THE W40 REGION IN THE GOULD BELT: AN EMBEDDED CLUSTER AND H ii REGION AT THE JUNCTION OF FILAMENTS

NIR photometric imaging observations were carried out using the Wide Field Camera (WFCAM; Casali et al. 2007) mounted on the 3.8 m United Kingdom Infrared Telescope (UKIRT). WFCAM contains four Rockwell Hawaii-II (HgCdTe 2048 × 2048 pixel) arrays spaced by 94% in the focal plane. With a pixel scale of 04, the field of view (FoV) of each array is ∼137. JHK-band data were obtained in the service mode during the period 2009 December 15–17. Observations at four positions are required to observe a contiguous (∼0.75 deg²) square field—called a WFCAM "tile." Observation at each of these positions was taken using a nine point jitter pattern and a 2 × 2 microstepping pattern, with an exposure time of 5 s each. The seeing was in the range of 13–18. We used the starlink software to do the initial processing, followed by mosaic making using the routine makemos in a manner similar to Davis et al. (2007; see this paper for details of the WFCAM processing).

SExtractor (Bertin & Arnouts 1996) was used to detect all the stellar sources in the region. Source detection and deblending is highly effective with this software. A detection threshold of 3 and a deblend contrast factor of 32 were used for our purposes. Subsequent to first pass of detection, the detection catalog was filtered using the FWHM, stellarness parameter, and ellipticity to remove spurious detections. Finally, some leftover spurious detections were removed manually to create a reliable detection catalog. Aperture photometry for the detected sources was performed using the phot routine in the IRAF apphot package. The aperture radius was set to 3 pixels (= 12) and the sky annulus was located 9 pixels away—with a width of 7 pixels. Sources without any centering error (cier), sky fitting error (sier), and photometric error (pier) were retained for further work.

Instrumental magnitudes are in the WFCAM native system (the Mauna Kea Consortium Filter System or the MKO system). Absolute photometric calibration of the J, H, and K photometry was done using the Two Micron All Sky Survey (2MASS) stars in the FoV, as explained below. We obtained 2MASS data with the constraints of "phqual = AAA" and "ccflag = 000" for all the stars in the FoV. This ensures detections that have the highest photometric quality without neighboring source confusion in all the three bands. We first constructed a J − H/H − K color–color (CC) diagram using these data to identify only the main sequence (MS) stars in the region. The MS stars obtained using the CC diagram were then used to carry out the photometric calibration of the WFCAM data. The 2MASS data were converted to the WFCAM system using the transformation equations given by Hodgkin et al. (2009). WFCAM data saturate for magnitudes brighter than 11.5 mag in all the three bands and these saturated points were replaced with 2MASS magnitudes.

Completeness limit tests were carried out for all the three bands by choosing subimages of size 5' × 5' located in the cloud region and the field region. The 90% completeness limits for the J-, H-, and K-band data are 19, 18, and 17 mag—respectively—for the field region and 18, 17, and 16 mag—respectively—for the cloud region, giving the mean as 18.5, 17.5, and 16.5 mag, respectively. These early observations of WFCAM have small differences in background matching between multiple chips and tiles, primarily resulting from the presence of a large-scale nebula contaminating the jittered frames. Therefore, the J, H, and K catalogs were clipped at 18, 17, and 16 mag limits for NIR CC, NIR color–magnitude (CM), and extinction map analyses to obtain conservative and robust classifications of both young stellar objects (YSOs) and extinction measurements. Since all NIR magnitudes had low errors associated with them, they were not filtered by any error criteria. The magnitudes of bright sources that were saturated in the WFCAM image but had good quality ("phqual=AAA") detections in 2MASS were taken from the 2MASS catalog after converting them to the WFCAM system using Hodgkin et al. (2009).

The Spitzer archival data for the Gould belt survey of Allen et al. (2006, PID: 30574) in all four Infrared Array Camera (IRAC) bands—3.6 (Ch1), 4.5 (Ch2), 5.8 (Ch3), and 8.0 (Ch4) μm were obtained using the Spitzer Heritage Archive. Final image mosaics were created with the MOPEX data reduction software, using the corrected basic calibrated data (cbcd) (S18.5 processed version), imask (bimsk), and uncertainty (cbunc) files. Mosaic images in all the four IRAC bands were created with a final pixel scale of 06.

Aperture photometry was carried out for the four IRAC bands. We first used SExtractor for source detection, as described above (see Section 2.1), and filtered to build a complete detection catalog. The detection catalog was used as input to the phot task in IRAF to obtain the aperture photometry of IRAC images. The aperture size, radius of the inner annulus, and radius of the outer annulus were taken to be 24, 24, and 72, respectively. The zero point magnitudes (with aperture corrections applied) were taken to be 18.593, 18.090, 17.484, and 16.700 for the four IRAC bands, respectively; these were calculated using the IRAC Instrument Handbook (version 2.0.2, 2012 March).

Average completeness limits were determined using the peak of the luminosity function for the four bands. As in the "Final Delivery Document For IRAC and MIPS data"^6,7 (see its Section 3.4.2), for the c2d legacy project (Evans et al. 2003), the peak of the observed luminosity function can be assumed to estimate the 90% completeness limit. We constructed the luminosity functions with a 0.5 mag bin size for each band and the peak (and thus the approximate 90% completeness limit) was found to be at ∼14, 14, 11.5, and 11.5 mag for Ch1, Ch2, Ch3, and Ch4, respectively.

The southeast region of the wide-field image where the extinction is low was the only region where both WFCAM and Spitzer had a high density of sources, which caused source crowding problems. However, as discussed in previous sections, extra care was exercised by filtering first at the level of SExtractor and then at the photometry level, allowing us to be quite certain that the detected sources were all real and uncontaminated. Finally, the cross-matching makes the detected sources even more robust.

The number of source detections in individual photometric catalogs of each band are different due to respective band extinctions, point spread function (PSF), sensitivity limits, and area of coverage. Different cross-matched catalogs were produced for different analyses. The final analysis was done on an area of size ∼3433 × 4000, centered at α₂₀₀₀ ∼ 18^h31^m4123, δ₂₀₀₀ ∼ −02°06'2921. We obtained the following cross-matched and merged catalogs for various analyses.

• 1.  
  Ch1–Ch2–Ch3–Ch4 catalog. The IRAC Ch1, Ch2, Ch3, and Ch4 band catalogs were cross-matched (within 06) with the constraint that all individual band errors are ⩽0.15 mag to produce an all-IRAC catalog of 1874 sources. This was used in the classification of YSOs.
• 2.  
  H − K − Ch1 − Ch2 catalog. Not all sources are detected in the IRAC Ch3 and Ch4 bands, but have good quality detections at H and K wavelengths. For the classification of such sources into different evolutionary stages, we obtained a cross-matched catalog of H, K, Ch1, and Ch2 magnitudes (NIR H and K catalogs were matched with a 04 matching radius; the Ch1 and Ch2 catalogs were matched with a 06 matching radius, followed by matching the resultant NIR and Spitzer catalogs with a 06 radius) with good quality detections in the IRAC Ch1 and Ch2 bands, i.e., sources with IRAC Ch1 and Ch2 magnitude errors ⩽0.15 mag. The resulting catalog contains 8585 cross-matched sources.
• 3.  
  J − H − K catalog. Next, we obtained a cross-matched (within 04) J − H − K catalog that was further clipped by conservative J, H, and K completeness limits of 18, 17, and 16 mag, as elucidated in Section 2.1. This filtered NIR catalog containing 10,990 sources was used for the identification of YSOs and for producing the extinction map of the region (see Section 3.2).
• 4.  
  H − K catalog. Many embedded NIR sources are detected only at H and K wavelengths. For identification of additional YSOs from amongst these, cross-matched NIR sources (within 04) with detections in only the H and K bands (clipped at H and K conservative completeness limits of 17 and 16, respectively) were obtained.

The catalogs described above are not mutually exclusive. Respective YSOs were identified using those in Section 3.1.

Narrowband imaging in the 2.12 μm H₂ (v = 1–0) S(1) filter and broad band imaging in the K' for continuum subtraction was obtained for a 19' × 14' region centered on the W40 embedded cluster. These data were obtained with the Calar-Alto 3.5 m telescope during the night of 2001 June 13 using the Omega Prime camera. The camera used an HgCdTe detector with 1024 × 1024 pixels with a plate scale of 039 pixel⁻¹ resulting in a 675 × 675 FoV. Observations were carried out using a 9 point jitter pattern. The exposure time was 30 s and 3 s—per jitter pointing—in the NB2122 and K' filters respectively, resulting in total integration times of 180 s and 18 s per pixel, respectively. Standard image reduction of dark subtraction, flat-fielding and sky-subtraction was carried out for each pointing. The K' images were PSF matched and scaled to the H₂ narrowband image to perform continuum subtraction.

The Herschel Space Observatory, a 3.5 m telescope, was launched to carry out observations in the wavelength range 51–670 μm using the Photodetector Array Camera and Spectrometer (PACS), SPIRE, and HIFI instruments (Pilbratt et al. 2010). We obtained the publicly available level 2_5 processed archival data for the PACS 100 μm (32 pixel⁻¹) and 160 μm (64 pixel⁻¹) bands—obtained as part of Proposal ID:KGBT_pandre_1. The "MADmap" (Cantalupo et al. 2010; Waskett et al. 2007) processed products are used, as opposed to "PhotProject" processing, since our aim is to analyze extended features in the region. A detailed overview of procedures involved, the data reduction pipeline, and the data products is available on the Herschel Web site.⁸,⁹ We use the images to examine the filamentary structures vis-a-vis the distribution of YSOs and radio continuum emission.

New radio continuum observations are obtained using the Giant Metre-wave Radio Telescope (GMRT) in two bands centered at 1280 MHz and 610 MHz. The data were obtained on 2011 November 15 and 2011 November 18 (Project ID 21_015), respectively, the details of which are given in Table 1. The GMRT array consists of 30 antennae located in an approximately Y-shaped configuration. Each antenna has a diameter of 45 m. A central region of about 1 km × 1 km consists of 12 (randomly distributed) antennae, while the rest are distributed along three radial arms (6 along each arm), up to a distance of ∼14 km. This configuration results in a baseline of ∼100 m–25 km, providing sensitivity to features ≲8' and 17' in the 1280 MHz and 610 MHz bands, respectively. Details about the GMRT array can be found in Swarup et al. (1991).

Table 1. Details of GMRT Observations

	1280 MHz	610 MHz
Date of obs.	2011 Nov 15	2011 Nov 18
Phase center	α2000 = 18h31m1575	α2000 = 18h31m1575
	δ2000 = −02°06'4930	δ2000 = −02°06'4930
Flux calibrator	3C286, 3C48	3C286, 3C48
Phase calibrator	1822-096	1822-096
Cont. bandwidth	32 MHz	32 MHz
Primary beam	24'	43'
Resolution of maps
used for fitting	4584 × 4440	4411 × 4067
Peak flux density	0.45 Jy beam−1	0.35 Jy beam−1
rms noise	6.34 mJy beam−1	5.52 mJy beam−1
Total integrated flux density
within a 4σ area	7.55 Jy	4.07 Jy
Integrated flux density
within a 50'' radius of the IRS 5 radio peak	109 mJy	121 mJy

Download table as:  ASCIITypeset image

Data reduction was carried out with the AIPS software. Bad data were flagged using the AIPS tasks "TVFLG" and "UVFLG." Multiple rounds of flagging and calibration were done to improve the calibration successively. Calibrated object data (W40) were subsequently split from the raw file (which also contains phase and flux calibrator data) and low-resolution (∼45'' × 45'') images were finally made using facet imaging with the help of the task "IMAGR." A few rounds of (phase) self-calibration were carried out to remove the ionospheric phase distortion effects.

The YSOs were identified using sources detected in the IRAC bands first, followed by using the catalog of sources with detections in the NIR and IRAC bands, and lastly using the NIR catalog of sources. The following is a list of the steps we followed.

• 1.  
  Identification of YSOs from the IRAC data was accomplished by computing the IRAC spectral index for the sources with detections in all four IRAC bands (i.e., sources from the Ch1–Ch2–Ch3–Ch4 catalog; see Section 2.3). The IRAC spectral index (α_IRAC = d log (λF_λ)/d log (λ); Lada 1987) for each source was calculated using linear regression. Thereafter, using the spectral index limits given in Chavarría et al. (2008) (Class I: 0 <α_IRAC, and Class II: −2 ⩽α_IRAC ⩽ 0), the sources were classified into Class I and Class II categories (see Figure 2(a)).
• 2.  
  The H − K − Ch1 − Ch2 cross-matched catalog (see Section 2.3) was used for the identification of additional Class I and Class II sources using the classification scheme of Gutermuth et al. (2009, 2010, see their Appendix A—Phase 2). For this, the dereddened (K − [3.6]) and ([3.6] − [4.5]) colors, calculated using the color excess ratios from Flaherty et al. (2007), were used. As stated in Gutermuth et al. (2009), to filter out possible dim extragalactic contaminants, we also applied limits on the dereddened Ch1 band magnitude to be ⩽15 and ⩽14.5 for Class I and Class II categories, respectively. We used the average extinction value estimated in Section 3.2 (converted to A_K using Rieke & Lebofsky 1985) to calculate the extinction for the Ch1 band in this region, and thus the intrinsic Ch1 magnitudes, to impose this condition. The reddening law from Flaherty et al. (2007) for Serpens (since it is closest to our region) was used for this purpose. Figure 2(b) shows the location of these Class I and Class II sources—which have been marked as filled black circles—on the (K − [3.6])_o versus ([3.6] − [4.5])_o CC diagram.
• 3.  
  In the next step, the J − H − K photometric catalog was used as follows. For the sources in this catalog, a J − H/H − K CC diagram was constructed. Figure 2(c) shows the NIR J − H/H − K CC diagram. The Classical T-Tauri Star (CTTS) locus (blue line) was taken from Yasui et al. (2008; which is derived from Meyer et al. 1997), while the locus for dwarfs (red curve) in the Mauna Kea Observatory (MKO) system was taken from Hewett et al. (2006). Similar to Yasui et al. (2008), the reddening laws of Rieke & Lebofsky (1985) (A_J/A_V = 0.282, A_H/A_V = 0.175, and A_K/A_V = 0.112) were used. The dashed green lines show the reddening vectors. The sources that were located in the "T" and "P" regions—similar to the method of Ojha et al. (2004a, 2004b) and Lada & Adams (1992)—were taken to be NIR Class II/III sources. There may be an overlap between the Herbig Ae/Be stars and Class II sources in this "T" region (Hillenbrand et al. 1992). To decrease any contamination in the "P" region, we took only those sources that were above the J − H = 0.89 threshold.
• 4.  
  Additionally, those NIR sources detected only in the H and K bands (i.e., using the H − K catalog from Section 2.3) were also used to identify YSOs using the H/H − K CM diagram (Figure 2(d)). From the H/H − K CM diagram, only the sources with an infrared excess upward of H − K = 1.5 were selected. This cut-off was chosen as—in the CM diagram—it marks a clear low-density gap between the field sources branch and the YSOs branch. CM diagrams of nearby non-nebular regions also show the sources in those regions (i.e., the field sources) to be confined below this value. YSOs selected using NIR photometry mostly fall into the Class II/III category.

Zoom In Zoom Out Reset image size

The YSOs selected from each of the above steps are not mutually exclusive and there are overlapping YSOs between various methods. The selected YSOs were hence merged (with the class of a YSO taken as being the class in which it was first identified in the above order of steps) to obtain a final catalog of 1202 YSOs, of which 40 are in the Class I category and 1162 are in the Class II/III category. The central high-mass Infrared Sources (IRS) sources (1A South, 2B, 3A, and 5) were detected as Class II in the above set of steps, but were expunged from the YSO catalog as they have already been identified as MS sources by Shuping et al. (2012). Table 2 gives the YSOs detected in the FoV. A sample of Table 2 is given here, whereas the complete table is available in electronic form as part of the online material.

The NIR photometry from the new WFCAM data is deeper by two magnitudes compared with the 2MASS data. The NIR J − H/H − K CC diagram of the region was used to select the reddened MS sources. Sources in the "F" region (see Figure 2(c)), which were not classified as YSOs in Section 3.1, were selected for this analysis (a total of 10,011 sources used here). We estimated the average visual extinction in the region using a method similar to the Near-IR-Color-Excess method of Lada et al. (1994), Kainulainen et al. (2007), and Ojha et al. (2004a). First, the (H − K) color excess for each of the selected sources was calculated by dereddening them, along the reddening vectors, to the dwarf locus that was approximated by a straight line asymptote. A_V for each source was subsequently calculated using the reddening laws of Rieke & Lebofsky (1985).

The dereddenned extinction values were used to produce an extinction map of the region, as shown in Figure 3. In producing this extinction map, at each position on the regular grid, a median extinction from the 20 valid nearest neighbors (NNs) was computed. The choice of the median instead of the average value is because the median works like an outlier rejector. Figure 3 shows two distinct features: (a) a low-extinction cavity around the ionizing stars and (b) higher extinction toward the west and north. The filamentary structures (whose locations have been marked with arrows) are partially traceable (with the one along the northeast being much more distinct than the one toward the south). Although the filaments are found to be very dense from PACS 160 μm emission (see Section 4.1), the extinction map fails to reproduce the same result because the NIR data are not deep enough to effectively detect many stars behind the filament. The depth of the NIR data is nevertheless good enough to trace extinctions of the order of A_V ∼ 20 mag, which is the maximum value found along the northeastern filamentary structure and northern regions.

Zoom In Zoom Out Reset image size

The average extinction of the diffuse, low-density remnant cloud covering most of the W40 H ii region is found to be A_V ∼ 15 mag. Note that the low-extinction cavity around the IRS sources in Figure 3 displays A_V ⩽ 12 mag, which agrees well with the previous extinction estimates for the dominant IRS sources (A_V ∼ 8–10 mag) in the region (Smith et al. 1985; Shuping et al. 1999; Rodríguez et al. 2010).

We carried out a YSO surface density analysis of the region using the NN method (Casertano & Hut 1985; Schmeja et al. 2008; Schmeja 2011). Figure 4 shows contours for 50 and 20 NN density overlaid on the PACS 160 μm image and the IRAC 3.6 μm image, respectively. A large value of NN is used to examine large-scale structures, while a lower value of NN is more sensitive to smaller-scale density variations and is used to trace sub-structures. This can be thought of some kind of smoothing process. Here, the 50 NN density contours are fairly circularly symmetric and were thus used to calculate the cluster radius (as opposed to using the 20 NN map). The dense central region for the 50 NN map was fit by a Gaussian profile whose FWHM comes out to be ∼3' (∼0.44 pc at a distance of 500 pc). There are 170 YSOs (from Section 3.1) within this cluster radius. The 20 NN density contours (being more sensitive to local density fluctuations) reveal the sub-structures of the cluster. The contours mirror the filamentary structures to a considerable degree, showing that the filaments also contain YSOs, albeit at low densities. The centers of both the 50 and 20 NN density contour maps are coincident with IRS 1A. Figure 5 shows the cumulative frequency distribution and the histogram for the 20 NN and 50 NN distances as the solid black line/curve and the dark gray dotted line/curve, respectively. As is expected, each feature (kink, peak, dip, etc.) that is observed in the 50 NN histogram/frequency polygon at, say, "x" NN distance, is observed at ∼"x/2" NN distance in the 20 NN histogram/frequency polygon. The cluster radius has been marked as a vertical dashed line. The cluster radius seems to be near an inflection point in the rising portion of the 50 NN frequency polygon.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

3.3.1. Comparison with Other Regions

The YSOs are not in particular found to be correlated with the lobes of the bipolar nebula, suggesting that the bipolar nebula is relatively young and not enough material has accumulated along the edge of the lobes to lead to star formation. There appears to be a prominent compact nebula (marked with a black arrow on 3.6 μm image in Figure 4) located at the edge of the southern lobe. The source associated with this compact nebula was not classified as any YSO (although there are Class II/III YSOs in its vicinity). It is likely to be a reflection nebula illuminated by the embedded star and seems to have been overrun by the expanding southern wall of the bipolar nebula.

Figure 6 shows the zoomed-in view of the central region with overlaid YSOs. Class I and Class II/III sources have been marked in blue squares and green circles, respectively. While Class II/III sources are distributed throughout the image, the Class I sources are located mostly along the filamentary structures extending out beyond the cluster radius (see details in Section 4.1). Additionally, protostars and starless cores from Maury et al. (2011) have been marked as blue diamond symbols and crosses, respectively. We note here that none of the Class 0/I sources or starless cores from Maury et al. (2011) had any YSO counterpart from our analysis within a 1'' matching radius, probably because these Maury et al. (2011) sources are younger (and/or are oriented edge-on), as opposed to the YSOs identified in Section 3.1, which are Class I or Class II/III sources. It is possible that smoothing by Maury et al. (2011) to a 13'' beam—while we are only matching up to 1'' here—might be the cause of the mismatch. Moreover, the classification by Maury et al. (2011) is using envelope mass versus bolometric luminosity diagram, inducing further uncertainty.

Zoom In Zoom Out Reset image size

The high-mass IRS sources have been marked with arrows and labeled. IRS 1A South—the main ionizing source—was found, using CC/CM diagrams, to exhibit excess infrared emission. This could be because it is located in the central region surrounded by high-density material. IRS 2B, 3A, and 5 (other high-mass sources, but of comparatively much lower mass than 1A South) were found to have a negative spectral energy distribution (SED) slope, with the respective spectral indices being ∼−1 for IRS 2B, −0.4 for IRS 3A, and −1.9 for IRS 5.

3.4.1. IRS 5 Nebular Region

3.4.2. Bright-rimmed Clouds

3.4.3. Comparison with Hα Emission

Figure 7 shows a color composite image made using PACS 160 μm (red), IRAC 3.6 μm (green), and Hα data (blue) from the SuperCOSMOS Hα survey (Parker et al. 2005). The dashed white box in the image marks the IRS 5 arc-shaped nebular region. The image shows the current star formation activity (traced by Hα emission) vis-a-vis the dispersed molecular material by the high-mass sources (traced by polycyclic aromatic hydrocarbons (PAHs) in the 3.6 μm band). As can clearly be seen, there is no correlation between Hα emission and the 3.6 μm emission, although this could partially be the effect of variable extinction. The Hα emission regions are also not correlated with the location of our detected YSOs. Hα is mainly present in the IRS 5 nebular region and in selected portions to the south of the midriff. Some of the Hα emission is correlated with BRCs and the so-called elephant trunks (to the east). It is likely that the stellar objects in these Hα emission regions are in extremely young stages.

Zoom In Zoom Out Reset image size

We determined the mass of the cloud within the central cluster radius and the IRS 5 region using the extinction map, assuming the empirically determined gas-to-dust ratio of 〈N(H₂)/A_V〉 = 0.94 × 10²¹ molecules cm⁻² mag⁻¹ (Ciardi et al. 1998; Lilley 1955; Jenkins & Savage 1974; Bohlin et al. 1978). This expression has been derived assuming that the total-to-selective extinction ratio (R_V) is 3.1 and that most of the gas is in the form of molecular hydrogen (Frerking et al. 1982; Krumholz et al. 2011). For the central cluster region (which has a radius of 3' centered on IRS 1A), first the column density at each pixel (of size 20'') was determined using the A_V value in that pixel. Thereafter, the column density values in all the pixels within the cluster area were integrated to obtain the total mass in that area. The same method was used to calculate the mass of the IRS 5 region—with the area taken as that extending from the edge of the central cluster circumference until approximately the edge of the arc-shaped nebula. The areas used for the calculation of masses in the central cluster and the IRS 5 region are mutually exclusive. The resulting masses obtained for the central cluster region and IRS 5 region were ∼126 M_☉ and ∼71 M_☉, respectively. These estimates are most likely lower limits as a non-negligible fraction of the molecular gas might have been expelled during the course of evolution of the cluster. Also, some of the gas will be present in form of ionized gas. Another caveat to be kept in mind is that R_V can, in general, have wide variations (Mathis 1990).

Figure 8 shows a Herschel PACS 160 μm image of the larger region (∼455 × 630) with overlaid YSOs and radio sources (in green plus signs) from Pirogov et al. (2012). The location of the molecular clump TGU 279-P7 from Dobashi et al. (2005) has been marked as a green cross. Apart from the circular lobes that join at the midriff, filamentary structures are also seen emanating toward the northeastern and southern directions from the central region. In Figure 8, three prominent parsec-scale filaments, marked on the image, can clearly be made out. Due to the high extinction in the region and by virtue of their intrinsically dense nature, the filamentary structures start becoming visible in emission only in the Herschel PACS 100 and 160 μm bands and at submillimeter SPIRE wavelengths (see Könyves et al. 2010 for the image of the larger region encompassing W40). They can also be seen in absorption against the dominant PAH (plus the continuum) emission at 11.3 μm and 12.7 μm in the WISE 12 μm band image (not shown here).

Zoom In Zoom Out Reset image size

Filament 1 is about ≳2.4 pc long and seems to be splitting into two parts near the midriff where it ploughs into the cloud. Possible signatures are the two bow shocks (the southern portion of filament 1, marked with an arrow) visible here (also see Figure 4 (left) for clarity). Filament 2, which is comparatively diffuse, extends up to ≳1.2 pc to the north of the central region, where it bifurcates into a filament to the north (≳1.6 pc in length) and another to the west (≳1 pc in length). Filament 3 is the longest visible structure, extending for ≳3 pc from the western part of the central region, and seemingly cutting across the southern lobe of the bipolar nebula. According to the column density map of Könyves et al. (2010), the value throughout—for all the three filamentary structures—is of the order of $N_{{\rm H}_{2}} \sim 10^{22}$ cm⁻². The filamentary structures are also seen in dust temperature as well as extinction maps (Bontemps et al. 2010; see their Figures 1 and 3).

Figure 8 also shows overlaid YSOs, with Class I sources, protostars, and starless cores marked as green squares, cyan diamond symbols, and cyan crosses, respectively. As can be seen in the image, most of the sources seem to be distributed along filaments 1 and 3 that point radially outward from the central region, which has a concentration of these YSOs. Filament 2 hardly contains any of these youngest sources, probably because of its diffuse nature, suggesting that it is most likely still not Jeans critical.

Myers (2009) had proposed a hub-filament model according to which a central "hub" with a column density of ∼10²² cm⁻² radiates filaments that can be seen up to a column density of ∼10²¹ cm⁻². It was suggested that the hub should have a minimum surface density of 25 YSOs pc⁻². The filamentary structures of the W40 region resemble this hub-filament structure. A possible "hub" region is marked in Figure 6 by a white rectangle. This "hub" region has a stellar density of ∼358 YSOs pc⁻² (the total number of YSOs divided by the total area), while the mean column density over this hub region is N$_{{\rm H}_{2}}$ ∼ 1.8 × 10²² cm⁻² (Maury et al. 2011). Among the sources from our analyses and the literature, a total of 12 Class I sources, 6 protostars, and 126 Class II/III sources are located in this "hub." This translates to the fraction of youngest sources (Class I/protostars) being ∼0.125 and thus, according to the criteria used by Myers (2009) (that this fraction should be ⩾0.1), this region can be deemed "young."

There are filaments to the west of this bipolar nebula that are a part of the Serpens South cluster (not shown here). Although they appear to be spatially correlated with the W40 region, this could be a projection effect since the distance to Serpens South is about 260 pc (Nakamura et al. 2011), while the distance to W40—although with varying values based on different techniques—has been estimated to be about 500 pc (Shuping et al. 2012).

In Figure 9, we show the continuum-subtracted H₂ narrowband image using an inverted gray scale. The brightest emission from the Herschel PACS 160 μm image is overlaid using dashed contours. Several important features are marked on this figure that are explained below. Comparing the H₂ emission with the 160 μm contours indicates that most of the observed H₂ emission traces the boundaries of filaments and dense gas traced by the 160 μm contours. The H₂ emission also reveals a faint bipolar-shaped cavity centered on the W40 embedded cluster. The northern lobe of this cavity is clearly brighter than the southern lobe, however, the 160 μm emission is also stronger in this northern region. BRCs can be seen along the southern boundary of this bipolar-shaped cavity in the IRAC images (see Figure 6). The morphology of the observed H₂ emission features are indicative of fluorescent excited emission in most of the observed field. We note that the arc-shaped bipolar cavity in the northern region, partially surrounding the IRS 5 source, is also the region where several starless cores were mapped by Maury et al. (2011; see Section 3.4). It may therefore represent a dense shell swept by the radiation from the central embedded cluster. While the lack of collimated bipolar features indicates the absence of shocked emission in general, confirmation of the fluorescent or shocked nature of the diffuse H₂ emission is not possible without obtaining spectroscopic observations.

Zoom In Zoom Out Reset image size

Figure 10 shows the Spitzer 3.6 μm image of the central portion of the W40 region. Overlaid are the 610 MHz and 1280 MHz radio contours from GMRT. The early-type stars in the region—IRS 1A South, 2B, 3A, and 5—have been marked with white squares and labeled. The ¹²CO(J = 1–0) peaks from Smith et al. (1985) have been marked with plus signs. It can clearly be seen that one of the ¹²CO(J = 1–0) peaks lies at the western edge of the H ii region. The ¹³CO(J = 2–1) and ¹³CO(J = 3–2) maps from Zhu et al. (2006) also reveal the presence of a molecular cloud to the west of the ionizing region. This molecular cloud to the west of the H ii region has also been observed in various molecular lines and in the continuum at millimeter wavelengths (Pirogov et al. 2013). The region thus appears to be density bounded to the west, as is evidently borne out by the CO isotopologues' peaks. However, there is no such constraint toward the east, leading to the proliferation of ionizing photons toward this direction. This is indicative of a blister morphology, also referred to as a champagne flow (Tenorio-Tagle 1979; Whitworth 1979) in the region. The radio contours depicted in the image also support this blister morphology for the region, with the head of the ionized flow toward the west (density-bounded) andextended emission toward the east. The peak of the 1280 MHz H ii emission is about 124 to the southwest of the peak at 408 MHz from Goss & Shaver (1970), but this could be due to their smaller beamsize (∼3'). As can be seen in Figure 10, there are no Spitzer Class I YSOs near this peak; only Class 0/I sources or starless cores from Maury et al. (2011) are seen. Also, one of the Class 0/I sources from Maury et al. (2011) coincides with the 1280 MHz radio peak. This is a probable indication of some high-mass star formation going on in this region, which is also supported by the outflow seen by Zhu et al. (2006). The timescale of the outflow from Zhu et al. (2006; ∼6 × 10⁴ yr) is of the order of the age estimate from Maury et al. (2011; ∼4–9 × 10⁴ yr) of the Class 0/I sources. However, it is plausible that the "outflow" observed by Zhu et al. (2006) might simply be the material swept up and pushed into the bipolar-shaped nebula. The H ii emission does not extend all the way up to the edge of the bipolar lobes (which marks the edge of the photodissociation region) or trace them at GMRT sensitivity levels. The extent of the central H ii region is almost the same as the cluster radius at half-peak surface density (see Section 3.4 and Figure 4), which is expected. The central ionizing region is limited to ∼2' (∼0.29 pc) on either side of the midriff of the bipolar lobes. Although they are isolated, the non-thermal radio emission sources have been discovered by Pirogov et al. (2012; see Figure 8). It seems unlikely that they are related to the W40 region due to their non-thermal nature. Another feature of note is the sub-region around IRS 5, which shows distinct radio emission with a roughly circularly symmetric ionization region around this source.

Zoom In Zoom Out Reset image size

The presence of these various morphological structures supports the schematic for this region by Vallee (1987; see their Figure 2). The high-mass stars formed are probably located at the edge of the parental giant molecular cloud toward the observer. As these sources have ionized the natal medium, the whole H ii region has broken out of the parental cloud. The filaments seem to be emanating from the midriff of the W40 region. This suggests that the density was high along the equatorial axis, as compared with the polar directions, leading to a high density contrast. Thus, the Lyman continuum radiation has escaped along the polar directions, flinging the ionized gas (which forms the bipolar nebula) partially toward the observer. This flung-off material seems to have encountered more resistance toward the north—as seems likely by the smaller diameter of the northern lobe, more extincted regions toward the north, and brighter 160 μm emission from the northern lobe (see Figures 8). This probably indicates that the main body of the cloud lies to the north of the W40 midriff region. The initial orientation and location of the high-mass stars seem such that the presence of this molecular cloud to the north has proven harder to disrupt, while the southern thin shell of molecular material (between the high-mass stars and the interstellar medium has been blown away relatively easily. We note that the southern lobe could be appearing larger than the northern lobe due to the projection effect from the viewing angle, i.e., the region might not be being viewed exactly face-on, but from a non-zero inclination angle. As has also been shown in Beaumont & Williams (2010) for various other regions, these northern and southern lobes could be more like rings than spheres, opening up the possibility that the molecular cloud could be oblate or sheet-like. The morphology of the filamentary structure shows its complete extent until the filaments join the central region. Since the filaments can be seen right until they join the midriff, they are most likely in front of the progenitor molecular cloud.

Radio continuum observation data were used to derive characteristic physical parameters like emission measure, electron density, dynamical age, and Strömgren radius for the region. Since this region contains multiple radio sources (Rodney & Reipurth 2008), we carried out a global analysis using the low-resolution images (∼45'' × 45'') to obtain an estimate of the parameters. The integrated flux density was calculated for both the frequencies within 4σ contours using the AIPS task "TVSTAT," giving a value of 4.07 Jy and 7.55 Jy at 610 MHz and 1280 MHz, respectively. Thereafter, using the model for free-free emission of Mezger & Henderson (1967), the flux density in a region is given by (adapted from Mezger et al. 1967)

$$\begin{equation} S_{\nu }=3.07\times 10^{-2}T_{e}\nu ^{2}\Omega (1-e^{-\tau (\nu)}) \end{equation} \tag{ 1 }$$

$$\begin{equation} \tau (\nu)=1.643\times 10^{5}aT_{e}^{-1.35}\nu ^{-2.1}n_{e}^{2}l, \end{equation} \tag{ 2 }$$

where S_ν is the integrated flux density in Jy, T_e is the electron temperature of the ionized core in K, ν is the frequency in MHz, n_e is the electron density in cm⁻³, l is the extent of the ionized region in pc, τ is the optical depth, a is the correction factor, and Ω here is the solid angle subtended by the beam in steradians. $n_{e}^{2}l$, called the emission measure, measures the optical depth in the medium (in cm⁻⁶ pc). For our calculation, we took T_e to be 8500 K (Shaver & Goss 1970; Quireza et al. 2006) and thus the correction factor a to be 0.99 (Mezger & Henderson 1967, Table 6). Using the two data points for 610 MHz and 1280 MHz, we fit the above equations using a non-linear regression with the emission measure ($n_{e}^{2}l$) as a free parameter (Figure 11(a)). This yielded an emission value measure of 1.63 ± 0.09 × 10⁶ cm⁻⁶ pc. The extent (l) of the central H ii region is approximately 70, equivalent to about 1.02 (±0.41) pc at a distance of 500 (±200) pc. Therefore, the electron density turns out to be ∼1265 ± 218 cm⁻³ for this central region.

Zoom In Zoom Out Reset image size

The Lyman continuum luminosity (photons s⁻¹) required for ionizing the gas was calculated using the formulation of Moran (1983, see their Equation 5) for the optically thin regime. As is evident from Figure 11(a), the 610 MHz and 1280 MHz emission lies in the optically thick and thin regimes, respectively. Therefore, using the flux density of 7.55 Jy estimated at 1280 MHz, the Lyman continuum luminosity was calculated to be ∼1.67 × 10⁴⁷ photons s⁻¹. Shuping et al. (2012) estimated that the radio emission is dominated by an O9.5 type star (IRS 1A South, see Figure 10) with a luminosity of 6.3 × 10⁴⁷ photons s⁻¹. They place an upper limit on the Lyman continuum luminosity of 1.6 × 10⁴⁸ photons s⁻¹. Smith et al. (1985), using infrared and low-resolution (∼11') radio observations of Altenhoff et al. (1970), inferred a luminosity of ∼1.5 × 10⁴⁸ photons s⁻¹. Our Lyman continuum luminosity estimate differs by a factor of the order of unity from the calculated flux of an O9.5 type star in Shuping et al. (2012) and the tabulated value from Martins et al. (2005). However, it is about 10% of the value from Smith et al. (1985) and the upper limit given in Shuping et al. (2012). This could be because of absorption of photons by the dust in the region, implying the presence of dense gas in the central region. As has also been noted by Kurtz et al. (1999), the presence of dust absorption can lead to an underestimation of the radio luminosity. Arthur et al. (2004) estimated that dust absorption can range from 60% up to 96% of the ionizing photons. A similar underestimation of the ionizing flux using radio data has been observed for other regions in Brand et al. (2011) and Alvarez et al. (2004). The luminosity calculated here, therefore, is most likely just the lower limit.

In the first stage during the ionization of a region by massive star(s), the ionization front propagates outward, leading to an expansion of the H ii region, until equilibrium is reached between the number of ionizations and recombinations. The radius of the H ii region at this point, assuming that the ambient medium has a spatially uniform density, is given by (Strömgren 1939)

$$\begin{equation} R_{s}=\left(\frac{3S_{{\rm Lyman}}}{4\pi n_{o}^{2}\beta _{2}}\right)^{1/3}, \end{equation} \tag{ 3 }$$

where R_s is called the Strömgren radius (in cm), n_o is the initial ambient density (in cm⁻³), and β₂ is the total recombination coefficient to the first excited state of hydrogen. The value of β₂, for T_e = 8500 K, was taken to be 2.94 × 10⁻¹³ cm³ s⁻¹ (Ward-Thompson & Whitworth 2011; Stahler & Palla 2005). In the second stage of expansion, after the Strömgren radius has been reached, the shock front overtakes the ionization front due to the vast pressure difference between material inside and outside of the ionization front. In this stage, the radius of the H ii region is given by (Spitzer 1978)

$$\begin{equation} R(t)=R_{s}\left(1+\frac{7c_{II}t}{4R_{s}}\right)^{4/7}, \end{equation} \tag{ 4 }$$

where R(t) denotes the extent of the H ii region at time t and c_ii is the speed of sound, taken as 11 × 10⁵ cm s⁻¹ (Stahler & Palla 2005). Since we have no way of gauging the initial ambient density (n_o) of the molecular cloud, we have plotted the parameters for ambient densities ranging from 1000 to 15,000 cm⁻³ to get an estimate of the ranges. Figure 12 shows the Strömgren radius (R_s) and dynamical age (t) plotted as a function of the initial ambient density of the medium. The Strömgren radius varies from ∼0.17 to 0.03 pc, while the dynamical age varies from ∼0.19 to 0.78 Myr. If we were to take the higher estimates of S_Lyman (the values discussed above), then, for a fixed density, R_s will increase and t will decrease. It should be noted that there is more than one massive star in the region and each massive star will have its own Strömgren radius and age. The multiple 3.6 cm Very Large Array (VLA) sources found in the region (Rodríguez et al. 2010) will also contribute to the total flux. Also, the radio morphology is not spherical here. Therefore, the above calculations should be taken as a representative value for the Strömgren radius and age estimate and n_e should be taken as the averaged out electron density for the ionized region.

Zoom In Zoom Out Reset image size

Since radio emission around the IRS 5 source is distinct from the rest of the morphology (see Figure 10), we can carry out a separate calculation to determine the physical parameters around this source. For this, we first determined the 610 MHz and 1280 MHz integrated flux density in this region in a circle with a 50'' radius centered at the peaks of the respective frequencies. This yielded a flux density of 121 and 109 mJy for 610 MHz and 1280 MHz, respectively. Rodríguez et al. (2010) detected IRS 5 source using VLA observations at 3.6 cm. However, they calculated a flux of 0.92 mJy for a sub-arcsecond area, while here we are concerned with the extended emission. Another VLA source from Rodríguez et al. (2010, Source 2 in their Table 2) lies close to the IRS 5 source. However, it has a flux of 0.90 mJy (at 3.6 cm) and is hence insignificant for our calculation here. Following a similar set of steps as above (Section 5.1), we determined the emission measure by fitting the free-free emission SED (see Figure 11(b)) to be ∼2.01 ± 0.04 × 10⁴ cm⁻⁶ pc. The extent of radio contours for this region is ∼100'' (∼0.24 ± 0.10 pc at a distance of 500 ± 200 pc), which gives an electron density of ∼288 ± 55 cm⁻³.

We estimated the Lyman continuum luminosity using the formulation of Moran (1983) and the 1280 MHz data point, as in Section 5.1 (in this case, since the 610 MHz data point is in the optically thin regime too, it yields a similar value), which results in a value of ∼ 2.40 × 10⁴⁵ photons s⁻¹. Comparing log (Lyman continuum luminosity) (45.38) with the tabulated values from Panagia (1973, Table II), we get the most probable spectral type of IRS 5 being B1V, in agreement with the values from Shuping et al. (2012). The slightly lower value of the calculated Lyman continuum luminosity, as compared with the value from Panagia (1973), is probably due to the absorption of photons by dust in the region.

Although the radio contours are seen within a ∼50'' radius of the respective radio peaks (Figure 10), it can be seen that the ionized region extends beyond that. This weak extended emission region is traced by PAH emission at 7.7 μm and 8.7 μm in IRAC Ch4 and at 3.3 μm in IRAC Ch1 (which, it must be noted, may also contain continuum emission; also see Figure 6). Most of Class 0/I sources and starless cores from Maury et al. (2011) are approximately along the circumference of a circle of radius ∼80'' centered at respective radio peaks. Therefore, taking the extent to be 160'' and the ambient density (n_o) to be a typical value of ∼1000 cm⁻³ (Stahler & Palla 2005), we use Equation (3) in conjunction with Equation (4) to obtain the dynamical age of the H ii region around the IRS 5 arc-shaped nebula to be ∼0.11 Myr. We note that though the ionized region appears more extended than what we have assumed in the above calculation and the ambient density of the medium could be larger, a higher value of both of these parameters will only increase the value of the dynamical age. Therefore, 0.11 Myr should be considered a lower limit for the dynamical age of this sub-region.