STRUCTURE AND COMPOSITION OF MOLECULAR CLOUDS WITH CN ZEEMAN DETECTIONS I: W3(OH)

Figure 1 shows integrated line maps of the five species, while Figure 2 shows channel maps. The continuum source was removed from the spectral line maps by averaging several offline channels together and subtracting that from each individual channel. Toward the continuum source, we see three of the species strongly in absorption. Besides the Turner–Welch object and the continuum source, there are two other regions that are notable in CN. The western side of the map contains very diffuse CN emission, while the eastern side of the map shows a compact and complex CN emission source. In order to compare the CARMA CN data with the IRAM CN data (Falgarone et al. 2008), we also generated a map where the CARMA channel data were convolved with a 23'' Gaussian beam, effectively smoothing the CARMA data to match the IRAM resolution (see Section 4.2). By comparing these maps, there is evidence that both spectra are dominated by diffuse emission.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Compared to the other molecules, the CN emission appears to be diffuse but "clumpy." Under closer scrutiny, it is much more complex. Our 8 MHz spectral window is centered at 113.490982 GHz, the frequency of CN's strongest hyperfine component. This frequency and window size was chosen to allow us to simultaneously image a second hyperfine component of the CN N = 1–0 transition at 113.488 GHz in the same spectral window. These two hyperfine components are separated by 7.9 km s⁻¹. Since this window covers approximately 21 km s⁻¹ of velocity space, we expect to fully resolve and image both lines. Multiple velocity components in the image make it difficult to determine which hyperfine component some emission belongs to. However, we have been able to identify at least three distinct velocity components, representing separate regions around W3(OH) (Figure 3).

Zoom In Zoom Out Reset image size

The northwest and southwest components appear to be very diffuse. They are centered at approximately −47.25 km s⁻¹ and −44.5 km s⁻¹, respectively. It is difficult to completely isolate these two velocity components due to blending and overlap of the hyperfine lines. The maps in Figure 3 reflect some of this "blending." It occurs because the second hyperfine component of one velocity partially overlaps the primary hyperfine component of the other velocity. For example, the second and third maps in Figure 3 are cross-contaminated. This is easily visible when the positions of the emission in the two maps are compared; they have a significant spatial overlap region.

The northeastern component is more complex. In addition to containing diffuse emission, multiple denser clumps are visible. The significance of some of these clumps is not clear from the integrated line map. Between the larger southernmost clump and the smaller, northern clump are a string of small, presumably unresolved clumps only seen in a single channel each. When viewing an averaged line map, many of these features are spatially smoothed and are not prominent (as in Figure 3). The densest portions of this region are concentrated in two lobes, not unlike the typical signature of an outflow. Figure 4 shows channel maps of the 15 channels around the strongest hyperfine component. There does not appear to be any blending as was seen in the south–west component, so we are confident that this channel map is not contaminated. An averaged contour plot (Figure 5) shows the physical comparison between the two "lobes" of the emission. The top (redshifted) lobe is averaged from the six channels from −47.6 km s⁻¹ to −49.2 km s⁻¹, while the bottom (blueshifted) lobe is averaged from the six channels from −49.2 km s⁻¹ to −50.8 km s⁻¹. While this does have the signature of a typical outflow, it is not accompanied by any coincident emission from any of the other species, nor is there any accompanying IR emission source visible in publicly available IR catalogs. Due to the extent of the low-level emission around the whole region, and lack of evidence of an outflow generating source, it is more likely that this is not an outflow, but rather bulk rotation of a clump of gas.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The maps for HCN, HCO⁺, C¹⁸O all appear to be tracing the same material, which does not coincide with either the CN or N₂H⁺ emission (see Figures 6–8 for channel maps). However, the peak position of these spectral lines coincides with the HCN detection of the Turner–Welch object first reported in Turner & Welch (1984). At 25 resolution, we do not appear to be resolving the object; previous high-resolution (subarcsecond) studies have been conducted in order to resolve this object in the continuum (Chen et al. 2006), which resulted in the detection of multiple cores. Previous HCO⁺ studies (Wink et al. 1994) had similar spatial resolution to our data, however, our data have significantly better S/N and sample material that was resolved out in the previous study. In the HCO⁺ spectrum of the Turner–Welch object, a weak, secondary velocity component at −44 km s⁻¹ is visible (Figure 9). This velocity component is not reported in any previous studies. The HCN spectra toward the Turner–Welch object shows an unexpected peak at an apparent velocity of −39 km s⁻¹ (Figure 9); comparing relative positions of these two peaks, the anomalous HCN peak would be the F = 0–1 transition of a secondary velocity component at −44 km s⁻¹. The other hyperfine lines from this velocity component are being masked by the hyperfine lines from the primary velocity component. This can only happen if the separation between the hyperfine lines is approximately the separation between the two velocity components. The C¹⁸O line may also show this secondary component, however, it cannot be confirmed due to the relatively low S/N of the C¹⁸O emission.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The C¹⁸O map does not show any absorption toward the continuum source, unlike HCN, HCO⁺, and CN. This is consistent with observations done by Wink et al. (1994), in which they surmised that the line excitation temperature of HCO⁺ is significantly lower than the kinetic temperature of 90 K, but the C¹⁸O excitation temperature is not low enough to produce absorption. However, our observed absorption in other lines is significant and can yield information about the regions around the continuum source. The HCO⁺ spectra, seen in absorption (Figure 9), looks similar to an inverse P Cygni profile which could indicate an expanding shell-like structure or a strong stellar wind. The multiple hyperfine components of HCN make it difficult to tell which gas is causing the absorption; however, it has the same central velocity and line width as the HCO⁺ absorption indicating that both features are being generated from the same region. The CN absorption also appears to have similar spectral features as HCN; however, the redshifted CN emission feature is at a level of less than 1σ of the noise level and cannot be considered as a positive detection.

Another notable feature of the continuum absorption lines is their shape. There is a double trough shape seen in HCO⁺, the strongest hyperfine component of HCN, and in CN (these spectra are centered on the strongest hyperfine component). There is also some evidence of this feature in the other hyperfine components; however, they are weaker and, as a result, noisier. According to the results of Wink et al. (1994), within the UCHII region of W3(OH) there is an embedded O7 star. It is very likely that W3(OH) is a region surrounding a bright massive star embedded in a dense cloud with a strong solar wind clearing out the region around the star. In looking at a velocity moment map in HCO⁺ (HCO⁺ was chosen due to its lack of hyperfine components and contamination), there appears to be an east–west velocity gradient on the order of 3 km s⁻¹. Since W3(OH) is unresolved in these maps, higher resolution data of W3(OH) are required to quantify and describe this effect (see below).

Figure 10 shows N₂H⁺ channel maps. If N₂H⁺ were tracing the same dense material as CN, we would expect to see very similar emission (and absorption) spectra. Most notable is the lack of an absorption feature toward the continuum source, or alternatively, of any emission in N₂H⁺ at the continuum position. Also, the rotation feature in CN noted above is not seen in N₂H⁺. The N₂H⁺ emission appears to be concentrated in the southwestern region of the map, roughly coincident with one of the velocity components of CN; however, this is where the similarity ends. In the N₂H⁺ spectra (Figure 9), only three peaks appear when there are seven hyperfine lines within our window. The central peak, at about −47.5 km s⁻¹, contains three hyperfine components which, having only 0.487 MHz (1.56 km s⁻¹) between them, are blended together. The peak centered at −42.5 km s⁻¹ contains another three hyperfine components whose transition frequencies all lie within 0.432 MHz (1.39 km s⁻¹) and would also appear blended. The last hyperfine component is seen at −56 km s⁻¹ and is the only one that is not blended. Our N₂H⁺ map contains some similarities to the ammonia (NH₃) map presented by Wilson et al. (1993) and Tieftrunk et al. (1998). The major difference between the two is the lack of N₂H⁺ emission around the Turner–Welch object. The major similarity comes from an east–west elongated clump present in the northeast corner of both the N₂H⁺ emission and the NH₃ emission. This stands out since none of the other molecules show any sign of emission from this region. The closest emission is the north lobe of the possible CN outflow. It is unclear if these regions are related.

Zoom In Zoom Out Reset image size

It is possible that the regions of CN emission much more closely match the regions of N₂H⁺ than is readily apparent, even though the spatial peaks do not coincide. Since CN is significantly weaker than N₂H⁺, it is possible that emission, particularly in the lower west side of the map that is detectable in N₂H⁺, is significantly below the noise threshold in CN. To emphasize this fact, which may not be noticeable in the individual channel maps, we constructed a channel map with 20% contours from all five detected species (Figure 11). Most noticeable in the −48 km s⁻¹ and −47 km s⁻¹ panels, CN (with several other of the species) does seem to trace much of the N₂H⁺ emission. In the −51 km s⁻¹ and −50 km s⁻¹ panels, it appears that the N₂H⁺ emission borders on the regions with emission in the other species. If we assume that the two molecules do trace the same density gas, it is quite possible that some of these regions are undergoing a chemical reaction that selectively annihilates N₂H⁺.

Zoom In Zoom Out Reset image size

From the observations described above, several significant features warranted even higher resolution observations. The continuum source is seen in absorption in some species; however, it remains unresolved. CARMA's B-array should be able to probe the region at significantly higher resolution than in our previous data. This will additionally allow us to study the velocity structure of gas near the continuum source. A velocity map of HCO⁺ shows an east–west velocity gradient of ∼3 km s⁻¹. HCO⁺ was chosen specifically since it does not have multiple hyperfine components which, due to blending, could pollute the velocity map, and shows absorption toward the continuum source. In addition, with these data we can see if we can resolve the individual cores of the Turner–Welch object. Previous high-resolution continuum maps have shown evidence of multiple cores within the Turner–Welch object which we would expect to also see in corresponding high-resolution spectral line maps. We previously chose to map the region with the C, D, and E arrays to study large-scale features; with B-array alone we are able to map features that require higher spatial resolution. If we combine the B-array data with the rest, we will produce a map with a synthesized beam that is larger than the beam of the B-array data alone, and larger than some of the smallest features, obscuring them. In addition, the necessity to self-calibrate the data and issues with gain calibration (discussed in detail below) make a separate analysis more informative.

Self-calibration results in several issues that need to be overcome. W3(OH)'s physical size is 0.01 pc in diameter (Kawamura & Masson 1998) which corresponds to an angular size of ∼1''. This means that many of the longer baselines in B-array will be resolving the W3(OH) continuum source. Data from these baselines will make it difficult to self-calibrate the data as it assumes a point source model for the source (or requires an accurate model of the source features which we do not have). Running self-cal on the full data set (including resolved baselines) generates a solution that is appropriate for the data up until a UV-radius of 130 kλ, at which the solution fails. This UV-radius corresponds to a physical size of

$$\begin{eqnarray*} \Theta & \sim & \frac{\lambda }{D} \\ & \sim & \frac{\lambda }{2 * \mathrm{UVr}} \\ & \sim & 0\farcs 79. \end{eqnarray*}$$

Therefore, running self-cal on the full data set assuming a point source will result in large errors in the self-cal solution due to long, resolved baselines. If we "cut" the data set so we only use the data with UV-radius <130 kλ for use in calculating the phase solution, we will not have this issue.

The self-cal technique fits the phase center of the data to the brightest point source in the map. Our C, D, and E array data is slightly offset in position from the pointing center, so this will result in a slight position offset between the two maps. This can be corrected for by manually fitting and entering the center position of the continuum source from our C, D, E maps, however, this is not necessary since we will not be directly combining our B-array data with the rest, since the resulting maps at full angular resolution would have a very low S/N.

This self-calibration technique also poses several problems with regards to flux and amplitude calibration. Several dishes primarily have long baselines, with only a few baselines that fall under our UV-radius limit of 130 kλ. While this worked well for phase calibration, the results of amplitude self-cal was very poor. Many of the amplitude gains, which should be flat and close to unity, were noisy and had many data points corresponding to unphysical gains that directly translates to poor image quality. The few remaining baselines also tend to have low S/N which further increases the problem.

In the attempt to apply correct gain calibration, several other techniques were tested to amplitude calibrate the data. We attempted to amplitude calibrate our data using the dedicated phase calibrator measurements and transferring the resulting solution to our source. This also failed, and while it produced better results than the amplitude self-cal technique described above, the high-resolution, low-intensity features were washed out in the final maps. This is most likely due to the gain calibrator being 10° from our source, far enough that the telescopes are viewing different enough atmosphere to negatively affect the calculated gains. This problem does not affect our C, D, E array data because the maximum baseline length is significantly smaller in these arrays and are not as sensitive to atmospheric fluctuations as the longer B-array baselines. Since these several techniques turned out poor results, it was decided not to use amplitude (flux) calibration on the B-array data, which is not an uncommon result when using self-cal. This is another reason why we do not combine this data with our C, D, E data as we would need an accurate gain solution to provide proper weights to combine the data sets.

Figure 12 represents the combined C, D, E array and B-array only continuum maps. Each map was produced using multi-frequency synthesis on two 500 MHz windows, one in the N₂H⁺ band (average frequency of 91.2 GHz, not shown), and one in the CN band (average frequency of 112 GHz, shown). In Figure 12(a), neither the Turner–Welch object nor W3(OH) are resolved. The peak flux of W3(OH) decreases by a factor of two from 91.2 GHz to 112 GHz, while the center of the Turner–Welch object increases in flux by a factor of two over the same range. In the B-array only data (Figure 12(b)), we resolve two individual clumps within the Turner–Welch object and slightly resolve structure within W3(OH) (note the off-center position of the highest level contour in Figure 12(b)). This structure supports the findings of Chen et al. (2006). We additionally see the same change in relative fluxes in the B-array maps as we saw in the C, D, E array maps, over the same frequency range.

Zoom In Zoom Out Reset image size

As stated above, W3(OH) itself is a UCHII region with an embedded O7 star. Our data show that several of the molecular tracers we have observed can be seen in absorption toward this source. The nature and structure of this colder absorbing gas may be able to give us some insight into the properties of W3(OH). Figure 13 shows three spectra sampled across the continuum source in HCO⁺. HCO⁺ was chosen for this since it does not have any hyperfine components that could cause line profile confusion. C¹⁸O, while also not having any hyperfine components, was not chosen to trace this feature as it is not seen in emission or absorption toward W3(OH). The composite emission and absorption spectra are consistent with a dense, hot region (containing HCO⁺) surrounded by one or more cold, less-dense layers which also contain HCO⁺. The absorption feature is evidence of cold, optically thick gas which is at a velocity of −46 km s⁻¹ and has an FWHM of 4 km s⁻¹. In looking at non-continuum subtracted spectra (Figure 13(b)), this absorption line is saturated and extends completely down to zero flux. This implies that this cold region is in front of the emission region and the continuum source. The absorption feature is also seen in CN and HCN, however, it is seen slightly in emission in C¹⁸O. The C¹⁸O emission feature is also centered at −46 km s⁻¹ and has an FWHM of 4 km s⁻¹, the same as the HCO⁺ absorption. It is very likely that it is from the same region, implying that the gas density is sufficient to raise the excitation temperature of the C¹⁸O line above that of the brightness temperature of the continuum source so the line appears in emission, while the higher critical densities of the HCO⁺, HCN, and CN transitions lead their excitation temperatures to be less than that brightness temperature, so they are seen in absorption. To note: Figure 13 is resolution limited since it is formed with C, D, and E array data and its three positions only cover an area of 1.5–2 beamwidths. This was part of the motivation to acquire B-array data as the same region would cover 5–8 beamwidths in a B-array map.

Zoom In Zoom Out Reset image size

From the B-array spectra (Figure 14(c)), we only see HCO⁺ in absorption against the continuum emission. The lack of emission either means that it is being resolved out at this high spatial resolution, or that the beam is completely filled by continuum emission, so we see no emission from the sides of the beam. In either case, this means that the emission does not come from the same gas that produces the absorption. To support this, we produced spectra that only includes E-array (largest beam size, shortest baselines) in order to inspect large-scale emission. The E-array data (Figure 14(a)) shows emission toward the continuum between −43 km s⁻¹ and −50 km s⁻¹ with a self absorption feature at −46 km s⁻¹. This high velocity gas is not seen in C¹⁸O and therefore seems to be hot, optically thin gas that is behind the continuum source.

Zoom In Zoom Out Reset image size

Therefore, we propose a multi-layered model of the region around W3(OH). Behind the UCHII region is a high velocity, hot, large spatial scale, optically thin region which appears to peak in intensity toward the Turner–Welch object (and may very well be associated with it). The UCHII region is optically thin, and has a small spatial scale of about 07 in diameter which is extremely bright in the continuum, enough such that the narrow band 2 MHz windows are contaminated by continuum emission. In front of the continuum is a very cold, optically thick region at a velocity of −46 km s⁻¹ which we primarily see in absorption, but we see slightly in emission in C¹⁸O.

Our maps are comprised of C, D, and E array data and do not cover the entire UV space. We do not have zero-spacing data which would be required in order to reconstruct the most accurate map of the region. The IRAM data consists of only a single pointing and its beam does not cover the full area of the CARMA maps. In order to quantify the amount of flux that could be resolved out by using CARMA, we examined the single dish data taken by Falgarone et al. (2008). Since the IRAM 30 m telescope has a resolution of 23'', we smoothed our CARMA maps to match by convolving the data with a 23'' Gaussian beam (Figure 15). Spectra (Figure 16) were then extracted from the position corresponding to the IRAM 30 m pointing center, as well as two other positions centered on the northeastern and northwestern components. According to the IRAM technical documentation (IRAM 2009), the main beam efficiency is 78% for the band that includes CN. To compare the flux values, an estimate of CARMA's main beam efficiency is needed (White & Zauderer 2008). CARMA does not have published main beam efficiency measurements, however, it can be estimated from single-dish aperture efficiency measurements which vary from 55% to 70% depending on the individual dish. From this, we estimate that the main beam efficiency (per dish) would be ∼70%–80%. By adopting a value of 75%, approximately the same as IRAM's, we can directly compare the IRAM and CARMA spectra. The IRAM peak was 2.2 K, while CARMA's was 1.4 K. This means that the peak of the CARMA spectra was at 64% of IRAM's (the area of the CARMA spectra is 48% of the IRAM spectra). CN is also seen in strong absorption toward the central source (W3(OH)), however, this absorption feature is not present in any of the IRAM data. This is possible for several reasons. First, the relatively small spatial size of the absorption feature covers a small fraction of the total IRAM beam. Any measured absorption would be diluted across the entire beam and masked by the significantly greater emission. In addition, the IRAM data baseline fitting procedure could contribute to hiding the presence of a weak absorption feature.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In regions with standard CO abundances (i.e., regions that are not depleted in CO), one of the formation mechanisms of HCO⁺ is the destruction of N₂H⁺ by CO (Bergin et al. 2002; Jorgensen et al. 2004). Since there is significant support that CN remains in the gas phase at densities greater than 10⁵ cm⁻³ (same as N₂H⁺) we would expect CN and N₂H⁺ to have a very similar spatial distribution, however, there are some regions and velocities where we would expect to see more N₂H⁺ emission than we do. The particular reaction that is thought to occur is

$$\begin{equation} {\rm N_2H^+ + CO} \rightarrow {\rm HCO^+ + N_2.} \end{equation} \tag{ 1 }$$

If we compare the regions of strong N₂H⁺ and HCO⁺ emission, we would expect them not to be coincident. In addition, we expect CO to be depleted in regions of strong N₂H⁺ emission, as the presence of CO would cause the formation of HCO⁺, while regions with significant CO emission must be weak in N₂H⁺, or the reaction between these two molecules would occur.

In the region surrounding the Turner–Welch object we can see considerable emission from both CO and HCO⁺, however, there is no N₂H⁺ within our detection limit. This depletion of N₂H⁺ could be caused by the amount of C¹⁸O present along this sightline. Since there is significant CN emission from this region, we would also expect to see N₂H⁺. The presence of CN and the relative strengths of HCO⁺ and C¹⁸O is significant evidence of this reaction occurring and that N₂H⁺ is being consumed by this reaction.

Above, we showed that there was a cold, dark region of gas toward the continuum source where HCO⁺ is seen in strong absorption along with CN and HCN. Discounting the contribution from the continuum source, C¹⁸O is seen slightly in emission and N₂H⁺ is seen slightly in emission and absorption. The presence of CN absorption implies conditions where N₂H⁺ should also be strongly detectable. If we assume that the primary formation mechanism of HCO⁺ in this region is by the CO and N₂H⁺ chemical reaction, the relative depletion of these two chemical species in this cold dark cloud can be understood.

In regions with conditions similar to that of W3(OH), CN may be a more reliable tracer than N₂H⁺ to directly sample regions at high densities. However, since the strongest CN hyperfine transition has significantly weaker line emission than the strongest hyperfine transitions of N₂H⁺, it would require significantly greater telescope time to achieve a comparable S/N. In our case, these sets of maps show not only that this chemical reaction is occurring, but is happening most strongly both along the sightlines associated with an active star-forming clump, and in a cold cloud in front of an extremely bright UCHII region.

Wilson et al. (1991) developed a model of the region surrounding W3(OH) based on C¹⁸O, C³⁴S, and methanol (among others). They proposed a multi-layer model of low-density molecular gas surrounding a high-density molecular core. This low-density cooler region extends in front of W3(OH) as well. Self-absorption in their CS data supports this model. In addition, the v_LSR of the self-absorption was up to 2 km s⁻¹ more positive than the v_LSR of the hotter emitting region. This implies that the cloud envelope is contracting relative to the cloud core. We see similar offsets in the E-array spectrum of HCO⁺ seen in Figure 14(a), however, the offset appears to decrease at higher angular resolutions (Figure 13).

We can compare the range of radial velocities at which we see molecular material. Wilson et al. (1991) data showed that C¹⁸O peaks toward W3(OH) at a radial velocity of −46.6 km s⁻¹, and peaks toward the Turner–Welch object at −48 km s⁻¹. We see the same radial velocities in our C¹⁸O data (even though our data set was at 3 mm, and theirs was at 1 mm). For other species that trace related density regimes (HCN and HCO⁺), we see them peak at a velocity of −47 km s⁻¹ toward the Turner–Welch object. These species are seen strongly in absorption toward W3(OH), which makes it difficult to determine its radial velocity. These data show that the region around W3(OH) is extremely complex.