STAR-FORMING ACTIVITY IN THE H ii REGIONS ASSOCIATED WITH THE IRAS 17160–3707 COMPLEX

In order to investigate the nature of the ionized emission toward this region, we constructed a spectral index map using data at 1280 and 610 MHz. The spectral index α is defined as S_ν ∝ ν^α, where S_ν is the flux density at frequency ν. As we are interested in a spectral index map that includes regions with large scale diffuse emission, we used 1280 and 610 MHz images constructed from visibilities within the same UV range: 0.2–50.0 kλ. The new 1280 MHz image was convolved and regridded to that of the new 610 MHz map using the AIPS tasks CONVL and LGEOM. Only pixels with flux densities larger than 5σ in both bands were retained for generating the spectral index map. The AIPS task COMB was used for this. An error map was also obtained to compare the pixel uncertainties in the spectral index image. We find that the errors in spectral indices are <0.5. The envelope exhibits relatively large spectral index errors: 0.3–0.4 as compared to the interior. However, we would like to mention that these errors do not include uncertainties due to calibration processes.

The 1280-610 spectral index map and the corresponding error map are shown in Figure 3. This map shows the variation of the spectral index toward each of the radio sources we have identified, enabling us to infer the mechanism behind the radio emission. The range of spectral indices in the map lies between −1 and 1.5, indicating the presence of both thermal free–free emission as well as non-thermal synchrotron emission. Thermal emission including those from inhomogeneous H ii regions with density gradients give rise to spectral index values −0.1 ≤ α ≤ 2 (Olnon 1975), while spectral indices α < −0.5 are believed to be mostly due to non-thermal mechanisms (Kobulnicky & Johnson 1999). Toward the radio compact sources: S5, S6, S7 and complex A, the spectral index is positive, lying between 0.7 and 1, suggesting thermal emission. Away from the peak, the spectral index decreases as we move toward the envelopes. On the other hand, the spectral indices toward S10 and S11 are flat, signifying optically thin emission. The spectral index decreases to a value of ∼−1 in the envelopes in some pockets, indicating that non-thermal synchrotron emission dominates here. However, we treat this with caution as the errors are also larger. In addition, we see negative values of spectral index toward the rim of CS-112, implying the influence of non-thermal emission. This is discussed further in Section 4.1.3.

Zoom In Zoom Out Reset image size

We have also compared the spectral indices using higher frequency maps: 4.8 and 8.6 GHz ATCA maps from Martín-Hernández et al. (2003) as well as the 5 GHz VLA map. The 5 GHz VLA image is pipeline-calibrated and processed using AIPS for observations carried out in the BC configuration on 1989 June 23 under the proposal ID AB544. The image was extracted from the NRAO VLA Archive Survey⁶ (NVAS). The resolution of this image is 93 × 55, which is the lowest among the five radio images. The resolution of the 4.8 GHz ATCA map is 31 × 17 while it is 14 × 08 at 8.6 GHz. All the images are convolved to the lowest resolution to compare the spectral indices. The range of visibilities in the higher frequency images are different from the UV range of the GMRT images and hence the comparison is just illustrative. The spectral indices using five frequencies (610, 1280 MHz, 4.8, 5 and 8.6 GHz) toward eight representative positions (five compact sources and three locations in the envelope (P1, P2, P3)) are compared with those obtained from the two-frequency GMRT spectral index map. These locations are marked in Figure 3 and the plots of spectral index are shown in Figure 4. A comparison of these spectral indices is listed in Table 4. Column 1 lists the locations, Column 2 lists the GMRT 1280-610 spectral index values from the map, and Column 3 lists the spectral index estimated using five frequencies: 8.6, 5, 4.8 GHz, 1280 and 610 MHz. The spectral indices (from five frequencies) consistently indicate thermal emission toward compact sources while the spectral indices are negative or flat near the locations in the envelope. In Figure 4, we observe that the flux density toward S11 at 4.8 GHz (ATCA) is considerably lower than the flux densities at other wavebands. This is likely to be due to the sensitivity of ATCA and VLA observations to different spatial scales.

Zoom In Zoom Out Reset image size

Table 4.  Spectral Indices of the Eight Locations shown in Figure 3

Location	${\alpha }_{1280\mbox{--}610}$	${\alpha }_{\mathrm{GMRT}+\mathrm{VLA}+\mathrm{ATCA}}$
S5	0.60 ± 0.06	0.27 ± 0.11
S6	0.78 ± 0.07	0.26 ± 0.19
S7	0.86 ± 0.11	0.93 ± 0.12
S10	−0.08 ± 0.06	−0.06 ± 0.20
S11	−0.05 ± 0.23	0.17 ± 0.27
P1	−0.67 ± 0.17	−0.27 ± 0.27
P2	−0.52 ± 0.13	−0.28 ± 0.14
P3	−1.07 ± 0.26	0.07 ± 0.19

Note. Note that the errors are likely to be larger than the quoted values as they do not include errors due to calibration processes.

Download table as:  ASCIITypeset image

It is to be noted that the emission at each frequency could include contributions from both thermal and non-thermal processes and this contribution varies as a function of frequency. Toward P1, P2 and P3, the lower frequency spectral indices (1280-610) are more non-thermal in nature (i.e., steeper negative index) when compared to all the five frequencies. The likely reason could be that thermal contribution dominates at higher frequencies.

Overall, we see that the compact sources are thermal in nature while the co-existence of a non-thermal component is seen toward a few pockets in the outer envelope as well as toward the rim of CS-112. Similar variations in the spectral index map (interior versus envelope) are witnessed in other star-forming regions (Curiel et al. 1993; Garay et al. 1996) where non-thermal emission is detected toward the outer envelope. Garay et al. (1996) propose that the emission from the diffuse envelope corresponds to synchrotron emission from electrons accelerated in the region of interaction between a central star wind and the ambient cloud material. According to Curiel et al. (1993), this can be expected in a shock wave where most of the electrons having a thermal distribution of velocity will produce the thermal free–free component while a fraction of electrons will be accelerated to relativistic velocities producing the non-thermal component.

A scrutiny of cold dust emission at 160 μm and longer wavelengths reveals the presence of a number of dust clumps. We prefer to use the 870 μm image to identify the dust clumps for two reasons: (i) this wavelength traces the coldest dust clumps, and (ii) the dust emission is optically thin. We used the 2D-Clumpfind algorithm discussed in Section 3.1 with a threshold of 15σ and contour spacing of 1σ to identify seven clumps, where the rms (σ) value is ∼50 mJy/beam. The clump structure has also been explored through the dendrogram algorithm (Rosolowsky et al. 2008) and the resulting tree structure shown in Figure 6 identifies these seven clumps for the same threshold and contour spacing. The seven clumps found in this region are labelled as C1, C2, ..., C7 (in order of decreasing brightness). The dendrogram format distinctly exhibits branches and leaves in the tree structure. The clumps C1, C2 and C3 lie within a large structure that also contains C4 and C5 as leaves. The hierarchial structure is evident (Houlahan & Scalo 1992) although the statistics are inadequate to comment on the general nature of the molecular cloud. In Figure 6 we show the clumps associated with the 870 μm emission in this region. The arbitrary apertures corresponding to the area covered by each clump in the 870 μm image retrieved from the 2D-Clumpfind algorithm are overplotted in Figure 6. As we are interested in the average properties of each clump, we have constructed a clump spectral energy distribution (SED) using wavelengths 70–870 μm by integrating flux densities in the selected apertures. An average background value is estimated from a nearby field that is ∼13' away (centered at α_J2000 = 17^h 20^m 1425, δ_J2000 = −37° 19' 234) devoid of bright diffuse emission and this is appropriately subtracted to account for the zero offsets at each wavelength. We fitted the SEDs with a modified blackbody model of the form (Ward-Thompson & Robson 1990; Battersby et al. 2011; Olmi et al. 2013):

$$\begin{eqnarray}&&{F}_{\nu }={\rm{\Omega }}{B}_{\nu }({T}_{D})(1-{e}^{-{\tau }_{\nu }}),\end{eqnarray} \tag{ 4 }$$

$$\begin{eqnarray}&&{\tau }_{\nu }={m}_{{\rm{H}}}\mu N({H}_{2}){k}_{\nu }.\end{eqnarray} \tag{ 5 }$$

Zoom In Zoom Out Reset image size

Here F_ν is the flux density at frequency ν, Ω is the solid angle subtended by the clump, B_ν(T_D) is the blackbody function at temperature T_D, ${m}_{{\rm{H}}}\mu$ is the mean particle mass, N(H₂) is the column number density of molecular hydrogen, and k_ν is the dust mass opacity. m_H represents the mass of a hydrogen atom and μ is taken to be 2.86 under the assumption that the gas is 70% molecular hydrogen by mass (Ward-Thompson et al. 2010). The dust mass opacity is estimated using the following expression (Hildebrand 1983; Beckwith et al. 1990),

$$\begin{eqnarray}&&{k}_{\nu }=0.1\,{\left(\displaystyle \frac{\nu }{1000\mathrm{GHz}}\right)}^{\beta }\,{\mathrm{cm}}^{2}\,{{\rm{g}}}^{-1}\end{eqnarray} \tag{ .6 }$$

In this equation, β represents the dust emissivity index. The best fits were obtained using a nonlinear least-squares Levenberg–Marquardt algorithm considering N(H₂), T_D, and β as free parameters. Based on other studies, we assumed flux uncertainties of the order ∼15% in all bands (Schuller et al. 2009; Launhardt et al. 2013). The resolution of the 350 and 500 μm images are lower than the 870 μm image. As a result, clumps located close to each other, such as C1–C2 and C4–C5, are not resolved in these images. Nevertheless, we have integrated flux densities at 350 and 500 μm in the apertures and find that the best-fit parameters change negligibly (<10%) when we exclude fluxes at these two wavelength bands for these clumps. The best-fit SEDs are shown in Figure 7 and the values of the derived parameters for all the clumps are listed in Table 5.

Zoom In Zoom Out Reset image size

Table 5.  Properties of the Cold Dust Clumps Identified in this Region

Clump	αJ2000	δJ2000	Area	β	Temperature	Column density	${\chi }_{\mathrm{red}}^{2}$	Mass	L	Σ
	(h m s)	(° ' '')	(pc2)		(K)	(1022 cm−2)		(×102 M⊙)	(×102 L⊙)	(g cm−2)
C1	17:19:25.65	−37:10:07.9	3.3	1.7 ± 0.1	24.2 ± 0.9	5.9 ± 0.8	0.6	48	106	0.3
C2	17:19:26.60	−37:10:23.3	2.3	1.8 ± 0.1	26.0 ± 0.5	7.6 ± 0.5	0.1	44	143	0.4
C3	17:19:26.81	−37:11:01.9	4.7	1.7 ± 0.1	29.6 ± 0.9	4.1 ± 0.4	0.2	48	317	0.2
C4	17:19:35.73	−37:10:55.2	1.5	1.9 ± 0.2	25.1 ± 1.2	5.1 ± 0.9	0.9	19	63	0.3
C5	17:19:33.12	−37:10:48.2	2.5	1.8 ± 0.1	28.4 ± 1.5	3.0 ± 0.5	0.7	19	118	0.2
C6	17:19:21.87	−37:10:41.9	1.2	1.8 ± 0.2	28.3 ± 1.9	2.2 ± 0.5	1.0	7	38	0.1
C7	17:19:35.12	−37:10:14.0	0.9	1.4 ± 0.4	25.9 ± 3.9	1.7 ± 0.8	5.7	3	9	0.1

Download table as:  ASCIITypeset image

The dust temperature in the clumps lie between 24 and 30 K, with C3 showing the highest temperature and C1 having the lowest dust temperature. The molecular hydrogen column densities range within 1.7–7.6 × 10²² cm⁻² with the clump C2 showing the largest column density value among them. The range of dust emissivity index β values is 1.4–1.9. These values represent average values over the entire clump region and the distribution of column density and dust temperature are discussed in more detail in Section 3.2.3. We have also estimated the mass of each clump from the column density value using the expression

$$\begin{eqnarray}&&{M}_{c}=N({{\rm{H}}}_{2})\mu \ {m}_{{\rm{H}}}\,A.\end{eqnarray} \tag{ 7 }$$

Here M_c is the clump mass and A is its physical area. The area and mass of each clump are listed in Table 5. The clumps C1 and C3 are the most massive, with a mass of 4800 M_⊙ each, while C7 is the least massive with a mass of 300 M_⊙. The total mass of the cloud is 18,800 M_⊙. We estimated the effective clump diameter as the geometric mean of the largest and smallest size of the irregular clump aperture from the 2D-Clumpfind algorithm. The diameters of clumps in this region range between 1.0 and 2.4 pc. Urquhart et al. (2013b) investigated the properties of a large number of clumps associated with compact H ii regions and they find that the clump masses range between 10² and 10⁵ M_⊙ for clumps of size 0.2–6 pc. Our clumps fall well within this range. In addition, Svoboda et al. (2015) studied the properties of starless and massive star-forming clumps from the Bolocam Galactic Plane survey and find that the masses of clumps harbouring ultracompact H ii regions lie in the range 30–30,000 M_⊙ with radii lying between 0.1 and 4 pc. Fallscheer et al. (2013) for instance adopt a lower limit of 40 M_⊙ for identifying potential molecular cores that can form a high-mass star. The sizes of cores in their sample range between 0.4 and 1.1 pc. In our case, all the clumps have masses larger than 40 M_⊙. This is not surprising as the clumps other than C6 and C7 are associated with compact radio sources lying close to the peak emission (within 5''). If we presume that each of these compact radio source is excited by ZAMS star(s), this would necessitate the presence of massive star(s) in these clumps. Thus, these parameters characterize those of massive star-forming clumps and lie within the mass range mentioned in above studies. We have also estimated the luminosity of each clump by integrating under the best-fit SED. The clump C3 has the largest luminosity of 3.17 × 10⁴ L_⊙ while C7 is the least luminous with 900 L_⊙.

Another parameter used to probe massive star formation in molecular clouds is the surface density (Σ) defined as M_c/A. According to Krumholz & McKee (2008), clouds with a minimum surface density of Σ ∼ 1 g cm⁻² would be able to suppress fragmentation to form massive stars. The surface density values for clumps in this region are listed in Table 5. The values of Σ for the seven clumps lie in the range 0.1–0.4 g cm⁻². The maximum Σ is observed toward clump C2 that also displays the largest column density among clumps. None of our clumps satisfies the massive star-forming criteria of Krumholz & McKee (2008). As the clumps C1–C5 have massive embedded (proto)stellar objects, their surface density would signify values where massive stars are forming. These are surface density values post massive star formation in a sense. Therefore, such a comparison of Σ makes sense under the assumption that the density of the clump has not changed notably. It is worth noting, though, that C6 and C7 have the lowest surface densities of 0.1 g cm⁻² while the values of clumps with massive embedded stars are in the range 0.2–0.4 g cm⁻². These values are consistent with the values obtained toward a large sample of massive star-forming cores by López-Sepulcre et al. (2010), Miettinen & Harju (2010) and Giannetti et al. (2013). These authors have shown that it is likely that massive star-forming cores possess lower surface densities of the order of ∼0.2 g cm⁻². However, it must be borne in mind that these values are average values given the large size of clumps, and the actual surface density could be higher near the peak emission or dense core. This was also emphasized by Chambers et al. (2009) who carried out a survey of massive star-forming clumps and attributed lower surface densities to massive protostars in cores of lower densities.

We decided to estimate the bolometric luminosity of this region and compare it with the IRAS luminosity obtained earlier by others considering that the recent scientific surveys and missions have enabled a larger wavelength coverage. The global SED of this region is constructed using flux densities from Spitzer-IRAC, IRAS-HIRES (Aumann et al. 1990), MSX, Herschel-HiGal, and ATLASGAL within a circular region of radius 32 centered on ${\alpha }_{{\rm{J}}2000}={17}^{{\rm{h}}}{19}^{{\rm{m}}}30\buildrel{\rm{s}}\over{.} 3$, ${\delta }_{{\rm{J}}2000}=-37^\circ 10^{\prime} 25\buildrel{\prime\prime}\over{.} 7$. This region is selected keeping in view the extent of diffuse emission as well as the low resolution of the IRAS-HIRES 100 μm image (∼21). We interpolate the flux densities of the observed SED using power laws and numerically integrate under the SED to obtain the bolometric luminosity. The total luminosity is 8.3 × 10⁵ L_⊙ which is nearly a factor of two higher than the previous estimate of ∼3.6 × 10⁵ L_⊙ (Peeters et al. 2002) where the latter value assumes a distance of 5.7 kpc. The larger flux density at 8 μm as compared to the 12 μm band is explicable on the basis of PAH emission in the former and silicate absorption in the latter. The full SED of this region is shown in Figure 8. A comparison of the bolometric luminosity with the total estimated luminosity of the clumps (Table 5) shows that the former is an order of magnitude larger than the latter. This is attributed to (i) non-negligible contribution of near- and mid-infrared fluxes to the bolometric luminosity unlike the modified blackbody fits to the clumps, and (ii) contribution of diffuse emission from the larger area used to estimate the bolometric luminosity. Although the bolometric luminosity corresponds to a single ZAMS star of spectral type O5 (Panagia 1973), the presence of seemingly isolated compact radio sources and YSOs coincident with few compact sources ascertains that this complex is excited by a cluster of massive stars.

Zoom In Zoom Out Reset image size

In addition to clump SEDs, we have also constructed the line-of-sight averaged molecular hydrogen column density, dust temperature and β maps by carrying out pixel-to-pixel graybody fits in the wavelength regime (70–870 μm). While in the previous section we obtained the clump-averaged properties, here we obtain the column density and dust temperature on the pixel scale. Although a few clumps are not resolved because of poor resolution (369), we can investigate the morphological distribution of dust properties. The 70, 160, 250, 350 and 870 μm images are convolved and regridded to the lowest resolution (369) and pixel size (14'') of the 500 μm image. The kernels used for the convolution of the Herschel images are taken from Aniano et al. (2011), while a Gaussian kernel is used for the 870 μm image. As earlier, the average background value estimated from a nearby field is subtracted. We fitted the flux densities at each pixel with the modified blackbody model defined earlier (Equations (2)–(4)).

The column density, temperature, and dust emissivity index (β) maps along with the reduced χ² (${\chi }_{\mathrm{red}}^{2}$) image are shown in Figure 9. For further analysis and comparison with average clump values, we consider pixels within the 15σ contour of the 870 μm emission, same as the threshold used for clump identification. The fits are relatively good as ${\chi }_{\mathrm{red}}^{2}\lt 2$ for these pixels. The peak column density is 1.1 × 10²³ cm⁻² toward clumps C1–C2, while the median column density in our region of interest is 3.0 × 10²² cm⁻². The column density values agree well with those of other massive star-forming regions (e.g., Garay et al. 2007; Dunham et al. 2010). The column density distribution shows two peaks: toward C1–C2 and C4, separated by 3.2 pc. The clumps are not resolved owing to the poor resolution of the map. A comparison of the column density and temperature values obtained within our region of interest is compared with the average values obtained from clumps in Figure 10. Clearly, the average clump estimates fall within the values occupied by the pixels.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The maximum temperature in this region is 35.5 K and peaks at ∼70'' to the west of the column density peak. The mean value of dust temperature is 28.9 K. We would like to emphasize that the ${\chi }_{\mathrm{red}}^{2}$ for pixels near the highest temperature are larger than the surrounding pixels. This is evident from Figure 9(d). The dust temperature peak can be understood when we take into consideration the 70 μm emission. The 70 μm flux density is the shortest wavelength used in the SED construction. Consequently, pixels with significant emission at 70 μm and negligible cold dust emission are weighed by the 70 μm emission leading to higher dust temperature. This explains the large temperature to the west of complex A. This region is enclosed within a dashed line ellipse in Figure 5 over the 70 and 500 μm images, emphasizing the contrast in emission. This is also conspicuous from the 870 μm and 70 μm emission contours overlaid on the images in Figure 9. In addition to thermal emission from warm dust that also contributes to the SED, the 70 μm emission traces the contribution of very small dust grains (Russeil et al. 2013) and many studies have excluded this wavelength in their analyses (Anderson et al. 2010; Peretto et al. 2010). Our attempts to exclude the 70 μm data point from the pixel-to-pixel SED fitting lead to unreasonably high values of parameters toward many pixels as only the longer wavelength edge of the graybody function is sampled (considering that the maximum mostly occurs between 70 and 160 μm). The temperature enhancement seen to the south-west of complex A is also associated with warm–hot dust emitting at 8, 24 and 70 μm and with diffuse ionized gas (see radio contours overlaid on the 70 μm image in Figure 5). This prompts us to speculate that this region could be more evolved compared to the rest of the complex associated with cold dust emission. This is based on the premise that the molecular cloud (hence cold dust) here has dispersed leading to relatively low column density values. Alternatively, it is possible that this region is not affiliated to the molecular cloud of IRAS 17160–3707 but lies along the same line of sight. Considering that IRAS 17160–3707 lies in the inner Galactic plane, this possibility cannot be ruled out.

The dust emissivity index β gives information about the properties of dust grains and hence the β map of a star-forming region is very useful in gauging the dust properties of the cloud. The variation of β in this region is within a relatively narrow range 1.4–2.0, unlike a few other star-forming regions where the range is larger, extending from 0.8 to 3.1 (e.g., Anderson et al. 2010; Tabatabaei et al. 2014; Veena et al. 2016). The mean value of β is 1.7, which is lower than the value of ∼2.0 typically quoted in the literature (Ward-Thompson et al. 2002; Sadavoy et al. 2013). The lower values of β have been attributed to the presence of fractal dust grains, which are created from the coagulation of smaller sub-units (Wright 1987). These fractal dust grains are of low mass and are porous, electrically conducting structures. Previous studies of dust temperature (T_D) and β toward molecular clouds and the interstellar medium in general have shown that there exists an anti-correlation between T_D and β (e.g., Dupac et al. 2003; Désert et al. 2008; Paradis et al. 2010; Anderson et al. 2012). An anti-correlation is also apparent from the β and T_D maps shown in Figure 9 where the region displaying high temperature shows lower dust emissivity values and vice-versa. While this effect may have implications on the composition and size of dust grains, there could be pseudo-anti-correlation effects introduced either due to the fitting procedure itself as shown by Sadavoy et al. (2013) or due to noise effects and dust temperature variations in the line of sight (Shetty et al. 2009a, 2009b). Hence we treat this anti-correlation effect between T_D and β with caution as the temperature map could have a contribution from very small grains at 70 μm.

In order to learn about the evolutionary stages of the YSOs, we have constructed their near- and mid-infrared SEDs and fitted them with the radiative transfer models of Robitaille et al. (2007). The models assume an accretion scenario for the star formation process where a central star is surrounded by an accretion disk, an infalling flattened envelope and the presence of bipolar cavities. We have used the command-line version of the SED fitting tool where a large number of precomputed models are available. The SEDs are constructed using wavelengths ranging from 1.25 and 24 μm (see Table 6 for fluxes up to 24 μm). Where the 24 μm point source counterpart is not detected, we have assumed an upper limit to the 24 μm flux density by carrying out aperture photometry for the parameters given by Gutermuth & Heyer (2015). For sources saturated in the 24 μm image, we used the flux density corresponding to the saturation magnitude of 0.5 (Gutermuth & Heyer 2015) as the lower limit. A few sources appear point-like at 70 μm, such as those associated with IRS14, IRS20 and IRS22. For these, we have estimated fluxes within apertures of size 12'' after subtracting the background from a nearby emission-free region.

Unlike the 24 μm image, the 70 μm image shows point sources as elongated, possibly due to different scan speeds in two directions (Olguin et al. 2015). However, the aperture size is selected such that the elongated beam is included within the aperture. For the other YSOs, we assume the 70 μm clump flux density as the upper limit as it is difficult to isolate flux density corresponding to the source.

For wavelength bands at 160 μm and longer, we used the associated clump fluxes as upper limits for constraining the models as the resolution is poor.

The inputs to the model include distance to the source, approximate visual extinction (A_V) and the magnitude in each waveband. The approximate A_V is estimated by using the column density values at the locations of these sources in the column density map. For converting column density to A_V in magnitudes, we apply the following relation: $N({{\rm{H}}}_{2})={A}_{V}\times 1.1\,\times {10}^{21}$ cm⁻² (Güver & Özel 2009). Here, we present the results for IRS2, IRS11, IRS14, IRS20 and IRS22 as they are positionally coincident with the radio compact sources S12, S11, S5, S6 and S7, respectively (see Table 7). We consider ten best fit models for each source and these are presented in Figure 14. The range of model parameters is listed in Table 8. The best fit models estimate the mass range as 6.8–29.4 M_⊙, with age ≤0.2 Myr, emphasizing the youth of these sources. This highlights the fact that these are massive objects. The best fit model obtained for IRS20 has a large effective temperature of the host object, 36,200 K. For sources other than IRS11, the models indicate high effective temperatures of ∼30,000 K. This is vindicated by the existence of compact radio emission seen toward these sources. IRS11 has a lower effective temperature of few thousand kelvin, suggesting a protostellar evolutionary phase for S7, which is also relatively isolated with lack of diffuse emission in its vicinity, implying an earlier evolutionary phase.

Zoom In Zoom Out Reset image size

We have also fit models to all the other YSOs using 2MASS, IRAC and 24 μm flux densities, or upper limits. For longer wavlength bands, we used the clump flux densities as upper limits. The models are shown and listed in the Appendix (Figure A.1 and Table A.1). According to the best-fit models, IRS1, IRS7 and IRS13 are evolved (age ≥1.9 Myr) compared to the other YSOs. It is important to realise that, although the models are consistent with observed flux densities, the results must be treated with caution. An important caveat is the absence of PAH or small-grain continuum emission, which forms an important source of emission to mid-infrared fluxes in YSOs with hot stellar sources (Robitaille et al. 2006). This is significant in our case as the SEDs of all the YSOs are constructed mainly using IRAC and 24 μm fluxes. Another primary assumption of these models is that the SEDs of the massive YSOs are scaled-up versions of their lower mass counterparts and the massive star formation scenarios have not been incorporated in them (Robitaille 2008). This also explains why YSOs associated with radio sources (other than IRS20) show low temperatures for best fit models in spite of the presence of ionized gas emission.

Complex A spanning ∼1.2 pc is the brightest region in IRAS 17160–3707 at mid- and far-infrared wavelengths extending from 8 to 250 μm. The infrared and radio nebulosity toward this region indicates the presence of a massive embedded cluster based on morphological evidence. In Figure 15, we see that the 8 μm mid-infrared emission is seen away from the cold dust emission peak suggesting a region of high extinction. In the absence of higher resolution mid-infrared images, it is difficult to ascertain if the six radio compact sources in complex A possess an intrinsic ionizing source or are externally ionized. Such externally ionized shells on ultracompact H ii region scales may be produced around turbulent self-gravitating cores when expanding H ii regions encounter such cores (Mac Low et al. 2007). Unlike our radio maps where complex A comprises compact sources, at higher frequencies only fragmented emission is seen, prompting Martín-Hernández et al. (2003) to label it as a classical H ii region. We differ from these authors as it is possible that the emission from the compact sources is resolved out at high frequencies due to the interferometric filtering. In addition, this region is near the peak of clump C4 implying that it suffers large extinction as the associated nebulosity is neither seen in the optical DSS nor in the 2MASS near-infrared images. Our view is that complex A comprises multiple compact H ii regions that are prior to the classical expansion phase.

The radio compact sources S5, S6, S7, and S11 have associated warm and cold dust counterparts implying the presence of embedded YSO(s) that have not yet shedded their dust cocoons. This is also evident from the SED models discussed in Section 3.3.1, where the age obtained from the best fit model of the massive embedded objects is 0.2 Myr or less. The radio luminosities indicate ZAMS spectral types of B0–O9.5 for S5 and S6, and B0.5–B0 for S7 and S11. S12 has weak radio flux density indicating a later ZAMS spectral type of B1–B0.5 of the intrinsic stellar object (Panagia 1973). However, it is relatively bright in mid-infrared (unlike radio) when compared to the other sources in this region. The 8 μm emission shows a bright emission but at shorter wavelengths only diffuse emission is discernible, implying relatively high extinction toward the embedded object. An arc-shaped emission is seen to the south-east of S10 in mid-infrared as well as ionized gas emission. Such mid-infrared arc-shaped features have been observed toward clusters of massive stars and could be attributed to (i) expansion of an H ii region, (ii) bow-shocks due to high velocity stars (Peri et al. 2012) or (iii) dust or bow wave (Ochsendorf et al. 2014).

We next attempt to understand the bubble-like morphology of CS-112 seen in ionized gas and warm dust emission. This bubble has a partially broken morphology and is oriented in the east–west direction, with the radio source S6 lying to the south of the bubble. We estimate the extent of the bubble by considering its size along the N–S direction close to the location of S6. The size of the bubble is approximately 2 pc with a thickness 0.3 pc in the 8 μm image. A color-composite image of CS-112 showing the distribution of mid-infrared, radio and submillimeter emission is presented in Figure 16. The region associated with CS-112 is saturated at 24 μm.

Zoom In Zoom Out Reset image size

The bubble morphology is demonstrated by using intensity profiles along the lines shown in Figure 16. We see an emission enhancement in the bubble in the vicinity of S6. To establish that this is the case, we consider two lines through the bubble: (i) AB passing through the emission enhancement near S6, and (ii) CD passing through another portion of the bubble. Both the lines are selected such that the contribution from stellar sources near the bubble edges is negligible. The intensity profiles at 4.5, 5.8 and 8.0 μm are shown in Figure 17. Also shown are the emission profiles from ionized gas emission at 610 MHz. As we are interested solely in the relative intensities along the edges of the bubble, the profiles are scaled and shifted by arbitrary factors for clarity. From this figure, it is evident that the edge near S6 (i.e., near point A in the line AB) is brighter than the bubble rim near B. This is unlike the intensity profile through the line CD where the emission contrast across the bubble edges is lower. We believe that the bubble rim near S6 is brighter in both radio and mid-infrared due to the gas from S6 filling the bubble region bereft of molecular gas leading to higher intensity in the rim. In other words, the ionized gas expansion and leakage of UV photons from S6 would occur more easily in a region where ionized gas (due to the bubble) is already present. The excitation of PAHs toward the bubble rim near S6 could be because of leakage of UV photons from the latter. This would inherently imply that the bubble lies at the same heliocentric distance as S6.

Zoom In Zoom Out Reset image size

A comparison of radio and mid-infrared emission reveals that, unlike the mid-infrared emission seen at the periphery of the bubble, the ionized gas occupies its interior and inner rim. This can be understood in terms of excitation of PAHs in the PDR, beyond the ionization front of the expanding H ii region of the bubble (Watson et al. 2008). Similar morphologies have been observed toward numerous other bubbles (e.g., Deharveng et al. 2010; Ji et al. 2012; Liu et al. 2015) It has been postulated that such infrared bubbles are excited by massive star(s)/cluster of stars where the expanding H ii region excites the ambient medium giving rise to a ring of PAH emission (Churchwell et al. 2006). An alternative theory proposes the interaction of strong stellar winds with the surrounding interstellar medium to produce a spherical ring of shocked gas (Weaver et al. 1977).

The radio emission within the region denoted by the partial dashed ellipse in Figure 16 corresponds to N_Lyc = 3 × 10⁴⁸ s⁻¹ for 1280 MHz. Assuming this to be optically thin, this is equivalent to a ZAMS star of spectral type O8–O7.5 (Panagia 1973). However, this is an underestimate as some UV photons are absorbed by dust (e.g., dust emitting at 24 μm). The identification of the exciting star(s) is difficult considering the population overdensity in the near-infrared. Nevertheless, we have attempted to search for the bubble progenitor(s) toward the center based on (i) its hemi-spherical shape and (ii) direction of pillar-like structures protruding within the bubble toward the central region both in ionized gas and mid-infrared dust emission, shown by arrows in Figure 16 (Pomarès et al. 2009). The latter structures are believed to be shaped by the UV radiation field of the exciting star(s). We visually scrutinized the central region of the bubble in the IRAC 3.6 μm image and located a favorable candidate: a faint group of stars. This cluster is encircled in Figure 16. The emission from the cluster at 4.5 μm is evident in the intensity profile through AB in Figure 17 (left). The GLIMPSE Archive contains three sources and these are marked in Figure 16. However, due to lack of flux density information in all the four bands, it is difficult to classify them through color–color diagrams. Toward this region, we observe flux enhancements corresponding to six sources within a small region (circular region of radius 5'') unlike its surroundings. Unlike the IRAC images, 2MASS detects only a single source at the location of the cluster. This is likely to be due to the lower resolution of 2MASS (4'') as compared to IRAC (2''). We speculate that these could be the exciting sources and further studies (high-resolution imaging and spectroscopy) are required to confirm this view.

We see a void in the cold dust emission toward the bubble, particularly near the cluster region. This could be explained if we consider dispersal of the ambient molecular cloud by stellar winds from the exciting sources. We note that although there is a clump toward the west of the bubble (C2, see Figures 6 and 16), it is larger and more massive than the one toward the east (C7). This probably explains the shell feature seen distinctly to the west in the broken bubble morphology. We also observe that the radio emission toward the periphery of the bubble shows the contribution of non-thermal component. This is evident from the spectral index map of CS-112, shown in Figure 18. The non-thermal radiation can be attributed to the presence of relativistic electrons produced in shocked regions of gas along the bubble periphery and accelerated by the magnetic field (Benaglia et al. 2013).

Zoom In Zoom Out Reset image size

We can estimate the dynamical age, t_dyn, of the ionized gas emission associated with the bubble based on a simple model of expanding photoionized nebula, in a homogeneous medium (Dyson & Williams 1980). If the cluster is indeed the exciting source, it is possible to estimate the age of the bubble based on the time elapsed for the ionization front to move from the location of the cluster to the present site through the ambient medium using the following formulation:

$$\begin{eqnarray}&&{t}_{\mathrm{dyn}}=\left(\left[\displaystyle \frac{4{R}_{s}}{7{c}_{i}}\right]\right)\left[{\left(\displaystyle \frac{R}{{R}_{s}}\right)}^{7/4}-1\right].\end{eqnarray} \tag{ 8 }$$

Here c_i is the isothermal sound velocity in the ionized gas and R is the radius of the bubble. R_s is the radius of the Strömgren sphere, given by

$$\begin{eqnarray}&&{R}_{s}={\left(\displaystyle \frac{3{N}_{\mathrm{Lyc}}}{4\pi {n}_{o}^{2}{\alpha }_{B}}\right)}^{1/3}\end{eqnarray} \tag{ 9 }$$

${\alpha }_{B}$ is the radiative recombination coefficient. We assume α_B = 2.6 × 10⁻¹³ cm³ s⁻¹ and c_i = 10 km s⁻¹ considered for typical H ii regions (Stahler & Palla 2005, p. 865). n_o denotes the mean number density of atomic hydrogen and we estimate its value using the expression $3N({\text{}}{H}_{2})$/2L where L is the molecular core radius. The three clumps surrounding CS-112 are C2 (west), C5 (south) and C7 (east). For estimating the dynamical time, we consider the mean number density of clump C2 as it is in the direction of the shell. Using the clump parameters, we get n_o = 4.3 × 10⁴ cm⁻³.

Taking 1.0 pc as the bubble radius, we estimate the Strömgren radius to be R_S = 0.04 pc. This corresponds to an expansion timescale of 0.7 Myr for the bubble CS-112.

A simple model such as this has also been used to estimate the age of other bubbles such as N22 (Ji et al. 2012) and N115 (Xu & Ju 2014).

In summary, the presence of compact and ultracompact H ii regions indicates the youth of this region, which is consistent with SED modeling of massive YSOs as well as with the approximate age of the bubble. This star-forming flurry is reflected in numerous studies of masers carried out for this region.