LOOKING THROUGH THE GALACTIC PLANE: IMAGING COLD DUST TOWARD ℓ = 44°

Dust grains are ubiquitous in the interstellar medium. Their existence was first established in the 1930s when it became clear that the extinction of starlight depends on wavelength. Even earlier, E. E. Barnard had cataloged (Barnard 1919) the extensive obscuration in regions of Ophiucus and Scorpius and hinted at its nature. The preferential scattering by dust at optical wavelengths results in the blue colors of reflection nebulae and the reddening of starlight when viewed through intervening dust. On larger scales, galaxies viewed edge-on commonly exhibit a very substantial equatorial band of obscuration resulting from dust.

Large-scale surveys of the Galactic plane (e.g., Jackson et al. 2006) show substantial differences in the distributions of the gas, dust, and molecular components. Early observations already revealed that the molecular gas distribution (as mapped via the CO (1-0) transition) is considerably different from that of the neutral hydrogen (H i). Subsequent surveys of CO showed that the molecular gas is concentrated in a ring at a distance of 5 kpc from the Galactic center (Clemens et al. 1988). The relatively dense gas in this "5-kpc ring" greatly facilitates star formation, and as a result is the source of virtually all of the associated energetic activity in the Milky Way, such as supernovae and ionized gas clouds.

The present data set includes a wide range of conditions in the interstellar medium, encompassing molecular clouds in cold quiescent regions to warm star-forming regions. In particular, it includes two very significant molecular cloud regions: W49, one of the most massive and energetic star-forming regions in the Galaxy, and a group of dense clumps at longitude 45.5.

The relationship between the molecular and dust components of interstellar clouds and their influence on the evolution of these clouds as they collapse under self-gravity are very current topics in astrophysics (see, e.g., Flower et al. 2005). These issues bear on the initial conditions for star formation in particular. As the contraction of a cloud proceeds toward collapse, many changes take place, including the deposition of volatiles onto grain surfaces ("freeze-out"). The densities at which these processes take place are quite low (e.g., several times 10⁴ cm⁻³). Under these conditions in the most extreme cases the only remaining free molecular material may consist of very simple species such as H₂D⁺. Thus, it is useful to study the contrast between the (possibly depleting) CO emission and the submillimeter emission produced by the dust itself.

Continuum emission from interstellar dust particles at millimeter/submillimeter wavelengths is optically thin under all realistic circumstances. Hence at these wavelengths the observation of this emission in any given direction provides a census of cold (that is, for temperatures typically within a factor of 2 of 20 K) dust along that entire line of sight through the Galaxy. Observationally, a negative aspect is that the signals are usually quite weak. In addition, water vapor in the terrestrial atmosphere both attenuates the signal and contributes substantial noise. As a result, observations such as those described in this paper require dry and stable atmospheric conditions.

In this paper, we report on bolometer array observations at two wavelengths, 850 and 1200 μm, of a segment of the inner Galactic plane, centered at galactic longitude 44°. Images at these wavelengths of individual star-forming regions, dust clouds, and external galaxies are well represented in the literature, but with the special exception of the Galactic center (Pierce-Price et al. 2000) the present work is the only one to date to address the general nature of emission at millimeter and submillimeter wavelengths from cold dust in the Galactic plane. We note, however, that in results yet to be formally published, Nordhaus et al. (2008; and earlier related references) report on an extensive survey of the Galactic plane at 1.1 mm wavelength with the Bolocam array.

Velocity information, coupled with a model of the rotation of the Galaxy, is essential to establish distances to the continuum emission sources. Spectral line observations obtained in this work provide much of this information, and consist mostly of either small images, or five-point samples, of approximately half of the dust clumps detected via the 850 μm mapping observations. As these data were obtained in the ¹²CO and ¹³CO (3–2) transitions, they are compatible with the 850 μm continuum image in terms of angular resolution, and together these two data sets form the foundation of the results presented in this paper. We use ¹³CO (1–0) data from the Boston University-FCRAO Galactic Ring Survey (GRS; Jackson et al. 2006) to supplement the molecular line data where CO (3–2) data are not available from our own observations.

The observations of continuum emission described here incorporate complementary observational techniques employed by two different instruments on separate telescopes. The results for both data sets appear in this paper, and we describe in detail the observation and analysis of the 850 μm data set. We anticipate that the data reduction of the 1200 μm data set and the relative merits of the two techniques will be discussed elsewhere; hence we include only a relatively brief description of these latter data, and the associated technical aspects, in the present work.

The observations cover an area of 2 deg² in one contiguous region 4° × 05 in extent, centered on galactic coordinates ℓ,b = 44.2154, −0.0248, at a position angle of −642. The region was selected with specific reference to the portion of the GRS described by Simon et al. (2001). The tilt of the observed region with respect to the galactic plane was adopted in order to include two well-known star-forming regions (G45.1/G45.5 and W49), and the overall distribution of gas and dust in this part of the Galaxy. At this galactic longitude the line of sight passes through two major galactic spiral arm features at velocities (with respect to the local standard of rest (LSR)) of about 10 and 60 km s⁻¹, corresponding to "near" and "far" kinematic distances of approximately 1 and 8–12 kpc, respectively.

2.1. Observations and Data Processing at 850 μm

The 850 μm observations were carried out at the James Clerk Maxwell Telescope (hereafter, JCMT), situated close to the summit of Mauna Kea, Hawaii, using the Submillimeter Common-User Bolometer Array (SCUBA; Holland et al. 1999) receiver during two periods between 2003 July 29 and August 1 and 2004 March 30 and April 5. In initial preparation for this program we made a test observation in 2000 centered on the bright star-forming region G34.3+0.2, the results of which, shown in Figure 1, are described in the Appendix.

Zoom In Zoom Out Reset image size

In the normal use of SCUBA, 450 μm observations would have been obtained in parallel with the 850 μm data; however, because the project was classed as a medium-weather program, the 450 μm data would have been unproductive and these bolometers were turned off during the present observations. In addition, the initial series of observations of the full program in 2003 demonstrated that the zenith atmospheric transmission limit ("weather band") initially granted resulted in insufficient signal/noise ratios, and we were able to obtain observations in 2004 under considerably better conditions. During the latter observing period half of the region observed in 2003 was observed a second time to offset the effect of the relatively poor conditions encountered during the early observations.

2.1.1. Observations at 850 μm

The region to be observed was divided into 16 segments each 30 × 15 arcmin in extent. The center positions of these are given in Table 1 together with an observing log for the observations and the mean zenith optical depth at 225 GHz (τ₂₂₅), normally available from the Caltech Submillimeter Observatory, located nearby on Mauna Kea. The optical depth at 850 μm (353 GHz) is approximately four times that at 225 GHz, and the relationship between these values is well determined (Archibald et al. 2002). To obtain the zenith opacity at 850 μm we used either the nightly polynomial fits to τ₂₂₅, available via the JCMT Web site, or extrapolations between values measured by skydips performed at the JCMT during the observations.

Table 1. Observing Log: SCUBA 850 μm Observations

Field	Observation Date	ℓa	ba	τ(225)b	rms noisec
01	2004 Apr 05	42.46241	−0.09627	0.085	0.07
02	2004 Apr 05	42.49036	−0.34470	0.085	0.07
03	2004 Apr 05	42.95928	−0.04037	0.085	0.04
	2004 Apr 04			0.090
04	2004 Apr 04	42.98723	−0.28880	0.080	0.07
05	2004 Apr 04	43.45614	+0.01552	0.075	0.06
06	2004 Apr 01	43.48409	−0.23291	0.065	0.05
07	2004 Mar 31	43.95301	+0.07142	0.050	0.06
08	2004 Mar 30	43.98096	−0.17701	0.055	0.05
09	2003 Aug 01	44.44987	+0.12732	0.147	0.07
	2003 Jul 29			0.150
10	2003 Aug 01	44.47782	−0.12111	0.135	0.08
	2003 Jul 29			0.170
11	2003 Aug 01	44.94674	+0.18322	0.149	0.07
	2003 Jul 29			0.150
12	2003 Aug 01	44.97469	−0.06521	0.146	0.07
	2003 Jul 29			0.170
13	2003 Aug 01	45.44360	+0.23911	0.146	0.10
14	2003 Aug 01	45.47155	−0.00932	0.146	0.10
15	2003 Jul 31	45.94047	+0.29501	0.146	0.15
16	2003 Aug 01	45.96842	−0.04658	0.146	0.12

Notes. ^aGalactic coordinates of the central position of each field. ^bZenith optical depth at 225 GHz (1300 μm); to obtain the equivalent value for the observing wavelength of 850 μm, multiply these numbers by a factor of about 4. ^crms noise value (Jy beam⁻¹ pixel⁻¹) in the final map, as determined from regions free of source emission. For regions where there was more than one observation this is the value for the combined data.

Download table as:  ASCIITypeset image

Each segment was observed at 850 μm in the standard scan-mapping mode used with SCUBA, in which the subreflector was nutated at 8 Hz while the telescope was scanned across the source (Jenness et al. 2000). Equatorial coordinates were used for the observing. A total of six chop angle/direction combinations provide a complete sampling of the image plane; in this particular case two orthogonal chop directions (0° and 90°) with each of three chop distances (30'', 44'', and 68'') were used. For these observations the scanning speed was increased beyond that normally used for full sampling by a factor of 2 to 48''s⁻¹; since only 850 μm data were being taken, this was sufficient for the present purpose. Each scan across the segment being mapped was sufficiently long to include the total extent of the segment in the scanning direction, and since the orthogonal scan directions were determined with respect to the bolometer array rather than the target region, the scan length was determined in real time. The observing process was quite efficient, each segment requiring slightly less than 1 hr for the observing. Together with calibration and other overheads, about 36 hr in total were used for this program.

For calibration purposes, one or two smaller maps (about 3' in extent) of standard sources were taken in the same manner during each observing night. Uranus and Neptune were used as the primary calibrators, although the galactic objects CRL 2688 and 16293 − 2422 also were used on occasion, depending on availability. The planetary data were used also to obtain the effective beamwidth at half-power; in the 2003 observing sessions the observations of Uranus gave a mean value of 152, while Neptune was used to derive a mean value of 146 during the 2004 observing sessions. The slight difference in the two values was likely a result of differing dish surface adjustments, ambient temperatures, and prevailing sky conditions at the two epochs.

2.1.2. Data Processing and Calibration

The initial stages of the reduction of the 850 μm data were carried out using the standard software available at the time (Holland et al. 1999) to flat-field the data, correct for extinction, and remove sky noise. The noise responses of the individual bolometers were first inspected to identify, and eliminate, pixels with high noise prior to the initial data processing. This initial processed data set was then used to identify other cases of high noise values before re-processing the data with poorer data omitted. The standard SCUBA reduction sequence (SURF; Jenness & Lightfoot 1998) was used via the online reduction facility ORAC-DR (Jenness & Economou 1999) to produce the final processed data set with extinction correction, baseline subtraction, and sky corrections included.

These data were then processed to construct images of the sky with a pixel size of 4 arcsec. We used both the standard facility rebinning/dual-beam removal process and a direct matrix inversion method (Johnstone et al. 2000) to produce versions of the final image with a pixel size of 4''. Since the largest chop throw employed in the observations was 68'', any structure in the reconstructed images on scales larger than this value is suspect; nevertheless, features with high aspect ratios should be correctly reproduced in any case. For the present rather inhomogeneous data set (as regards data quality), the standard data reduction tools introduced undulations in the images which are very unlikely to be real. The matrix inversion technique produced images of rather better quality in this regard and form the basis of the analysis in the present paper.

The JCMT SCUBA data were calibrated with a single value of the instrumental gain (203 Jy V⁻¹ beam⁻¹) derived from the ensemble of calibrator observations (see Section 2.1.1). The latter were observed in the same way as for the program fields, although covering a much smaller area. The observations of Neptune and Uranus were given the stronger weighting in this determination. Values from individual observing sessions varied by as much as 15% about the mean, and the overall calibration accuracy of the final image is set by the latter uncertainty.

2.1.3. Image Quality at 850 μm

The products of the matrix inversion are the reconstructed image of the sky, together with maps of the noise and the number of measurements at each point. For the present observations these data are shown in Figure 2 for an overview of the results. On the scale presented in this figure (note the beam size in the lower left corner of the left panel), there is little discernible detail, although the brightest submillimeter wavelength sources are clearly visible. The middle and left panels, respectively, show the rms noise level and number of measurements at each data point in the reconstructed image. Observational data for each field contain information somewhat (about 1') beyond the formal edges of the field. The resultant overlap regions between neighboring fields therefore have correspondingly more observations per point, and hence lower noise values, resulting in the pronounced orthogonal pattern evident in the "noise" and "number" images.

Zoom In Zoom Out Reset image size

There is substantial variation overall in the rms noise across the 850 μm image. The maximum noise values occur in the northern part of the field, where there are fewer observational samples per data point and the sky transmission was relatively poor during the time in which these data were taken (see Table 1). In the least noisy region (field 3) the rms noise level is below 0.1 Jy beam⁻¹ pixel⁻¹, as compared to almost 0.3 Jy beam⁻¹ pixel⁻¹ in the worst case. In the former region, two observations were made, both under good sky conditions, whereas only a single observation under relative poor conditions was obtained for the latter. The maximum number of observations per point in the reconstructed image, 51, occurs in this region, together with fields 9–12; however, in the worst cases (fields 13–16, in the northern part of the region), the sky transmission was relatively poor, and the number of observations limited, for all the observations.

2.2. Image Refinement and Observational Results at 850 μm

The sky image obtained at 850 μm (shown in the left panel of Figure 2) clearly shows residual low-level undulations on scales larger than the maximum chop throw. These are most likely due to bolometer gain drifts and short-term opacity variations. As an aid to subsequent analysis, we filtered out these effects by subtracting a smoothed version of the image, including only the low-level emission, from the original. We first clipped the image in Figure 2 at ±0.5 Jy beam⁻¹ and convolved the resulting image with a two-dimensional Gaussian of width 31 pixels (i.e., 124''). This smoothed image was subtracted from the original data. Further, to facilitate comparison with the 1200 μm data, the smoothed image was convolved to an effective beam of 22''. The calibration of the final image was checked against the JCMT Legacy Catalogue (Di Francesco et al. 2008) and the source peaks were found to be in agreement to within 20%.

Figure 3 displays the final image, revealing a significant improvement in image quality; in particular, the base level in these data is considerably flatter on a scale of a few arcminutes than that for the initial image derived from the matrix inversion procedure. Also, the variation in the noise level across the image is now rather obvious, and a substantial number of submillimeter wavelength sources are clearly visible even at this reduced angular scale.

Zoom In Zoom Out Reset image size

To summarize these basic data, the measured rms noise for each segment ("Field") of this final map is included in Table 1. These values were determined in each case over a square area 50 pixels on a side from a region judged largely free of localized emission from the individual dust clumps. The noise values obtained are significantly improved over those for the original reconstructed map shown in Figure 2, mostly due to the smoothing to larger effective beam size.

On the scale of the entire image shown in Figure 3, it is not possible to obtain a good assessment of the image quality in regard to the detection of emission from dust in this part of the Galactic plane. For this purpose, we select three relatively small regions, indicated by the squares overlaid on Figure 3, and show these sections of the image on a common intensity scale (−0.6 to 0.3 Jy beam⁻¹), chosen to emphasize low-level structure in the data, in Figures 4, 5 and 6. Each region is about 17' square.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 4, labeled "blank," shows a section of the image in which there are no obvious emission sources, and which therefore illustrates the nature of the background noise in the data at 850 μm. Faint linear structures are due to the residual noise characteristics of individual bolometers resulting from the orthogonal pattern in which the SCUBA array was scanned across the region.

The second field is shown in Figure 5; seven compact structures in this part of the image are subsequently identified by an automated clump-finding algorithm, as indicated by the crosses, and as described in Section 3. The number of dust clumps found in this map segment is typical of the average for the entire surveyed region. There are additional structures visible in this image which are likely to be weak 850 μm emission from dust, evident from comparison of the characteristics of this image with the "blank" field in Figure 4.

The third map section, in Figure 6, contains the star-forming region W49, the source of the most intense emission in the entire surveyed area. This image reveals in detail extensive ridges and clumps of dust emission surrounding the core of W49. The latter is situated within a depression in the base level caused by the chop-scanning observing method used. The flattening process described above cannot significantly remove this instrumental artifact. In this image, in particular, there is substantial evidence of clumped and filamentary dust structures below the formal clump sensitivity limit, especially to the southeast of W49. Again, comparison of this last image with Figure 4 indicates that these weak features represent real emission from dust rather than instrumental artifacts.

2.3. Observations at 1200 μm

The final 850 μm image was used to locate and obtain the basic parameters of the compact objects within the region. For this purpose we used a two-dimensional version of the clump-finding algorithm clfind (Williams et al. 1994). In this implementation, the shapes of individual clumps are not pre-defined and each map cell in which the signal is above a given threshold is assigned to a particular "clump" or coherent structure. Johnstone et al. (2001) present an extensive discussion of this particular approach, and their article should be consulted for detailed information on the methodology.

As noted in Section 2.1.3, the 850 μm image has significant differences in the noise level between the individual sections observed. However, the noise level within any given map section does not vary significantly, except close to the perimeter, where there are fewer data points per beam area. The noise is also enhanced in pixels which record strong continuum emission sources. In order to define a set of clumps according to a coherent minimum standard, we first used the clfind program to search for signal clumps with peak emission five times the lowest overall noise level in the map. From these we eliminated clumps with peaks less than five times the local (that is, within each section) noise level. Next, to eliminate spuriously identified clumps, we removed those with areas smaller than that of the beam (i.e., incorporating less than 25 pixels) and objects very close to the map edge, where the local noise level is especially elevated. Finally, we removed remaining apparent noise spikes through visual inspection.

A total of 132 clumps survived the detection criteria. The data for these are presented in Table 2, where each clump is identified by a running number and its galactic coordinates. The list is ordered in decreasing galactic longitude. The spatial distribution of these objects is displayed in Figure 7, where we indicate also the relative sizes of the clumps. This shows that there are three major groupings of submillimeter wavelength continuum clumps, around the G45.1 and G45.5 regions in the north, and associated with the W49 star-forming region in the south, where 26 clumps were identified within the field of view shown in Figure 6.

Zoom In Zoom Out Reset image size

The major uncertainty in the flux densities obtained for the dust clumps arises from the calibration of the data. We estimate that this effect is about 20%, applicable to all the clumps identified in Table 2. In addition, the noise characteristics vary across the image, as is clearly apparent in Figure 3. Using the rms noise values in the sections of the image with the least good noise values, we obtained estimates of the errors in the clump flux densities, from which we derived estimates of the fractional errors in the total flux ranging from 4% to 8%. The varying noise level for each segment of the image results in relatively fewer weak sources being identified in the noisier regions, in particular in the northern sections of the region.

The most intense single clump (entry 90 in Table 2) is associated with the core of W49, and accounts for 37% of the total flux density from the entire region observed. Even aside from these concentrations, the distribution of clumps is far from uniform. At the south end of the map clumps appear to be confined to a relatively narrow band, while above W49 and at the higher-latitude end of the region surveyed the distribution is more extended in galactic latitude. There are no sources identified in a region somewhat larger than 05 in extent centered at declination ∼10°30' (see also Figure 4); this appears to be a real lack of objects, as the noise levels are similar to those in the rest of the region observed.

In Table 2 we include in the final two columns the integrated and peak flux densities at 1200 μm. These data were derived from the 1200 μm image using the position and footprint for each of the clumps identified at 850 μm, rather than using an independent clump search at 1200 μm. The integrated flux densities follow a linear relation for which S₈₅₀ ≈ 2.3 × S₁₂₀₀, as shown in Figure 8. The peak flux relationship shows the same scaling. We estimate the likely uncertainty in the flux densities at 1200 μm to be similar to those at 850 μm, i.e., about 20%, with a minimum sensitivity of about 0.1 Jy beam⁻¹.

Zoom In Zoom Out Reset image size

Most of the clumps have integrated flux densities at 1200 μm of less than 1 Jy, and there is a substantial scatter about the mean value relation shown in the figure. The latter is at least in part due to measurement errors, but we would expect that dust temperature differences within the clump ensemble also contribute significantly to this scatter. The data are not sufficiently precise to enable us to draw firm conclusions in this regard; however, the low conversion factor of 2.3 between 850 μm and 1200 μm fluxes does require both a low dust temperature, T_d < 20 K, and a low value for the dust emissivity power-law exponent, β < 2.

The region we observed is shown projected in galactic coordinates in Figure 9. Here the SCUBA 850 μm and SIMBA/SEST 1200 μm images are placed in context with the molecular emission from the region by comparison with a velocity slice (50–70 km s⁻¹) from the GRS ¹³CO (1–0) data. The latter velocity range includes the majority (two-thirds) of the dust clumps observed in the region, as determined from the GRS data and additional molecular line observations obtained in the present work.

Zoom In Zoom Out Reset image size

As noted in Section 2.3, the 1200 μm data were obtained in a direct-detection mode. In principle, this technique should ensure that no artifacts remain in this 1200 μm image, such as those which can result from the removal of the dual-beam signature in the case of the JCMT 850 μm image following removal of the dual-beam signature.

In principle, larger-scale structures should be faithfully registered in the 1200 μm data, and the base level well determined. Inspection of the 1200 μm results thus provides confirmation that there are no substantially extended bright emission structures in the 850 μm data set which, if present, would have been diminished by the chopping technique used in the latter case. Further, the relative sensitivity of the two images to small-scale structures is quite comparable; this is illustrated by comparing the detailed 850 and 1200 μm images in the following section.

However, as is obvious from Figure 9, only the highest column density regions of ¹³CO (1–0) emission are associated with detectable dust emission, and, conversely, the majority of the gaseous material is not associated with observable amounts of cold dust at the sensitivity levels attained in the present work. Note that W49 is not registered in the ¹³CO (1–0) velocity range presented in Figure 9, as its LSR velocity is about 7 km s⁻¹.

4.1. Results for Individual Regions

Two sections of the 850 and 1200 μm images are presented in closer detail (and in equatorial coordinates) in Figures 10 and 11, where the continuum images from the present work are compared with the GRS ¹³CO (1–0) (using the same velocity range as for Figure 9) and 8 μm images of the same regions taken from the MSX mission (Price et al. 2001).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The G45.1/45.5 region (see Figure 10) includes several compact H ii regions which are known to be associated with OH and H₂O maser sources. At millimeter and submillimeter wavelengths the region consists of two main groups of compact objects, along with several weaker components, which appear in both images. These structures are well correlated with their appearance at mid-infrared wavelengths (e.g., at 8 μm, from the MSX mission) and with emission in the ¹³CO (1–0) line. Similarities in the appearance of these objects in dust, CO, and mid-IR emission are typical of images of active star-forming regions.

The latter situation contrasts with a region about a degree further south, shown in Figure 11. In this case, although there is considerable complexity in the ¹³CO (1–0) emission as seen by the GRS, most of the latter molecular structures are not visible in continuum emission at millimeter and submillimeter wavelengths, implying that they possess a relatively low dust column density. The exceptions are a relatively bright compact structure consisting of two main clumps to the west and an elongated feature to the east about 5' in length, both of which appear in the 850 μm and 1200 μm images, as well as in the ¹³CO (1–0) data. However, neither of the latter objects are visible at 8 μm or 24 μm (see Section 8.2), suggesting that they are relatively cold, dense objects which are yet to form stars, in contrast to the group of dust/gas clumps in the G45.1/G45.5 region (Figure 10).

Figure 12 focuses on the elongated feature mentioned above which is centered at G44.48 − 0.13. An expanded view of this structure shows it to consist of a series of (probably) six distinct dust clumps, terminated at the eastern extremity by an IRAS object (19110+1007) which also shows relatively weak indications of emission from dust at 850 μm. The positions determined for these dust clumps substantially predate the formal clump-finding analysis presented in Table 2 (Section 3); nevertheless, the six clumps indicated by "plus" signs in Figure 12 were independently identified in Table 2 as clumps 46–51 (from east to west). Possible dust emission associated with IRAS 19110+1007 lies below the formal detection level.

Zoom In Zoom Out Reset image size

The latter region was subsequently observed using the JCMT in service mode in the ¹³CO (3–2) transition, for which the angular resolution attained is the same as for the continuum image. These data show (Figure 13) more extensive emission from gas than that seen from dust in the 850 μm continuum, a result of the instrumental sensitivity to gas (CO) relative to that of dust emission in these observations, in addition to the more pronounced clumpiness of the latter. With the exceptions of the clumps, most of the ¹³CO emission is quite optically thin and contains relatively little, if any, cold dust.

Zoom In Zoom Out Reset image size

Using these data we can associate LSR velocities with five of the more compact clumps, for which the measured parameters are listed in Table 3. The clump velocities are very similar, within about 0.5 km s⁻¹ of the mean value of 60.0 km s⁻¹, and the velocity widths are on average within about 0.25 km s⁻¹ of a mean value of 3.6 km s⁻¹. These data imply that this group of clumps constitute a single elongated structure in which several potentially star-forming regions are collapsing, a unique feature within the extent of the 850/1200 μm images in this paper. The kinematic distance of this structure implied by the mean radial velocity is about 7.5 kpc (see Section 6).

Table 3. G44.484-0.133: Peak ¹³CO (3–2) Parameters

Clump	Name	R.A. (J2000)	Decl. (J2000)	Amplitude K T*a	Line Width km s−1	LSR Velocity km s−1	Integrated K km s−1
A	G44.515 − 0.168	19:13:24.1	10:12:58	3.0	3.5	60.3	12.4
B	G44.505 − 0.165	19:13:22.3	10:12:32	2.4	4.0	59.7	9.2
C	G44.488 − 0.151	19:13:17.2	10:12:00	2.9	3.8	59.4	13.1
D	G44.485 − 0.134	19:13:13.3	10:12:20	3.1	3.3	60.3	10.6
E	G44.473 − 0.103	19:13:05.1	10:12:32	2.7	3.3	60.3	11.6

Notes. These data represent the initial effort in the present program of observations to study the range of gas properties of dust clumps in star-forming regions in the Galactic plane presented in this paper. The results are nevertheless typical of the much larger field subsequently studied; rms errors for the peak amplitude and velocity are about 0.3 K T*_a and 0.2 km s⁻¹, respectively.

Download table as:  ASCIITypeset image

In 2005 August and September we undertook a more extended program of spectral line observations with the JCMT of the ¹³CO and ¹²CO (3–2) transitions of a substantial subset of the objects identified in the 850 μm continuum image. With the single-beam dual-channel receiver system available at the time, a complete map of the entire region would have required a prohibitive observing time; we therefore focused on individual targets already identified via their 850 μm continuum emission, as in the earlier observations described above.

We obtained either small maps two arcminutes in extent in the ¹³CO (3–2) transition, or simple integrations of five points arranged in a cross (center, and four positions offset in the cardinal directions by 30'') of the ¹²CO (3–2) line. The central positions observed are given in Tables 4 and 5 for the small maps and the five-point observations, respectively. We indicate the positions for these observations in Figure 14. The reference position used for these data was 5 arcmin either east or west of the central position, chosen in an endeavor to avoid probable confusion with nearby spectral line emission, as determined from visual inspection of the 850 μm continuum image of the region. For the small maps, we chose fairly bright 850 μm continuum targets, and obtained fully sampled images. A set of objects with generally weaker continuum emission were chosen for the five-point maps; in this case, our intent was simply to attempt to obtain spectra of typical weaker continuum objects, to determine kinematic distances and search for possible outflows.

Zoom In Zoom Out Reset image size

The locations of these five-point and map observations are indicated by small crosses and squares respectively in Figure 14. The central positions used were determined prior to the automated clump-finding process (Section 3) using visual inspection of the 850 μm image and the centroid-finding tool within the GAIA¹³ package. Nevertheless, the positions observed are in all cases within a few arcseconds of those ultimately adopted; in the tables we give the equatorial coordinates of the observed position and the reference number of the associated submillimeter clump.

5.1. ¹³CO (3–2) Data

The results of the ¹³CO (3–2) imaging observations are collected together in Figure 15. The velocity range covered by the spectrometer was from −10 to +110 km s⁻¹ with respect to the LSR. Thirteen positions were observed in total, several of which were selected to include more than one continuum feature based on initial inspection of the 850 μm image. The images show that in all cases there is close correspondence between the positions of the continuum emission features determined subsequently by the clump-finding algorithm and strongly concentrated molecular emission. In three cases, the datacubes reveal additional weak CO emission at a velocity well separated from the principal emission; images of the integrated emission from these velocity components, included in Figure 15, do not show any correspondence with the continuum emission clumps in these cases.

Zoom In Zoom Out Reset image size

For some of these objects we provide additional comments below. In general, it is worth noting that the CO emission, integrated over a velocity range of several km s⁻¹ (see Table 4), obscures the complexity of these structures seen with finer spectral resolution in the individual spectral channel maps.

CO42.436 − 0.258. The weak continuum emission clump G42.434 − 0.272 (see Table 2) at the eastern edge of the field is associated with CO emission at velocities of about 63 km s⁻¹.

CO43.101+0.055. The two continuum sources are related to low-velocity material associated with W49; weaker emission from gas at the more red-shifted velocities (58–60 km s⁻¹) has no continuum counterpart. CO43.238+0.046. The bright central source has significant outflowing material with relative velocities of ±6 km s⁻¹. This object may be part of the W49 region.

CO43.530+0.019. The two components associated with continuum emission are separated in velocity by about 3 km s⁻¹.

CO43.796 − 0.126. The bright central feature has significant outflow velocities, extending by about ±8 km s⁻¹. Emission at near velocities (15–18 km s⁻¹) is not associated with detectable continuum emission in these data. This image has significant overlap in declination with that of CO43.818 − 0.117, which forms an extension to the latter.

CO44.310+0.042. The bright northern component has outflow velocities of ±8 km s⁻¹; neither continuum source is associated with the higher-velocity emission near 70 km s⁻¹.

5.2. ¹²CO (3–2) Data

A total of 43 positions were observed within the continuum map survey area in the ¹²CO (3–2) transition. The integration time was 30 s per point, using a position-switch in azimuth. Since ¹²CO (3–2) emission is extensive within the galactic plane, we used a widely spaced five-position cross to attempt to isolate the particular velocity component of the gas associated with the dust accompanying each continuum emission clump. Subsequent examination of the data indicates that this approach was quite successful, in that in almost all cases there was very little doubt about the velocity of the molecular gas associated with any given clump. A single spectrum on the clump position would not have provided the necessary degree of discrimination in this regard, as there is often more than one velocity component visible within the spectra.

The results of the analysis of these data are given in Table 5. For each position we provide the intensity of the ¹²CO (3–2) transition integrated over the velocity range, and the velocity centroid of the emission line. In cases where the line shape is relatively simple we also give the line width at half power derived from fitting a Gaussian to the line emission profile. In most cases, however, the line profiles are complex, often consisting of two components, most probably due to self-absorption in the core of the line structure. We use one such case (CO45.569 − 0.118) in Figure 16 as a typical example of the data obtained from these observations; in this particular instance the object was close to the edge of the continuum image, where the noise is relatively high, and although it was identified in the initial visual search for continuum clumps it fell outside the constraints of the subsequent automated clump-finding process.

Zoom In Zoom Out Reset image size

CO45.569 − 0.118 shows evidence of molecular outflow, with visible line wings at the central position extending to perhaps 20 km s⁻¹ from the central velocity. Relatively few of the continuum clumps, however, have significant extended line wings resulting from outflowing gas, as is commonly found in star-forming cores. This is most likely because we selected against the more active star-forming regions by observing relatively weak continuum targets, for the most part. We identify the continuum clumps subsequently associated with the observed objects in Table 5. Eight of the original ¹²CO (3–2) targets were not ultimately selected by the automated continuum emission clump-finding technique, but nevertheless all show significant ¹²CO (3–2) emission apparently centered at the nominal continuum emission peak.

A determination of the masses of the dust clumps we have observed requires that the distance to each be known. For a given galactic longitude within the plane of the Galaxy, excepting the tangent point, there are two distances corresponding to a given velocity for each of the clumps identified in the 850 μm continuum observations. In principle, this dichotomy can be resolved by inspection of H i line data associated with each line of sight; the velocity of H i self-absorption (HISA) in the direction of each clump then indicates whether a particular molecular cloud (identified via ¹³CO (1–0) data from the GRS data) resides at the near or far distance. This technique was illustrated by Jackson et al. (2002) in the case of one region (GRSMS 45.6+0.3) included in the present work, and more recently further developed by others, e.g., Goldsmith & Li (2005). We show an example of the relation between the ¹³CO (1–0) GRS data and H i self-absorption in Figure 17 for one of the objects in the present work.

Zoom In Zoom Out Reset image size

The initial formal HISA processing for the 132 dust clumps was carried out by Alexis Johnson (during the tenure of a temporary student position at Boston University in 2007) using existing code developed for this purpose. Table 6 contains the results of this analysis; for each object (identified by its clump number in Table 6) we list the peak ¹³CO (1–0) intensity within the bounds of the clump at the emission match, the velocity of this emission, the determination of whether the emission is at the near (N) or far (F) distance, and the distance associated with this determination using the standard galactic rotation model. The "cloud number" given is the identifier obtained from the H i catalog maintained locally at Boston University. Individual clouds were assigned to either the near or far ("N" or "F") distance, in the final two columns of Table 6. A few clumps could not be positively identified with any particular cloud—these are therefore not assigned to a cloud number, do not have a near/far determination, and cannot be assigned a formal distance as a result. These objects are identified by zeros in the final three columns of Table 6. Some have associated low-level emission not associated with any cataloged cloud; others are best matched with negative-velocity components in the CfA longitude–velocity diagram; ¹³CO (1–0) data are not available for these velocities. In addition to the distance found for the larger associated H i cloud, distances for individual clumps within such clouds can also be estimated using the H i velocity observed at the location of the clump. We identify these two distances as the "cloud" and "clump" distances; based on the standard galactic rotation model, both are given in Table 6. While these distances are subject to the motions of the individual clumps in the Galaxy, we anticipate that they should be accurate to within 10%.

Subsequent comparison of the subset of the HISA results with those for which there are also JCMT ¹²CO and ¹³CO (3–2) data (55 clumps in total) revealed that the velocity determinations obtained from the latter substantially differed in 11 of these objects. In four of these cases the HISA analysis found no suitable velocity match; these objects appear in Table 6 with an assigned velocity of −15 km s⁻¹. In light of this issue the H i and ¹³CO (1–0) spectra of this group of sources were re-examined by Julia Duval (Boston University), using the ¹²CO (3–2) velocity obtained from the JCMT data. In all cases ¹³CO (1–0) emission could be identified at the same velocity as observed in the ¹²CO (3–2) line. We then used H i self-absorption data obtained from the VLA Galactic Plane Survey (Stil et al. 2008) to search for H i self-absorption in order to resolve the near/far kinematic distance dichotomy obtained from the Clemens galactic rotation curve.

From the "cloud" and "clump" distance data in Table 6 we derive two estimates of the clump masses M_clump, using the relation (see, e.g., Johnstone et al. 2001)

$$\begin{eqnarray} M_{\rm clump} &=& 12.2 \times S_{850} \left[ \exp \left({17\,{\rm K} \over T_d}\right) -1\right]\, \left({\kappa _{850} \over 0.02\, {\rm cm}^{2}\,{\rm g}^{-1}}\right)^{-1}\nonumber\\ &&\times \left({d \over 1700\,{\rm pc}}\right)^{2} \,M_\odot \end{eqnarray} \tag{ 1 }$$

where we assume that the emission arises in optically thin dust particles having a dust temperature T_d and a dust opacity κ₈₅₀ = 0.02 cm² gm⁻¹. We further assume that the dust has a typical temperature T_d ∼ 15 K, based on results obtained from other investigations of the physical properties of infrared-dark clouds (Egan et al. 1998; Carey et al. 1998; Johnstone et al. 2003). Based on our estimates for the uncertainties in the flux density and distance scales of 20% and 10% respectively, formula 1 above suggests that the masses we derive are uncertain by a factor of approximately 2.

These cloud/clump mass data are summarized in Table 7. In a few cases (7 of 132 in total) we were unable to determine a value for the "cloud" distance, and only the "clump" distance allows us to derive a value for the mass. Since the masses determined using the "cloud" and "clump" distances are very similar in objects where both values have been determined, we conclude that for the few clumps where only a "clump" distance is available, the associated mass estimate is reasonable.

The fractional uncertainty in clump mass can in turn be determined from the cloud distance and the uncertainty in the flux density. The latter provides the primary source of error, which we estimate is about 20%. Secondary sources of error arise from the dust opacity assumed, which is likely to vary by at least a factor of 2, and may well vary significantly across the range of clump properties included in this work. In combination with the uncertainty in the distances of individual clumps (at least 10%), it is clear that clump properties (in particular, mass) cannot be very precisely determined.

From the clump masses, one can compute the mass function of dust clumps within the observed region of the Galactic Plane. The cumulative mass function, N(M), is usually parameterized as a power law with N(M) ∝ M^−α. When α < 1, the bulk of the mass resides in the largest clumps. Alternatively, when α>1 the bulk of the mass resides in the smallest clumps. Thus, the Salpeter (1955) mass function for stars, with α = 1.35, puts most of the stellar mass in low-mass stars. For cores in nearby molecular clouds, observed in the submillimeter wavelength regime a similarly large value for α has been measured (e.g., Motte et al. 1998; Johnstone et al. 2000; Alves et al. 2007). At larger distances, however, the mass function of submillimeter clumps is much more ambiguous (e.g., Reid & Wilson 2006; Rathborne et al. 2006; Ragan et al. 2009).

Figure 18 depicts the distribution of dust mass versus radius for all the clumps identified in the present data at 850 μm. The dash-dotted line denotes the mean surface brightness sensitivity of the survey; below this line clumps are too faint to be detected. A majority of the identified clumps with mass below 500 M_☉ are near the observational sensitivity limit, which suggests that substantial incompleteness may begin well before the threshold for detection is reached. Thus, cumulative mass plots, such as those in Figure 18, should be interpreted with caution. The solid line shows where gravity is balanced by thermal pressure. This indicates that all of the clumps identified would require additional support (e.g., via nonthermal motions or magnetic fields) in order to avoid gravitational collapse. This situation holds even for rather higher temperatures than we have assumed in this work (15 K). The dashed line shows where gravity is balanced by the typical nonthermal motions observed around each clump (using the mean CO line width of 4 km s⁻¹ FWHM).

Zoom In Zoom Out Reset image size

It is to be expected that there should be a relationship between the 850 μm flux density and the ¹²CO (3–2) emission, and for those objects where we have mapping data in the ¹³CO (3–2) line, this is confirmed. The continuum emission at these wavelengths is optically thin, and thus the flux density scales directly to the dust mass within the beam. For low optical depths, the ¹²CO and ¹³CO (3–2) should also exhibit the same behavior, assuming a constant gas-dust ratio. In more extreme circumstances, at very low densities the dust may well be evaporated by stellar radiation, and the relationship would break down. At higher densities the emission from ¹²CO (3–2) rapidly becomes optically thick with increasing column density, while the ¹³CO (3–2) line emission remains optically thin to much greater associated continuum column densities. Furthermore, it is highly probable that a significant fraction of the CO contained within these clumps is simply invisible, being frozen out at the low temperatures (we use 15 K in the absence of more complete data) expected in cores, in particular those which do not contain stars. We cannot reliably predict the degree to which this might be occurring in the sample of objects we have examined in this work.

We use the present limited ¹²CO and ¹³CO (3–2) observations to illustrate this point, by comparing the peak 850 μm flux density for those clumps for which we obtained molecular line data. We show the results in Figure 19. The ¹³CO (3–2) data, shown in the frame to the right, exhibit a linear relationship within the probable errors over the range of continuum flux densities up to 10 Jy beam⁻¹. The ¹²CO data, at left, also indicate a linear relationship with the continuum flux density, but the relationship breaks down at quite low flux densities, probably at less than 2 Jy beam⁻¹.

Zoom In Zoom Out Reset image size

In this paper, we employ the ¹³CO (1–0) GRS data as the primary means to obtain kinematic distances of the continuum sources in this part of the Galaxy, if the line and continuum emission can be associated with the same object. However, the ¹³CO (1–0) transition tends to emphasize relatively cool, more diffuse, material, whereas the millimeter and submillimeter continuum emission arises in compact regions with relatively high column density. Furthermore, the observing beam sizes for the ¹³CO (1–0) and the 850 μm data sets are substantially different. For these reasons we cannot always be entirely sure of the physical relationship between a given continuum feature and an object seen in the spectral line data.

The 850 μm image shown earlier in Figure 11 also illustrates this point. While there is generally a good correspondence between the continuum emission clumps in the images and structures in the ¹³CO (1–0) survey data, the converse is not true; there are a number of ¹³CO features which do not appear in the continuum data. This may be simply a result of the limited sensitivity of the continuum images, or it may be that the dust/gas ratio diminishes when the optical depth in the cloud is relatively low, perhaps due to evaporation of grains due to incident starlight.

8.1. The Physical Content of Dust Clumps

8.2. Spitzer Images

Data from the Spitzer satellite in particular are of considerable interest in the context of the current work. In a subsequent paper we intend to discuss the present results in light of the 24 μm images from this instrument. For the present, we show two small sections each about 12' square of the field which contain interesting extended structures at the millimeter and submillimeter wavelengths.

The first of these, in Figure 20, shows an object which has an apparent head-tail structure more clearly seen in the 1200 μm image. The single millimeter and submillimeter wavelength clump in the field is seen at 24 μm, and there is evidence at the mid-IR wavelength of associated structure in the region of the tail. Based on the CO velocity, we estimate this object (G45.651 + 0.272) to lie at a distance of about 10 kpc (see Table 6).

Zoom In Zoom Out Reset image size

The second object for which we have Spitzer data is the "ridge" of emission centered at about G44.48 − 0.13 (see also the discussion in Section 4.1 and Figures 12 and 13). This consists of about six clumps of millimeter and submillimeter emission, in which the structure is clearly seen at both 850 and 1200 μm. The velocity w.r.t. the LSR for all these clumps is about 60 km s⁻¹, and we conclude that these objects are part of a coherent structure. In this case, the 24 μm image (see Figure 21) shows no obvious counterparts to the millimeter and submillimeter wavelength images; however, a bright stellar object lies off the eastern extremity of the submillimeter wavelength ridge; this object corresponds to IRAS 19110+1007.

Zoom In Zoom Out Reset image size

We have presented the results of parallel efforts to obtain relatively large-scale millimeter and submillimeter wavelength images of continuum emission from cold dust in a section of the Galactic plane 2 deg² in extent, toward ℓ ∼ 44° ± 2°, using complementary chop-scanning and Fastscanning techniques.

The observed emission from the present region, typical of star-forming clouds in the Galaxy, is strongly clumped. 132 dust clumps were identified, and using distances for these objects derived from CO emission and H i self-absorption data we find that they range in mass from a few M_☉ to almost 3 × 10⁵ M_☉. We find that the cumulative mass spectral index of these objects is about unity over those distance bins which contain sufficient clumps for a determination. We note also that, for the assumed dust temperature of 15 K, the clumps are too massive to be supported by thermal pressure alone, and hence one can expect that such clouds will naturally form star clusters via fragmentation and subsequent collapse of the individual fragments.

The present observations are a fairly small effort to gain some insight into the behavior of large dust concentrations which are expected to collapse under self-gravity into star-forming clumps. Much more extensive surveys of the submillimeter wavelength continuum emission sky are planned by various groups using the JCMT once the next-generation imaging receiver SCUBA2 becomes fully operational in the near future.

We are grateful to the staff of the JCMT and to various visiting observers at the telescope for assistance in carrying out the 850 μm observations under the flexible service observing scheme. In particular, we thank the telescope operators for their fine attention to detail during the observing periods. We also greatly appreciate the efforts made by local staff during the latter half of this program to maintain SCUBA in working condition at a time when the instrument was having operational difficulties.

The JCMT observations were obtained during the execution of a joint Canada–UK project under programs M03BU18, M04AC15, S04BC11, and M05BC11. The JCMT is operated by the Joint Astronomy Centre, on behalf of the Science and Technology Facilities Council of the United Kingdom, the Netherlands Organization for Scientific Research, and the National Research Council of Canada. The SEST observations made use of SIMBA, a 37-element bolometer array, and the observing program was carried out on site in Chile by several of the co-authors of the present paper.

We are indebted to Alexis Johnson and Julia Duval (formerly term assistants at Boston University) for providing us with distance estimates based on H i self-absorption and ¹³CO (1–0) data, and for the resolution of problematic cases of disagreement between the CO (1–0) and (3–2) velocities. In addition, we thank Ronak Shah, also formerly at Boston University, for his assistance with processing images from the Boston University/FCRAO Galactic Ring Survey (GRS) data.

We also thank the anonymous referee for providing detailed and helpful comments, which resulted in a substantially improved paper over that originally submitted.

This publication makes use of molecular line data from the Boston University-FCRAO Galactic Ring Survey (GRS). The GRS is a joint project of Boston University and Five College Radio Astronomy Observatory, funded by the National Science Foundation under grants AST-9800334, AST-0098562, and AST-0100793.

Helen Kirk was supported by NSERC CGSD and NRC GSSSP awards (Canada). Doug Johnstone is supported by a Natural Sciences and Engineering Research Council of Canada grant.

Finally, we acknowledge the following organizations and facilities. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. GAIA was developed by the Starlink Project and funded by the Particle Physics and Astronomy Research Council (UK). It is a derivative of the SkyCat image display and catalogue browsing tool, developed as part of the VLT project at ESO.

Prior to obtaining the data presented in this work, we made 850 μm observations with SCUBA at JCMT in 2000 to verify system performance and test our observational procedures. For this, we imaged a region centered on the strong continuum source G34.3+0.2, using the same parameters as we planned to use for individual sections of the Galactic Plane survey (see Section 2.1.1). The results of these observations are presented in Figure 1. The data were reduced using the same procedures as for the 16 individual fields which constitute the JCMT observations in this paper.

Although G34.3+0.2 is frequently used as a calibrator at millimeter and submillimeter wavelengths, there are rather few observations which provide images of the continuum emission from the region on the larger scale. One of the earliest observations of the region obtained with SCUBA imaged the central 2 arcmin (Strong-Jones et al. 1991). At the center of G34.3+0.2 is a cometary nebula which has been imaged at radio wavelengths on arcsecond scales (e.g., van Buren et al. 1990). Our observations show the structure of the region on a much larger angular scale. There are several clumps of cold dust seen in these data, and filamentary structure extends to the southeast from the central region. In the southern part of the image there is a dust clump at the edge of the image, and a clumped filamentary structure extends to the north over a similar projected distance (about 12').