3.1. The distribution of ammonia cores in the W 3 HDL
The contour plot of the velocity integrated emission (including all hyperfine [hereafter HF] components) toward the W 3 HDL is shown in Fig. 1. We have overlaid the plot onto a greyscale map of the integrated CO (1-0) emission toward the region taken during the FCRAO second quadrant survey (Heyer et al. 1998). The strongest CO emission arises in two regions: the W 3 Main and the W 3(OH) clouds.
To the east of the W 3(OH) and W 3 Main clouds, the CO emission drops off; this is the edge of the W4 H II region (Lada et al. 1978). To the west of the two clouds, diffuse CO emission is apparent. The diffuse CO emission extends westward as far as one degree beyond the borders of Fig. 1. The high density layer (HDL) of Lada et al. (1978) consists of the W 3 Main and W 3(OH) clouds and a third cloud to the south of W 3(OH) which was not included in our map. Although CO emission is detected throughout the region covered in our map, between the W 3 Main and W 3(OH) clouds there is a hole in which only weak CO emission is detected. This hole is coincident with the IC 1795 H II region which is seen in optical images (Dickel et al. 1983).
The outline of the map in Fig. 1 shows that we confined our survey to the high density layer and that we have included most of the W 3 Main and W 3(OH) clouds and a small section of the IC 1795 H II region. We have detected line emission at a 0.5 K km s-1 level over 27% of the mapped region, and at a 1 K km s-1 level over 9% of this region. These levels correspond to the 3 and 6 levels for the velocity integrated emission . We note that the weaker emission regions typically exhibit narrow linewidths, hence, most of the emission at the velocity integrated 3 level results from line emission with peak temperatures 5 or more above the noise level.
There are three strong peaks of emission within the map. We find strong emission toward the west of the W 3 Main cloud, coincident with a massive core detected in previous maps of the region. At the southern end of our map, we find the W 3(OH) core which is centered on the W 3(OH) H II region. In addition, we find one previously undetected core at R.A. , Dec. (1950.0). This core, W 3 SE, is located southeast of W 3 Main. Outside of these three cores, we find only regions of weak emission.
3.2. The -band mosaic and the identification of stellar clusters
The -band mosaic is reproduced in Plate 1, and in Fig. 2 we display the emission overlaid onto the -band mosaic. Fig. 2 shows that the mosaic encompasses the HDL from W 3 Main to W 3(OH) and most of the positions observed with the 100-meter telescope. In the -band mosaic, regions of recent star formation are traced by the bright nebulosities and compact clusters of stars. The -band is highly sensitive to embedded stars, and at the 3 detection limit, we can detect 0.1 , 1 Myr, pre-main sequence stars at the distance of the W 3 GMC. Hence, while the H II regions detected in centimeter surveys of the cloud trace high mass star formation, the clusters detected in the -band image are composed primarily of young, low mass stars.
Through visual inspection we have identified five regions which show obvious stellar density enhancements, or clusters. Since each cluster is detected toward the molecular gas within the W 3 HDL, we associate these clusters with the W 3 region. We also find an extended distribution of stars not associated with the five clusters. It is not clear whether these stars are isolated young stars in the W 3 GMC or unrelated field stars in the line of sight. Consequently, we restrict our analysis to the young stellar populations in the clusters.
Three of the five embedded clusters have been reported previously in the literature. Between W 3 Main and W 3(OH) we find a cluster of stars surrounding the O8 star BD 411. This cluster has been studied through optical photometry (Ogura & Ishida 1976) and is not deeply embedded. The cluster contains the exciting stars of the diffuse H II region IC 1795 (Dickel et al. 1983), which appears as a low surface brightness nebulosity in our K' mosaic (cf. Plate I). Megeath et al. (1996) studied the cluster in W 3 Main and Rayner et al. (1990) presented the detection of the cluster toward W 3(OH). In addition to these, we have identified two previously unreported clusters which are located to the northeast of W 3(OH).
3.3. The morphology of the W 3 main region
In Fig. 3, we show a contour plot of the emission overlaid onto a grey-scale image of the 6 cm continuum emission toward W 3 Main (Tieftrunk et al. 1997). We use a grey-scale in which the bright sources appear saturated to enhance the low brightness of the extended sources (for the true brightness scale, see Tieftrunk et al. 1997). The individual H II regions traced by the 6 cm emission have been labeled using the nomenclature of Wynn-Wiliams (1971) and Harris & Wynn-Wiliams (1976). A cluster of compact, ultracompact, and hypercompact H II regions is apparent in the northern half of Fig. 3. These H II regions trace the most recent sites of OB star formation in W 3 Main. The two diffuse H II regions in the southern half, J and K, appear to be more evolved regions (Tieftrunk et al. 1997).
Fig. 4 shows a contour plot of the emission overlaid onto the -band data toward W 3 Main. Comparison of the 6 cm map with the -band mosaic shows that, with the exception of W 3 C, all of the H II regions appear as nebulosities in the -band. In addition to the nebulosities, a dense cluster of stars is observed. The cluster density peaks toward the luminous infrared source IRS 5, which is labeled in Fig. 5. A recent NIR study of the cluster around IRS 5 has found that the cluster has a stellar density of 1300 - 4300 pc-3 and is composed of primarily low mass pre-main sequence stars (Megeath et al. 1996).
Two cores are detected toward W 3 Main. The brightest of the two cores is found near the cluster of H II regions and the NIR sources shown in Figs. 3 and 4, with the emission peak clearly offset from the H II regions and the NIR cluster. This particular molecular core has been detected in a number of previous molecular line studies (e.g. Hayashi et al. 1989, Richardson et al. 1989, Hasegawa et al. 1994, Oldham et al. 1994, Tieftrunk et al. 1995, Roberts et al. 1997), and we will refer to it as W 3 West. Interestingly, these studies also detected a somewhat smaller but denser core centered on the IRS 5/W 3 M region and surrounded by the (ultra-)compact H II regions W 3 A, B, and F. However, this core does not show strong emission in the line. We will refer to the core surrounding IRS 5/W 3 M as W 3 East. The H II regions W 3 C and D appear to be embedded in the W 3 West core, while the stellar cluster and the H II regions W 3 M, F, and B appear to be embedded in the W 3 East core (Tieftrunk et al. 1997, Megeath et al. 1996).
The second, newly detected core, which we denote W 3 SE, is found at R.A. , Dec. (1950.0) in the map. The central line velocities of this core are similar to those found for the weak emission toward W 3 East (Fig. 8 & Table 1); consequently, the molecular core W 3 SE is most likely associated with the W 3 Main region. W 3 SE lies just northeast of the evolved, low surface brightness H II region W 3 K. No ultracompact H II regions or stellar clusters are detected in or near W 3 SE. The -band data in Fig. 4 shows several stars coincident with W 3 SE, but it is not clear whether these stars are embedded stars or unrelated stars in the line of sight. However, the -band mosaic also shows a narrow north-south elongated nebulosity toward W 3 SE, which we will refer to as a jet. We have obtained near-infrared narrow-band line images in the 2.12 µm H2 (1-0) S(1) line and K-band GRISM spectra, which have shown that this jet is dominated by vibrationally excited H2 emission. It is not known whether this emission is excited by outflow driven shocks or from fluorescing gas near a UV source. The source driving the H2 emission has not been identified; however, the close positional association of the jet with the core suggests that the driving source is embedded in W 3 SE. Hence, we take the detection of the jet as evidence for ongoing star formation in W 3 SE. A second jet-like nebulosity, detected in our -band data (cf. Plate I), extends from R.A. , Dec. (1950.0) northwest toward the south of W 3 SE. This jet has also been shown to be comprised of vibrationally excited H2 emission; however, as the emission does not envelop it, its association with W 3 SE is not as straightforward.
Table 1. & (2,2) spectral line parameters, temperatures and densities of selected positions toward W 3.
3.4. The morphology of the W 3(OH) region
In Fig. 5 we show an overlay of the velocity integrated emission mapped with the 100-m ( FWHP beamwidth) in thin contours onto the velocity integrated emission mapped with the VLA ( FWHP beamwidth) in thick contours (Wilson et al. 1993). The filled circles indicate the location of the TW object and an ultracompact H II region (Guilloteau et al. 1985). The cross indicates the center of the OH maser complex W 3(OH). The 100-m data shows that the is extended over a area ( pc), much larger than previous maps indicated. Within this region, the VLA map traces only a small fraction of the total emission due to the lack of short baselines. The largest angular scale reliably imaged by the VLA at 1.3 cm in the D-array is . For more extended regions, the VLA data will emphasize peaks and edges in the emission. Since the VLA primary beam at 1.3 cm is FWHM, the VLA data is also not sensitive to emission displaced more than from the pointing center of the observations (i.e. the W 3(OH) H II region). An interesting detection from the VLA data is the extremely narrow plume of ammonia north of W 3(OH) extending to the east and then bending sharply to the north (Wilson et al. 1993). The 100-m data shows that this plume is part of a structure which extends as far as 1.5 pc away from the center W 3(OH) region. The VLA data seems to be tracing a sharp cutoff in the emission at the eastern edge of the ammonia core, perhaps showing the wall of a cavity to the east of the W 3(OH) core.
Fig. 6 is an overlay of the velocity integrated emission onto the -band image of the W 3(OH) region. Coincident with the W 3(OH) region, we identify three clusters in our -band image. Toward the W 3(OH) H II region and directly east of the peak, we detect a rich, compact, cluster of stars which was first identified by Rayner et al. (1990). Two more clusters of stars, which were previously unidentified, are observed toward the northeastern edge of the plume. These clusters are located pc away from the W 3(OH) H II region. If the newly detected clusters are indeed associated with the W 3(OH) molecular cloud, as their proximity to the plume suggests, then it appears that star formation associated with the W 3(OH) region is more extensive then previously thought. Interestingly, these clusters are the westernmost component of a chain of nebulosities and stars extending eastward toward the W4 H II region (cf. Plate I). To date, there are no photometric studies of the three stellar clusters and thus their properties are not known. However, the stars in these clusters are comparable in brightness to the stars in the well-studied cluster surrounding IRS 5, suggesting that the clusters are composed primarily of low mass stars.
3.5. Determination of kinetic temperature, column densities, and virial masses
In Fig. 7, we show the (1,1), (2,2) and (3,3) inversion transition spectra for the positions labeled in Figs. 3&5. When the optical depth is small, the ratio of intensities of the HF satellites to the main peak in the (1,1) line is 0.28 for the inner and 0.22 for the outer group of HF components. For the (2,2) inversion line, this ratio is 0.06 for both groups. For all positions observed toward W 3 Main, the main component optical depths are . To determine the total column densities of toward the cores in W 3, we follow the approach of Wilson et al. (1993). The rotational temperature obtained from the inversion transitions (J',K') and (J,K) is expressed in terms of the excitation temperatures, , and the main-line opacity, , of the upper and lower transition (cf. Martin & Barrett 1978, Gaume et al. 1992, Wilson et al. 1993):
where J' and K' are the total and projected angular momentum quantum numbers for the upper excitation level, J and K the quantum numbers for the lower excitation level, E the excitation energy, g = (2J+1), and f the ratio of the main line intensity to the total intensity of all the HF components. For the most reliable estimate of one must use the metastable inversion transitions, i.e. J = K. In our case, we use the (1,1) and (2,2) excitation levels:
If the population is in LTE, characterized by a single rotational temperature as determined above, the total column density of can be determined as a func- tion of the column density in the (1,1) doublet. We have assumed that only the metastable levels are populated. A significant population in the non-metastable levels (J K) would require very high densities (n(H2) cm-3 ); it is unlikely that these conditions are typical for the extended cool ammonia gas we mapped toward W 3 SE, W 3 West and W 3(OH):
If the non-metastable levels are populated, then the tabulated column densities will be a lower limit. For a given , the population in each state will increase as 2J+1, until the energy E above the ground state becomes sufficiently large compared to that higher order terms may be neglected (cf. Mauersberger et al. 1988). Given the values of we determined for W 3 Main and W 3(OH), we only need to consider levels up to (7,7), and the resulting value of lies between 2.24 and 3.13 (cf. Wilson et al. 1993).
In Table 1, we give the main beam brightness temperature of the main components of the and (2,2) inversion lines measured toward the regions numbered in Figs. 3&5. The spectra for each position are plotted in Fig. 7. We also give the linewidths and central velocities, which are in agreement with those previously ob- served for C18O and C34S (Tieftrunk et al. 1995). The to ratios are used with Eq. 2 to determine for each position. We then used to determine the kinetic temperature (see e.g. Walmsley & Ungerechts 1983, Danby et al. 1988). From the integral over the entire line, including HF satellites and the tabulated values of we compute the beam averaged column densities of using Eq. 3.
For the ammonia gas toward W 3(OH), we find column densities of the order cm-2 and kinetic temperatures of = 27 K. These temperatures are significantly lower than those determined from other molecular tracers of this cloud (cf. Helmich 1996) and higher level transitions (Mauersberger et al. 1988). However, the tabulated values for are derived from the and (2,2) emission. This emission will by biased to cool gas which is extended over sizes of the beam or more. Since we find that varies little over the entire extent of the emission toward the W 3(OH) molecular cloud, we believe that the low level transitions presented here are tracing primarily an extended, cool component of the gas and do not reflect the physical conditions in the hot and dense W 3(OH) core detected in higher angular resolution observations (Mauersberger et al. 1988).
Masses for each of the cores can be derived from the virial theorem by assuming a spherical symmetry, uniform density and virial equilibrium. For the diameter of the cores, we use the full width at half maximum of the emission, and we take the linewidths from Table 1. Given an observed core diameter of 1 pc, we calculate a virial mass of 1100 for W 3 West. This is marginally smaller than the mass of 1400 determined from C18O (2-1) line emission by Tieftrunk et al. (1995) for this core. For W 3 SE, we use the observed diameter of 0.5 pc and determine a virial mass of 300 . Since the W 3(OH) core does not have a circularly symmetric morphology, we separate the extended cool emission mapped in into two components: a 1 pc core centered on the W 3(OH) H II region and a 0.65 pc region centered on the plume. Again, assuming a spherical symmetry, we estimate a virial mass of 1500 for the central core and an additional in the plume. We estimate the total virial mass of the W 3(OH) molecular cloud as derived from the emission to be . Note that a more centrally condensed distribution in any of these cores would result in a lower mass estimate. Using the virial masses and the observed diameters, we have determined H2 column densities in W 3 SE and W 3(OH) for comparison with the column densities (Table 1.)
In Table 2, we give the of the inversion lines from Fig. 7, measured toward the regions numbered in Figs. 3 & 5. We note that the (3,3) main beam brightness temperature is relatively strong toward the position of IRS 5 when compared to the weak (1,1) or (2,2) emission from this region.
Table 2. line parameters toward W 3
Relative abundances of can be determined toward each core given the H2 column density. In the case of the W 3 East and W 3 West cores, the C18O maps of Tieftrunk et al. (1995) and Roberts et al. (1997) can be used to determine N(H2) since the ratio of C18O to H2 is thought to be relatively constant (Dickel et al. 1996, Tieftrunk et al. 1995, Helmich et al. 1994, van Dishoeck et al. 1993). Since the density of the gas traced by C18O was found by Tieftrunk et al. (1995) to be two orders of magnitude higher than the critical density of the inversion transition, we expect the same volume of gas to be traced by the emission. To correct for the possible effect of beam dilution in the relative abundance determination, we have smoothed the data of Tieftrunk et al. (1995) to the FWHP beam of the measurements. In Fig. 8, we compare this smoothed C18O (2-1) emission to the emission toward W 3 Main.
The C18O emission originates from two spatially and kinematically distinct cores, W 3 East & West. In the smoothed, integrated intensity map shown in Fig. 8, the C18O emission from these two cores appears somewhat blended. The emission, however, clearly peaks toward W 3 West and no emission is apparent toward W 3 East in the inte- grated intensity map. However, from Table 1 and Fig. 7, it is clear that weak emission has been detected toward W 3 East, since the velocity of the weak emission is consistent with the velocity found from other molecular tracers toward W 3 East. In Table 3, we show the relative abundances derived from the and lines, using the smoothed C18O determined H2 column densities for W 3 East and W 3 West.
Table 3. line intensity ratios and relative abundances toward W 3.
We find that W 3 East shows an relative abundance roughly an order of magnitude lower than W 3 West or W 3(OH). We have also included the predictions of the chemical model of Helmich (1996), which were based on the same N(H2) column densities. While our observations agree with the predictions of these models for W 3 West/IRS 4, we derive a relative abundance an order of magnitude lower than predicted by Helmich (1996) for W 3 East/IRS 5. However, the reader should note that the observed abundances are averaged over a beam.
From the and data we estimate = 150 K in W 3 East. However, if the gas in W 3 East is significantly hotter than 150 K, then we may have underestimated the population of molecules in the higher energy states. Consequently, the derived relative abundance could underestimate the true relative abundance for W 3 East in Table 3.
The presence of hotter gas in W 3 East is indicated by the strong emission toward W 3 East (cf. Table 2). In Table 3, we list the and ratios for the integrated line emission. Toward W 3 East, the ratio is an order of magnitude higher than toward W 3 West and W 3(OH). The observed ratio would require values for of several hundred Kelvin, much higher than the = 150 K derived from the ratio. However, the derived from many other molecular tracers by Helmich (1996) are similar to the 150 K derived from the & (2,2) emission. Furthermore, we believe it is unlikely that the entire 800 W 3 East core can be heated to a temperature of several hundred Kelvin. Instead, we find it more likely that the is tracing localized regions of exceptionally hot gas. This hot component may be heated by radiation from the embedded stars or by shocks produced by stellar winds and outflows (Choi et al. 1993). The acceleration of the gas by shock waves may also explain the higher linewidths of the lines compared to the or lines. Alternatively, a group of masers, similar to the masers detected by Kraemer & Jackson (1995) in NGC 6334, may be the source of the strong lines. An unusual ortho- to-para ratio may also result in the observed line ratios and an underestimate of the total density, but there are currently no grounds for assuming such an unusual ortho-to-para ratio.
Future confirmation of a lower abundance in W 3 East will require a more detailed study of the excitation of ammonia using observations of the line as well as interferometric observations in the line. However, support for a lower abundance in W 3 East is found in recent, resolution VLA observations in the and line of W 3 East and West, from which Tieftrunk et al. (1998) derive values for the relative abundance consistent with those listed in Table 3. Thus, on the basis of the current data and recently analyzed VLA data (Tieftrunk et al. 1998), we believe a small relative abundance is the most likely reason for the weak and emission toward W 3 East and we will discuss this in Sect. 4.2.
Since similar C18O measurements are not available for W 3(OH) and W 3 SE, we used the virial column density to de- rive a relative abundance for these cores. We find that W 3 SE has a relative abundance consistent with that of W 3 West, toward W 3(OH) it is roughly twice as high. We note that our measured relative abundance for W 3(OH) is still lower than that predicted by Helmich (1996). This may be due to the larger beam of the 100-m tracing the extended, cool component and the fact that Helmich (1996) derived abundances assuming hot core chemistry for the hot compact W 3(OH) core observed by Mauersberger et al. (1988).
In Table 1 and Fig. 7, we present the spectra and line parameters toward a position of weak emission in W 3 Main at R.A. , Dec. (1950.0). The line toward this region is substantially narrower and weaker than the emission from the dense cores. Although the derived of 25 K is similar to the in W 3 SE and W 3(OH) (but much lower than in W 3 West and W 3 East), we find that the column density is less than a quarter of that found toward W 3 West or W 3(OH). Due to the small observed linewidth, the virial column density derived for the region is also much smaller than those found in the cores. Interestingly, we find that the virial mass and column densities are consistent, given a relative abundance of . This value of the abundance is similar to that derived in W 3 West and W 3 SE, suggesting that the virial column density is indeed close to the actual H2 column density. From the measured diameter of the region, one arcminute or 0.72 pc, we estimate a density of cm-3 , the critical density of the inversion transition. Thus, we find that the measured linewidth, column density, and size of the region are consistent with a virialized (and consequently, gravitationally bound) clump with an abundance similar to that measured in W 3 West and a density just sufficient to excite the inversion transition. We have found that the extended weak emission regions apparent north of W 3 Main, east of W 3 SE, and north of W 3(OH) show similar characteristics. In general, the and (2,2) lines toward these regions are extremely narrow ( km s-1 ) and weak (K & K, respectively). The central line velocities of the clumps are similar to the velocities of the nearest dense cores.
© European Southern Observatory (ESO) 1998
Online publication: July 27, 1998