solid line represents an extraction from the temperature distribution shown in Fig. 10. The dashed line was extracted from a temperature distribution derived in the same way as described in Sect. 3.6, but from the simulated maps. The dotted lines denote temperatures of the different sorts of dust grains computed from the RT model. e  Radial intensity distribution at 11.7 µm . The solid line is from our observation, the dashed line from the simulated map. f  Radial distribution of the silicate optical depth as described in Sect. 3.4. Again, the solid line represents extractions from observational data while the dashe line denotes an extraction from the simulated maps. The arrow in parts C-F marks the radius of the inner rim of the dust shell in the model calculations.

To be able to compare the results to those of CWW, we used the same dust properties as were used for their model 5 and started off from the same geometrical assumptions. It turned out that we mainly needed to change the total mass of dust involved and the density distribution as well (slightly) the radius of the inner dust-free cavity.

The parameters used for the best fit are summarised in Table 4. Fig. 13 shows the main results of the model calculations. We will now briefly discuss this figure and the main consequences for the very simple model of a spherically symmetric dust shell.

Table 4. Parameters used in the modelling of the source

Fig. 13b shows the resulting spectral energy distribution (SED). The solid line gives the resulting SED for the parameters listed in Table 4. Although these parameters are significantly different from those of CWW, the result looks almost identical to those of CWW's model 5. The diamonds show observed fluxes of the source within a mask that is shown in Fig. 13a (except for the 1.3 mm flux which is the total measured flux after subtraction of the free-free contribution). The crosses denote fluxes measured in the same aperture in the simulated maps of the model. The representation of the observed SED by the model SED is very good, except in the near-infrared.

Fig. 13c compares the measured mass column densities of hot dust as derived in Sect. 3.6. The solid line gives the profile as measured in the map shown in Fig. 10. The profile was extracted in the following way: The source was assumed to be radially symmetric and profiles of 6" length were laid from the centre at R.A. , DE. -24^o 03´56:007 in 19 directions from -45^o to 45^o around north (19 profiles in 5^o steps). The arrangement of these profiles is sketched in part a of the figure, the centre was determined such that all extracted profiles do not shift with respect to each other by more than 0:0017 (one pixel). The profile shown is the mean of the 19 extracted profiles. The same procedure as described in Sect. 3.6 was applied to the simulated 11.7 and 21.0 µm maps and the resulting extracted profile is denoted by the dashed line. The dotted line shows the "real" model mass column density in the line of sight, integrated from the input dust configuration inside the aperture region of Fig. 13a where hot dust is visible. The temperature distribution which results from the same procedure is compared in part d of Fig. 13. The solid line is again extracted from the contour map shown in Fig. 10, the dashed line from the result of the procedure applied to the simulated maps. The dotted lines show the temperatures of the different sorts of dust grains in the model. We note that the deviation of these grain temperatures from the colour temperatures derived from the maps at small radii is not an effect of optical depth, but is due to averaging the observed temperature along the line of sight. This effect is thoroughly discussed, e.g. in Schreyer et al. (1996).

Fig. 13e compares the radial profile of the intensity distribution at 11.7 µm . The solid line gives the profile as measured in the map shown in Fig. 3, the dashed line represents the profile as extracted from the corresponding simulated map.

Part f of Fig. 13 compares our pseudo silicate optical depth by showing the extracted profile from the map in Fig. 8 as solid line and the result of the same operation on the simulated 10.6 and 11.7 µm maps as dashed line.

We will now briefly discuss the main aspects of the model:

• The main goal of the modelling was to show that this simple geometry can reproduce the radial intensity distribution. This is clearly shown in Fig 13a and e. The shape and the position of the maximum flux density is governed by the location of the inner radius of the dust shell and the exponent of the power law which determines the decrease in density towards the outer regions of the dust shell.

• The SED is practically identical to that in CWW's model 5. It fits the MIR observations nicely, as is especially visible by the marks for simulated and real observational data. Note that the solid line represents the total flux of the spherical dust shell, while the simulated observations rely on only half the flux measured through a virtual telescope with a 1" beam. The resemblance to CWW's model 5 is somewhat astonishing as our calculation relies on only about 43% of the mass compared to that of CWW. In addition, we used a density gradient exponent of -0.6 compared to CWW's constant density distribution. Note that when using CWW's original parameters for model 5 with our code, we get a much deeper silicate feature and an earlier and steeper falloff towards the NIR. Harvey et al. (1994) pointed out that the overall mass doesn't play a decisive part in the modelling, which is true as long as only the SED is concerned. For their own models, they used density gradients of and . With such steep gradients, we cannot fit our observed radial profiles.

• The radial profiles of density and temperature show basically that the simulated observations reproduce the real ones within a factor of two. As far as the mass is concerned, the radial dependence also follows nicely the model input quantities. When comparing the "real" temperature to the observed ones, one has to be careful that we never see only the inner edge but always an average along the line of sight (at best). The effects of the finite optical depth given in Table 4 seem not to be serious. The optical depths in the table are valid for the line of sight towards the central source, whereas we are only dealing with the inner 4% of the dust shell as far as the hot dust is concerned.

• The radial dependence of the pseudo depth of the silicate feature is completely different in the model and the observations. While the observational data show that silicate optical depth is high in the same places were emission is generally high at all wavelengths, the computed from the RT model rises only outwards of 3" (7800 AU). This behaviour of the model is easily understood in terms of line-of-sight effects. Contrary to this simple explanation, the observed high silicate depth at the inner edge of the dust shell indicates the non-applicability of spherical symmetry in this case: Starlight escaping through the channel of the outflow - visible as a breach in symmetry in the radio maps (see below) - and heating dust grains at the channel walls can generate MIR-emission at locations, where less and less forground dust exists to produce an absorption feature (The channel opening is assumed to be pointing north, exactly the region where our profiles are extracted). Thus the falloff of optical depth towards the outside of the shell is understandable only in terms of non-spherically symmetric dust-distribution.

• The low temperature of only 200 K at the inner boundary of the dust shell implies that the dust-free cavity must be made and maintained by some mechanism other than the destruction of dust particles through mere heat.

Surely the simple model of a spherical dust shell is far from perfect. The observed maps show that in the end the source is not spherically symmetric, even when assuming to see only half the sphere due to foreground dust. The radio maps show a "channel" aligned roughly in north-south direction and the outflow(s) from the object need openings in the shell to get out (see next section). Interestingly, most non-symmetric features in the near- and mid-infrared maps are located close to the northern end of the "channel". This might indicate processes of scattering and reflection of light leaking out of the sphere through an opening. Figs. 1 and 2 show that in the NIR the maxima in flux density are on both sides of the channel's opening, while in the MIR maps the maximum shifts towards the opening itself with longer wavelengths. The SED shows that in the NIR the predicted fluxes are too low by a factor of around 100 or more. This can also be explained by direct starlight leaking out of the shell through this channel. Because the northern end of the channel is pointing away from the observer as infered from the direction of the outflow (see nect section), this light could still not reach the observer directly. It would have to undergo one or more scattering processes, a fact which might also help to explain the extremely red colour of source 16 in Fig. 11.

Albeit, we have seen that the spherical model can explain most of the observed properties and we now know that the basic asymmetry - that only half the sphere is visible at infrared wavelengths - is due to extremely massive foreground dust concentrations. This is exactly the scenario which was argued to be unlikely by Harvey et al. (1994).

4.2.2. The outflow(s) of G5.89

G5.89 is thought to be the origin of one of the most massive outflows within the Galaxy. However, there is an ongoing discussion as to what the actual orientation of this outflow is and whether there is only a single one or if multiple outflows are responsible for the puzzling observational results. Among the first groups to examine this phenomenon were Harvey & Forveille (1988) who found a strong outflow in ¹²CO(1-0). They concentrated on a red wing towards the east of the source and a much stronger blue wing towards the southwest. However, their original data show two red wings, one of which is coincident with the source. The blue wing shows up about 4" slightly west of south from the source. This indication for an outflow slightly off the N-S direction agrees much more with later findings than the originally anticipated E-W direction.

Further interesting findings come from Zijlstra et al. (1990) and Cesaroni et al. (1991). Zijlstra et al. (1990) found indications for a N-S outflow from OH maser velocity fields. They derive the northern cone to be directed towards us, which is opposite to most other measurements where the blue shifted emission is detected south of the source. Cesaroni et al. (1991) mapped the outflow in C³⁴S. They find the orientation of the flow to be slightly west of north with the blue wing south of the source. Their red wing centre also almost coincides with the source position.

Acord et al. (1997) contributed high-resolution SiO data obtained with the VLA. They find an overall orientation of about 45^o east of north with the blue-shifted emission just southwest of the source position and the red-shifted emission about 4" to the northeast. They also speculate on the cause for the blue-shifted emission to be much stronger and closer to the source than the red shifted counterpart and conclude that this is due to higher density of the surrounding material to the south. This results from their assumption that SiO is produced in a shock when the outflow hits the surrounding material.

Our findings concerning the outflow can be drawn from Fig. 6: We assume that the H₂(1-0)S1 line is due to shocks when the outflow hits the surrounding material. Three features can be seen in this line, indicated by red emission in Fig. 6. The first and strongest, marked "A" is in position exactly coincident with the centre of the blue shifted C³⁴S emission found by Cesaroni et al. (1991). The counterpart ("C") is about 5" north of Cesaroni's red shifted position. Feature "C" is quite weak and extended. With the assumption that H₂ is shock excited, this supports the view that the density of surrounding material is much higher to the south than to the north. The feature "A" is about 4" closer to the source's centre than "C". The higher density towards the south is also consistent with our findings in Sect. 4.1.1 and our assumptions on the general structure of G5.89. Another H₂ feature, marked "B" originates from the star at position (-14",+4"). We do not have any colour indication on that star, but the feature might indicate that it is also a young object.

4.2.3. No disk in G5.89?

Yorke & Welz (1996) introduced an outflow driving mechanism that relies on the photo-evaporation of circumstellar disks in combination with a stellar wind. Their disk sizes are of the order of 300 AU (they are considering B stars!). We have no direct evidence for the presence of such a disk inside G5.89 and the necessary resolution of 0:0011 is inaccessible to us. However, the broken symmetry of the radio shell and the presence of the channel opening might point to a disk-like or toroid structure, as was first pointed out by Zijlstra et al. (1990). This structure may well continue into a small-scale disk inside the otherwise dust-free cavity. A rough simulation of a configuration with an O6 ZAMS star surrounded by a disk of 6 M with a radius of 1000 AU and an inclination of 30^o , shows that the disk contributes only between 10 µm and 60 µm to the spectral energy distribution significantly. However, since the flux never reaches the 10% level of the measured flux, it would not alter the measured (and modelled) SED, so we cannot detect such a disk inside G5.89's dust shell just from the spectrum. At cm wavelengths, a disk comparable to the candidate disk of G339.88-1.26 ² might be visible in G5.89. Taking G339's 3.5 cm flux of 14 mJy (Ellingsen et al., 1996) and accounting for the smaller distance and brighter central star, a comparable structure in G5.89 should emit around 190 mJy at 3.5 cm and thus be visible in the WC89 2 cm map. However, two arguments militate against the presence of such a disk in G5.89:

• The orientation of the disk plane should be east-west, as suggested by the outflow. An inclined disk should thus, according to simulations by Kessel et al. (1998) appear as a north-south elongated structure. Nothing like this is detected in the maps.

• A disk in east-west direction should absorb all UV photons emitted in its plane. Thus it would prevent ionization of the spherical shell to the east and the west of the central object. This is however exactly the direction where the ionization appears to be strongest. This strong ionization towards the east and west is on the other hand exactly what one would expect if an east-west oriented disk is being evaporated by UV radiation and disrupted by strong stellar winds (Hollenbach et al., 1994). The pronounced structure of the ionization might thus indicate that the disk is either in a late stage of its destruction by the central star or has already vanished completely.

As we practically rule out the presence of a disk inside G5.89, the question of a driving mechanism for the outflow (and indeed for the formation of the central star) remains. The possible answer is that a disk was present inside G5.89 and served as a driving mechanism for the outflow. It might be completely evaporated by now, while the outflow is still escaping the shell due to the high pressure of the ionized gas. The channel opening can serve as a collimation mechanism. Little is known about disks around massive stars, let alone their lifetimes. However, given the extreme youth of G5.89 - the dynamical ages of the shell and the outflow are 600 yr (Acord et al., 1998) and 3000 yr (Acord et al., 1997) - either renders the lifetime of such a disk extremely short or questions the dynamical age as a reliable age estimator.

4.3. G5.89 and the models of ultracompact H II regions

4.3.1. Sketch

We will first put all the discussed facts together to form a complete picture of the source. The spatial distribution of all the features discussed so far is shown in Fig. 14. It shows G5.89 as a spherical shell of dust surrounding the hot star with a dust-free inner cavity and two emerging outflow cones. The object is located at the rim of a large cloud of dense cold dust, which obscures its southern half from the view of the observer who is located to the right. One additional source is shown: This is an example for one of the surrounding stars, e.g. the wavefront reference star. It is illuminating the rim of the cloud and its wind presumably shocks H₂ molecules there.

Fig. 14. Sketch of G5.89's configuration. The large figure is a side view where the observer is located to the right. The inset on the upper right gives the observers view of such a model. Only the ends of the outflow cones are visible here, as is expected for shocked emission. The dashed southern part of the ionized shell should be visible only at radio wavelengths.

4.3.2. A strömgren sphere?

The viewpoint summarised in Fig. 14 states that G5.89 is an ideal UCHII . An O6 ZAMS star has formed an ionized shell around it of roughly 4:005 or 0.05 pc diameter. From WC89 we learn that this is a typical initial size of a Strömgren sphere. Intriguingly, the inner radius of our modelled dust shell is 5050 AU or 0.025 pc. Thus, the ionized sphere is obviously identical with the dust-free inner cavity. We can easily see that this must be the case: The model predicts a dust density of 5 g cm^-3 at the shell's inner edge. Estimating the UV optical depth by

we obtain unity after a distance AU or 0:0003. This uses a dust mass absorption coefficient for 2.2 µm from Ossenkopf & Henning (1991), the factor of 44.8 converts the K optical depth to the desired UV depth and is taken from Mathis (1990). 99% of all UV photons are absorbed 590 AU or 0:0022 beyond the inner boundary of the dust shell. Thus, only the very inner edge of the dust shell can be ionized.

On the other hand, we have learned that the ring structure at infrared wavelengths is due to a combination of the dust and temperature distribution and a line-of-sight effect through the hollow sphere. Since exactly the same structure is visible at radio wavelengths, it is unlikely that the dust-free, inner sphere is uniformly filled with ionized gas. Obviously, the gas (plasma) density is highest close to the inner edge of the dust shell.

One more hint, that G5.89 is not a classical Strömgren sphere after all, is its expansion rate. It was measured by Acord et al. (1998) to be 41 mas yr^-1. Estimating the pressure contrast between the ionized gas ( cm^-3 from the electron density, K) and the innermost part of the dust shell ( cm^-3, K, both from the RT model), we derive a pressure contrast of about 300:1, i.e. the region should be freely expanding via an R-type ionization front. Using Eq. 9 from WC89, we derive an expected expansion rate of mas yr^-1 for an UCHII of 0.025 pc initial radius. Thus, the expansion cannot be driven by pressure differences between ionized and non-ionized material alone. The stellar wind that disrupted the circumstellar disk may serve as an explanation here: From Puls et al. (1996) we learn that an O-star of 42000 K has typical mass-loss rates of 3 per year. This mass has a speed at infinity from the star of approximately 2500 kms^-1. Thus, the influx of kinetic energy into the hollow sphere is of the order of 10³⁷Ws per year . The energy thermally stored in the ionized gas is of the same order of magnitude, so the wind may well play a decisive role in accelerating the expansion.

4.4. The dust-free cavity

A possible explanation for the rapid expansion would be if the inner surface of the dust shell is being photo-evaporated by the UV radiation from the central star as proposed by Hollenbach et al. (1994) for disks. The inside of the cavity would in this case not only be free of dust, but also possess a much lower gas density than the border region. Other models that explain the presence of an ionized shell at the boundary layer ia-Segura & Franco between dust and dust-free regions are described by Garc (1996) and by Dyson et al. (1995). Especially the latter paper deals with the formation of shell structures due to a strong stellar wind driving out the ionized material. Hollenbach et al. (1994) also show how the evaporation of the circumstellar disk leads to this spherical morphology when interacting with a strong stellar wind. Thus, one might well infer the presence of such a strong wind - and even the former presence of a now evaporated disk - inside G5.89 that formed its shell-like morphology in both the distribution of the dust and of the ionized gas.

As already mentioned in Paper I, we again learn that dust plays an extremely important role in UCHII s. Hot dust is usually present in regions where also strong ionization is maintained. In Paper I, we had a problem of maintaining the observed ionization through the amounts of measured dust. Now that we can take G5.89 as a detail of a larger region, we see a possible mechanism how the depletion of dust might lead to the formation of hollow spheres and typical, spherical UCHII s. We note that projected to the distance of G45.45+0.06, G5.89 would appear similar to one of the embedded radio peaks in G45.45+0.06 in both size and intensity (i.e. the radio peaks close to sources a and and m in Paper I).

4.4.1. A cluster nevertheless

As already stated above, G5.89 is obviously embedded in surroundings where more massive stars form. The Br emission region 22" southeast, the identified hot stars in the vicinity and the feature attached to a star northwest of G5.89 support our reasoning that the surrounding complex is indeed very similar to that of e.g G45.45+0.06. The extreme youth of G5.89 close to some "normal" O stars again raises the question of sequential star formation. As G5.89 is considerably closer than G45.45+0.06, we now have the chance to look also for low-mass stars in the cluster but all sources in Fig. 11 that might be shifted to a ZAMS spectral type later than A are either thought to be reflections on the rim of the large cloud of dust or are not obviously associated with the cluster. If G5.89 is indeed as young as its associated dynamical ages suggest, the presence of main-sequence low-mass stars in the cluster is highly unlikely, but deeply embedded objects might be detected in the future.

© European Southern Observatory (ESO) 1999

Online publication: May 6, 1999
helpdesk.link@springer.de