3.1. Radio continuum of W3
Fig. 1 shows a natural weighted image of the continuum emission of the entire W3 region at 4.9 GHz with an angular resolution of . The individual H II regions have been labeled following the scheme introduced by Wynn-Williams (1971) and Harris & Wynn-Williams (1976). The northern part of this region, the W3 core, consists of the compact H II regions W3 A, B and D, the ultracompact sources C, E and F, and the diffuse sources G and H. There is very little diffuse continuum emission between these sources, but there is dense molecular gas (Hayashi et al. 1989, Tieftrunk et al. 1995, Roberts et al. 1996). The southern part of W3 consists of the two diffuse regions W3 J and K, with the former superimposed on the larger low brightness ring-like H II region G 133.7+1.1 extending to the south (Roelfsema et al. 1987).
We produced uniform weighted images of the continuum emission of the W3 core at 4.9 GHz, 14.9 GHz and 22.5 GHz with angular resolutions of , and , respectively, by combining data obtained from the VLA D,C,B and A array configurations. The peak and total flux densities of these sources at different wavelengths are given in Table 1. In proceeding from lower to higher frequencies, the largest object that can be imaged reliably depends on the minimum baseline spacings used for full synthesis observations. For the VLA at 1.3 cm this is about , , and for the A,B,C and D configurations respectively (for a high declination source). Our images do not properly reproduce source structures larger than these values. Thus, one has to keep in mind that the VLA may record significantly less flux for the extended H II regions at 2 cm and 1.3 cm. Observations by Wood & Churchwell (1989) show that, on the average, ultracompact H II regions become optically thick for cm.
Table 1. Radio continuum flux densities at 6, 2 and 1.3 cm for the compact and ultracompact H II regions
Table 2 lists the continuum optical depths , the peak emission measures EM, source diameters d, peak and mean electron densities , excitation parameters U, the mass of ionized gas M(H II) and the number of Lyman continuum photons . The estimated zero-age main-sequence (ZAMS) spectral type of the exciting stars, assuming ionization by a single star and no absorption of ionizing photons by dust Panagia (1973), are also given. The electron temperatures, , of the H II regions are taken from Roelfsema & Goss (1991) and our own H66 recombination line data. Data with good signal-to-noise ratios were obtained only toward the regions with highest electron density, namely W3 B, C and F. Our results are given in brackets. The only significant discrepancy for is found for W3 F; we calculated the resulting source parameters using both electron temperatures. Source properties have been estimated using the formulae of Panagia & Walmsley (1978) and Turner & Matthews (1984) for a distance of 2.3 kpc.
Table 2. Radio continuum properties of the H II regions in W3
3.2. H66 radio recombination line emission of W3 at 22.369 GHz
Together with the continuum emission at 22.5 GHz, we imaged the entire W3 core at 22.369 GHz, the wavelength of the H66 recombination line emission. For those regions with good signal-to-noise ratios, we determined electron temperatures under the assumption of local thermodynamic equilibrium (LTE). Shaver (1980) had proposed that there is a relationship between the emission measure of an H II region and the frequency of the recombination line emission at which the derivation of the electron temperature based on the assumption of LTE conditions is valid, i.e. = . For the H66 recombination line emission this corresponds to an EM of for the H II regions. Shaver (1980) finds / for regions with EM an order of magnitude higher than this limit. From the EMs measured toward W3 this indicates that, for the H II regions observed, stimulated effects are not large. Thus the LTE approximation is valid and is close to the the actual electron temperature, . For W3 B, C and F we find values of K, 8000 K, and 10000 K. The value for W3 F is significantly higher than the previous value determined by Roelfsema & Goss (1991).
Fig. 2 shows images of the H66 recombination line emission observed toward W3 B for individual velocity channels. Emission toward the filamentary tail extending to the NW is clearly associated with thermal ionized gas. From the first-moment image (intensity-averaged line-center velocity) and the position-velocity cut along the axis shown in the 6 cm image, the filamentary structure has a more positive velocity. The compact object IRS 3a (Megeath et al. 1995) is associated with the double-peaked brightness distribution (N to S) in the southern bulge of W3 B.
In Fig. 3, we show images of the H66 recombination line emission observed toward W3 C for individual velocity channels. As toward W3 B, we detect a double-peaked brightness distribution (NE to SW). A position-velocity diagram shows the velocity for the brightened edges of the teardrop-shaped emission (R.A. offset and ) to be 20 km s-1 blue-shifted in respect to the center's emission ().
In Fig. 4, we analyze the H66 recombination line data of W3 F associated with the compact source IRS 7. Here, we present the peak of spectra along the axis indicated. From this presentation we see that the bright rim of W3 F appears to be moving -5.5 km s-1 relative to the tail region. Measurements of C18 O show that the ambient molecular gas has a of -39 km s-1. Thus, the ionized gas in the head of W3 F is moving with a velocity of -4.5 km s-1 relative to the molecular gas, whereas the tail gas is almost at rest.
3.3. Spectral indices
We derived spectral indices from our 6 cm, 2 cm and 1.3 cm data. In order to compare the flux densities at different frequencies for the H II regions, the data has to be convolved to the same restored beam. Furthermore, a direct comparison of large scale structures is limited by the largest angular scale reliably imaged with a given configuration. Errors could be introduced by the lack of short interferometer spacings, since these are sensitive to large spatial scale sizes. For the 6 cm, 2 cm and 1.3 cm data the shortest spacings are in D configuration, which reliably images spatial scale sizes up to , and , respectively. Thus, combining the data obtained from B, C and D array configurations we are sensitive to large scale structures with a high resolution.
Fig. 5 shows a spectral index image of W3 A, B, F and M obtained from the spatial alignment of the 2 cm and 1.3 cm images. The 1.3 cm image was produced from the combined C and D continuum channel. This has been convolved with the restored beam of the 2 cm image taken in the C and D configuration. Thus, the angular resolution is . Since the largest spatial scales in W3 A are , short interferometer spacings are not a significant problem for this spectral index map.
The compact and ultracompact H II regions show spectral indices in the range (). This is consistent with sources with a normal thermal spectrum with positive 's being an indication of large optical depth. Gaume et al. (1995) showed that asymmetrically bright, arc-like, and edge-brightened H II sources exhibit spectral index gradients which closely follow the continuum intensity gradient. This correlation can be seen toward the H II regions W3 A, B and F. These gradients are spatially correlated with the H2 density gradients found in the molecular gas (Hayashi et al. 1989, Oldham et al. 1994, Tieftrunk et al. 1995) as the result of higher electron densities where neutral gas is being ionized (Gaume et al. 1995).
For the spectral index of the radio continuum region W3 M toward IRS 5 we find between 2 cm and 1.3 cm. We also find a spectral index of for the source W3 Ca near IRS 4. If we compare the peak flux densities of the individual hypercompact continuum regions, given in Table 3 (obtained at maximum resolution for each given frequency), we find spectral indices of between 6 cm and 1.3 cm.
Table 3. Radio continuum flux densities at 6, 2 and 1.3 cm for the hypercompact radio continuum regions
© European Southern Observatory (ESO) 1997
Online publication: July 3, 1998