A VERY LARGE ARRAY STUDY OF ULTRACOMPACT AND HYPERCOMPACT H ii REGIONS FROM 0.7 TO 3.6 cm

The VLA was used to observe H53α, H66α, H76α, and H92α RRLs toward G10.96, G28.20, G34.26, and G45.07. Table 1 gives the instrumental parameters of the observations. Table 2 summarizes the transitions that we observed toward each source (Sewiło et al. 2004a, and this paper) and the adopted distances.

The first channel of VLA spectral line data ("channel 0") contains the average of the central 75% of the observing bandwidth (generated by vector averaging channels; Wilcots et al. 2003). The broadband uv data from "channel 0" were edited and calibrated using the Astronomical Image Processing System (AIPS) software package, and the resulting calibration solutions were applied to the multi-channel spectral-line data. All sources were self-calibrated to increase the dynamic range of the images. The continuum level was determined from a linear fit to the line-free channels. Continuum subtraction was done in the uv plane with the AIPS task UVLSF. A line-free continuum data set and a 31 channel cube of recombination line emission were constructed. Both data sets were Fourier inverted using weighting intermediate between natural and uniform (Robust = 0 in AIPS task IMAGR) and CLEANed in the standard manner. The images were corrected for primary beam attenuation using the AIPS task PBCOR. Integration times and continuum noise values of the final images are listed in Table 3.

Source positions, deconvolved angular diameters, position angles, and total flux densities (Table 3) were determined by elliptical Gaussian fitting using the least-squares AIPS routine JMFIT. Continuum-subtracted line images were used to generate the spatially integrated H53α, H66α, H76α, and H92α line profiles. For each line, the integration was performed approximately over the area enclosed by the 50% contour with respect to the peak value. The subsequent spectral analysis of the RRLs was done using the IDL Programming Language.⁷ The spatially integrated line profiles were fitted by a single Gaussian. The results for individual sources are presented graphically in Figures 1–6 for the H53α, H66α, H76α, and H92α RRLs. The continuum images are shown in the same figures; the projected linear scales are based on the kinematic distances listed in Table 2.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In Sewiło et al. (2008), we reported the results of high-resolution (015) observations of G28.20 N that revealed the shell-like morphology of this HC H ii region. To investigate the morphology of G28.20 N at even higher resolution, here we use VLA K-band A-array archival data (Program ID AZ168) to create the high resolution (009) continuum image shown in Figure 4. These observations were taken on 2006 March 14, with a 3.125 MHz bandwidth of 63 channels of 48.828 kHz each. We used line-free channels of these spectral line data to make a self-calibrated continuum image with an rms noise of 85 μ Jy beam⁻¹. The measured 1.3 cm flux density of G28.20 N is 630 ± 63 mJy. The synthesized beam is 010 × 008 at position angle of −05.

We also present H₂O, OH, and CH₃OH maser positions relative to the radio continuum for each HC H ii region based on high resolution (interferometric) data from the literature. The relative positional accuracy of 22 GHz H₂O masers detected toward G45.07 NE is 01 (Hofner & Churchwell 1996). The 1665 MHz and 1667 MHz OH masers observed in G10.96 W and G28.20 N have an absolute positional accuracy of ∼03 in right ascension and ∼05 in declination (Argon et al. 2000). The absolute positions of the 6.67 GHz CH₃OH masers in G10.96 W and G28.20 N are accurate to ∼1'' (Walsh et al. 1998), while the positional uncertainty of the 44 GHz CH₃OH masers in G45.07 NE is 06 (Kurtz et al. 2004). The relative positional accuracy of 22 GHz H₂O masers detected toward G45.07 NE is 01 (Hofner & Churchwell 1996). The distribution of masers is shown in Figures 1(b), 4, and 6 for G10.96 W, G28.20 N, and G45.07 NE, respectively.

We used VLA K-band spectral-line archival data for G28.20 N HC H ii region (Program ID AP465) to supplement the single-dish water maser data available for this source in the literature. The observations were taken on 2008 April 4 with the VLA in C configuration. A 3.125 MHz bandwidth with 255 channels of 12.2 kHz each was used. We followed standard VLA data reduction procedures. The synthesized beam is 119×094 at position angle of −128. The typical rms of a line-free channel is 22 mJy beam⁻¹. We determined the positions and peak flux densities of the masers from Gaussian fits to the peak channel of each maser. The H₂O maser positions and observed parameters are listed in Table 4. The uncertainty of the H₂O maser positions in G28.20 N is 01.

We also investigate dust emission for our sample using data from the Spitzer Space Telescope GLIMPSE survey ("Galactic Legacy Infrared Mid-Plane Survey Extraordinaire"; Benjamin et al. 2003; Churchwell et al. 2009). The inner Galactic plane was imaged at 3.6, 4.5, 5.8, and 8.0 μm using the IRAC camera (Fazio et al. 2004) on the Spitzer Space Telescope. The IRAC spatial resolution ranges from 15 to 19 between the 3.6 and 8.0 μm bands. The IRAC image mosaics were created by the GLIMPSE pipeline. The absolute accuracy of the GLIMPSE point-source positions is 03 (Meade et al. 2009).

Physical properties of the sources were derived from the observed continuum parameters reported in Table 3 using the equations of Mezger & Henderson (1967) and Panagia & Walmsley (1978) for spherical, optically thin, homogeneous, and ionization-bounded H ii regions. Panagia & Walmsley (1978) presented modified versions of the Mezger & Henderson (1967) equations for geometry-dependent physical parameters such as the rms electron density (n_{e, rms}), the emission measure (EM), and the mass of ionized gas (M_H ii). The excitation parameter (U) and the Lyman continuum photon flux required to maintain ionization of the nebula (N'_c) do not depend on the source size or its morphology and were derived using the formulae from Mezger & Henderson (1967). The continuum optical depth (τ_c) was calculated from the Oster (1961) formula in a form presented by Mezger & Henderson (1967), using EM derived based on the results of Panagia & Walmsley (1978). The derived physical parameters for each source are presented in Table 5. The linear diameter of a sphere listed in Table 5 is the diameter of the spherical model for the source, rather than the apparent Gaussian half-power width of the source reported in Table 3. An optically thin homogeneous sphere has the same peak EM and total flux density as the observed source. The calculations take into account the dependence on the beam size as described in Panagia & Walmsley (1978).

The N'_c values were used to estimate a lower limit to the spectral type of the exciting star given in the last column of Table 5. The classification is based on the non-LTE, spherically expanding (owing to the stellar wind), line-blanketed (to account for the effects of metals on the atmospheric structure and emerging ionizing flux) model atmospheres derived by Smith et al. (2002) under the assumption that a single star ionizes a dust-free, ionization bounded H ii region. The assignment of a given spectral type with N'_c depends strongly on assignment of effective temperature with spectral type and how accurately stellar atmosphere models reflect reality. For these reasons, spectral types listed in Table 5 are approximations. The stellar spectral type corresponds to the value of radio flux density at the highest frequency observed.

A second set of physical properties of the nebulae was derived based on their peak brightness temperature; these are "peak physical properties" (Wood & Churchwell 1989). These quantities depend only on the peak brightness, not the measured diameter or morphology. Thus, they can be derived for all the sources from our sample including the blended components B and C in the G34.26 complex and the overresolved component E of G10.96. The brightness temperature T_B is given by the Rayleigh–Jeans approximation.

The peak optical depth was estimated from $T_{B} = T_{e} (1- e^{-\tau _c})$ assuming a kinetic temperature of T_e = 10⁴ K that is constant over the synthesized beam. The peak EM was calculated using the optical depth equation from Mezger & Henderson (1967). The relation n_e = 〈n²_e〉^1/2 = (EM/Δs)^1/2 was used to calculate the rms electron density (n_{e, rms}) averaged over the beam. The quantity Δs is the source size along the line of sight through the peak position. We assume that the sources in our sample have spherical geometry based on their projected circular symmetry; for Δs we adopt their geometrical mean angular diameters (half-power widths, or "source size" in Table 3) assuming that the extent of the source along the line of sight is the same as its extent in the plane of the sky. The peak physical properties are given in Table 6.

The Gaussian fit parameters of the line profiles are given in Table 7. V_LSR is the central line velocity relative to the local standard of rest, FWHM is the line full width at half-maximum, S_L is the flux density in excess of the continuum at line center, S_C is the continuum flux density, ${\frac{S_{L}}{S_{C}}}$ and $\int\!\frac{S_{L}}{S_{C}}\,{d}v$ are the line-to-continuum ratios at line center and integrated over the line profile, respectively. The continuum flux densities are determined by integrating over the same areas of the source used to generate the line profiles; thus they may be less than the flux densities given in Table 3 (integrated over the entire source). The uncertainties given in Table 7 are the formal 1σ (68.3% confidence level) errors of the nonlinear least-squares fit. These uncertainties are lower limits because they do not account for systematic errors.

Table 7. Gaussian Fits to the H53α, H66α, H76α, and H92α Radio Recombination Lines

Source		RRL	VLSR	FWHMa	SL	SbC	${\dsty\frac{S_{L}}{S_{C}}}$	$\int\! \dsty\frac{S_{L}}{S_{C}}\,\mathrm{d}v$
			(km s−1)	(km s−1)	(mJy)	(mJy)	(peak)	(km s−1)
G10.96+0.01	E	H66αc	11.4 ± 0.7	18.7 ± 1.8	12.6 ± 1.0	46.3 ± 2.3	0.27 ± 0.03	5.4 ± 0.7
	...	H76αc	15.8 ± 1.0	25.3 ± 2.5	9.7 ± 0.8	53.8 ± 2.7	0.18 ± 0.02	4.9 ± 0.7
	W	H66α	13.6 ± 0.4	28.5 ± 1.0	35.9 ± 1.1	147 ± 7	0.24 ± 0.01	7.4 ± 0.5
	...	H76α	13.3 ± 0.5	29.6 ± 1.3	19.9 ± 0.7	131 ± 7	0.15 ± 0.01	4.8 ± 0.4
G28.20−0.04	N	H53α	88.5 ± 0.5	39.0 ± 1.4	144 ± 4	372 ± 19	0.39 ± 0.02	16.1 ± 1.1
	...	H76α	80.1 ± 0.8	55.7 ± 2.2	17.0 ± 0.5	235 ± 12	0.07 ± 0.01	4.3 ± 0.3
	S	H53α	99.6 ± 0.4	19.9 ± 0.9	63.6 ± 2.4	65.3 ± 3.3	0.97 ± 0.06	20.7 ± 1.6
	...	H76α	99.3 ± 0.5	24.8 ± 1.3	7.3 ± 0.3	34.9 ± 1.7	0.21 ± 0.01	5.6 ± 0.5
G34.26+0.15	A	H53α	50.2 ± 0.4	21.9 ± 1.0	46.3 ± 1.7	62.9 ± 3.1	0.74 ± 0.05	17.2 ± 1.3
	B	H53αc	64.9 ± 1.0	34.8 ± 2.6	20.5 ± 1.2	58.6 ± 2.9	0.35 ± 0.03	12.9 ± 1.4
	C	H53α	47.9 ± 1.2	49.4 ± 3.7	589 ± 32	2587 ± 129	0.23 ± 0.02	12.0 ± 1.2

Notes. ^aFull width at half-maximum. ^bS_C is a continuum flux density of the same area of the source over which an RRL was integrated. It was determined using AIPS task IMEAN. ^cThe spectrum was Hanning smoothed.

Download table as:  ASCIITypeset image

The measured line widths given in Table 7 are broadened because of the finite velocity resolution of the observations. To determine the intrinsic line widths, the instrumental spectral resolution was deconvolved from the observed lines. The convolution function in the spectral frequency domain was assumed to be a Gaussian function with FWHM of 1.2 times the channel width (or twice the channel width if the spectrum was Hanning-smoothed off-line) and deconvolved from the observed line widths. The deconvolved line widths (FWHM_deconv) are given in Table 8.

Table 8. Corrected H53α, H66α, H76α, and H92α Radio Recombination Line Parameters and the LTE Electron Temperatures

Source		RRL	FWHMadeconv	$({\frac{S_{L}}{S_{C}}})_{\rm corr}$b	T⋆e c
			(km s−1)	(peak)	(K)
G10.96+0.01	E	H66α	15.5 ± 2.1	...	11400 ± 1600
	...	H76α	19.6 ± 3.2	...	8800 ± 1400
	W	H66α	27.8 ± 1.1	0.25 ± 0.01	7500 ± 500
	...	H76α	28.0 ± 1.4	0.16 ± 0.01	7400 ± 500
G28.20−0.04	N	H53α	38.5d ± 1.5	0.39 ± 0.02	7000 ± 400
	...	H76α	54.9d ± 2.3	0.08 ± 0.01	7300 ± 500
	S	H53α	18.8 ± 0.9	...	5900 ± 400
	...	H76α	22.9 ± 1.4	...	6800 ± 500
G34.26+0.15	A	H53α	20.9 ± 1.0	0.75 ± 0.05	6800 ± 500
	B	H53α	33.1 ± 2.8	...	8800 ± 900
	C	H53α	49.0 ± 3.7	0.24e ± 0.02	8700 ± 800

Notes. ^aFull width at half-maximum (FWHM) of the spectral lines corrected for broadening introduced by spectrometer and smoothing. The convolving function in the spectral frequency domain was assumed to be a Gaussian function with FWHM of 1.2 times a channel width (or twice a channel width if a spectrum is Hanning smoothed) and deconvolved from the observed line widths. Velocity resolutions for all frequencies are given in Table 1. ^bWe adopt the same uncertainties as for the fitted line-to-continuum ratios given in Table 7. ^cFor optically thin sources, the values of $(\frac{S_{L}}{S_{C}})_{\rm peak}$ from Table 7 were used to calculate T^⋆_e. ^dFor G28.20 N, the H53α and H76α line widths are slightly smaller that those reported in Sewiło et al. (2008) based on the same data: 39.7 ± 1.3 km s⁻¹ and 57.6 ± 2.2 km s⁻¹ for H53α and H76α lines, respectively. The difference in linewidths is due to different integration areas used in both papers. While Sewiło et al. (2008) integrated the lines over the entire source, the lines reported in this paper were integrated over the area enclosed by the 50% contour. ^eThe line-to-continuum ratios corrected for opacity for this source was derived from the peak optical depths listed in Table 6.

Download table as:  ASCIITypeset image

We calculated the LTE electron temperatures T^⋆_e for all the sources with reliable RRL measurements using the formula from Rohlfs & Wilson (1986):

$$\begin{equation} T_{e}^{\ast } = \bigg [ \frac{6.985 \times 10^3}{a(\nu, T_e)} \cdot \nu ^{1.1} \cdot \frac{(S_L/S_C)_{{\rm peak}}^{-1}}{\rm{FWHM}_{{\rm deconv}}} \cdot \frac{1}{1+Y^+} \bigg ]^{0.87}, \end{equation} \tag{ 1 }$$

where ν is in GHz, FWHM_deconv is in km s⁻¹, and T^⋆_e in K. Y⁺  =  N(He⁺)/N(H⁺) is the ratio of ionized He to H atoms and is assumed to be 0.1. The factor a(ν, T_e) is approximately 1 (Altenhoff et al. 1960). This equation was derived under several assumptions about H ii regions: (1) the structure is plane-parallel, homogeneous and isothermal, (2) all optical depths are small: |τ_l + τ_c| ≪ 1 and τ_c ≪ 1, and (3) the lines can be treated in the LTE approximation.

However, several of the observed sources have relatively high optical depths (τ_C from 0.02 to 0.2), causing attenuation of the line-to-continuum ratio. In these cases, the LTE electron temperatures have been calculated using the line-to-continuum ratios corrected for opacity rather than the measured values:

$$\begin{equation} (S_L/S_C)_{{\rm corr}} = (S_L/S_C)_{{\rm peak}} \cdot (e^{\tau _c}-1)/\tau _c. \end{equation} \tag{ 2 }$$

The corrected line-to-continuum ratios and the LTE electron temperatures are listed in Table 8. The uncertainties in T^⋆_e were estimated by a linear propagation of errors.

High-spatial resolution mosaics from GLIMPSE (Churchwell et al. 2009) are a unique tool for the study of massive star formation regions (MSFRs). The IRAC 3.6, 5.8, and 8.0 μm bands contain strong polycyclic aromatic hydrocarbon (PAH) features that probe conditions in photodissociation regions (PDRs) that separate the ionized gas in H ii regions from the surrounding molecular gas (e.g., Tielens et al. 1993). PAHs are large organic molecules containing ≳50 C atoms and are very abundant in the interstellar medium (e.g., Sellgren 1984; Allamandola et al. 1989). PAHs are excited by sub-Lyman ultraviolet photons but are destroyed in H ii regions by extreme-ultraviolet (EUV) radiation (Voit 1992; Povich et al. 2007). Nevertheless, H ii regions can be traced by the IRAC 4.5 μm band, which does not contain PAH features. The 4.5 μm emission of H ii regions is a combination of thermal emission from small grains, scattered star light, and the Brα recombination line. The 3.6 and 4.5 μm bands are sensitive stellar tracers. The stellar contribution at 5.8 and 8.0 μm is relatively low because the photospheric emission is on the Rayleigh–Jeans tail of the energy distribution and declines rapidly with increasing wavelength.

We present IRAC images in Figures 7–10. All the images are presented in Galactic coordinates and the intensity is scaled logarithmically. The positions of radio sources are marked or the radio contours are overlaid on the images for reference. Equatorial coordinates are indicated to facilitate comparison with the radio images shown in Figures 2–6. The GLIMPSE Point Source Catalog (green circles) and Archive⁸ sources (red circles) are overlaid on the 3.6 μm images which show the stellar distribution in each field. UC and HC H ii regions are usually bright in the IRAC bands; thus they are often saturated in the IRAC images. G28.20 N is saturated in the 8.0 and 4.5 μm bands, G34.26 is saturated in 8.0 and 3.6 μm bands, G45.07 is saturated in all bands. G10.96 and G28.20 S are not saturated in any band. The fact that the sources are saturated does not affect the conclusions we draw in this paper.

Zoom In Zoom Out Reset image size

The bright IRAC sources suffer from the "electronic bandwidth effect."⁹ This instrumental effect is exhibited only in the 8.0 and 5.8 μm IRAC bands for saturated or bright sources; it produces artifacts appearing as a sequence of sources of decreasing intensity. This effect is seen in the 8.0 μm image of G28.20 N, and 5.8 μm and 8.0 μm images of G34.26 and G45.07 NE. As an example, the source appearing to be blended with G28.20 N slightly to the south in the 8.0 μm image of Figure 8 is not the S component, but rather is an image artifact produced by this effect.

Zoom In Zoom Out Reset image size

A discussion of the IR emission from individual sources is presented in Section 4.

The G10.96 MSFR consists of two compact radio continuum components G10.96 E and G10.96 W surrounded by more diffuse ionized gas. They have been detected in radio continuum between 1.4 GHz and 8.6 GHz (Becker et al. 1994; Walsh et al. 1998; Paper I). In Paper I, we reported the detection of a broad H92α line with an FWHM of 43.8 ± 1.5 km s⁻¹ toward the W component. The E component has a linewidth of 33.6 ± 2.8 km s⁻¹, typical for UC H ii regions. Based on the 3.6 cm and H92α observations, we identified G10.96 W as an HC ii region candidate (Paper I).

The deconvolved linewidths of the H76α and H66α lines from G10.96 W reported in Table 8 are 28.4 ± 1.5 and 28.0 ± 0.6 km s⁻¹, respectively (see Figure 1). These results indicate that the H92α line is strongly affected by pressure broadening (see the discussion in Section 5.1). For G10.96 E, both H76α and H66α lines are narrow with FWHMs of 19.4 ± 6.5 and 15.0 ± 4.9 km s⁻¹, respectively (see Figure 2). These lines are narrower than their H92α counterpart by over 10 km s⁻¹, even if the 30% uncertainty in the FWHMs of the higher frequency lines is accounted for. The difference in FWHMs over a range of RRL transitions in a relatively low density gas indicates that other line broadening mechanisms may be at work in addition to the thermal and turbulent broadening and a small contribution from pressure broadening (see Section 5.1). Most of the diffuse 3.6 cm radio continuum emission observed near component E (Paper I; HPBW ∼ 23) is resolved out in our 1.3 and 2 cm images (see Figure 2).

Several maser species have been detected toward G10.96. Methanol (CH₃OH) 6.67 GHz masers were first reported by Schutte et al. (1993) and later confirmed by Walsh et al. (1997), both based on single-dish observations. The high resolution interferometric survey of Walsh et al. (1998) revealed two CH₃OH maser spots in the W component, near the continuum emission peak (see Figure 1(b)). Water masers and CS(2-1) emission are reported by Codella et al. (1995) and Bronfman et al. (1996), respectively, based on single-dish observations.

In Figure 7 we present 3.6 and 8.0 μm images of the G10.96 MSFR. These images indicate a close correspondence between the distribution and morphology of dust and ionized gas. The diffuse dust emission from G10.96 E is filamentary, with a size of ∼60'' ×40'' (∼4.1×2.7 pc at a distance of 14 kpc; Table 2) in R.A. and decl., respectively. This filamentary structure is more prominent at 24 μm (MIPSGAL survey; Carey et al. 2009), but is also noticeable at shorter wavelengths. In the 3.6 μm image a possible central ionizing source (high-mass (proto)star) is prominent in the W component. The 3.6, 4.5, 5.8, and 8.0 μm flux densities of the IRAC source (GLIMPSE Archive ID: SSTGLMA G010.9584+00.0219) are 9.7 ± 0.9 mJy, 58.7 ± 7.5 mJy, 174.2 ± 12.1 mJy, and 395.3 ± 30.7 mJy, respectively.

The IRAC images also reveal a cluster of sources around the radio emission peak in component E (see Figure 7). Some of these objects may be ionization sources; however, we cannot rule out that some may be background objects. The multiplicity of ionizing (proto)stars may give rise to the multiple-peak morphology of the UC H ii region shown in Figure 2. The mid-IR source located near the 3.6 cm peak of G10.96 E in Figure 7 corresponds to one of the two peaks (E2 to the southwest) of the 2 cm emission (Figure 2(b)). The GLIMPSE Archive provides the 3.6, 4.5, and 5.8 μm flux densities for this source (SSTGLMA G010.9646+00.0098) of 8.2 ± 2.7 mJy, 13.1 ± 4.0 mJy, and 58.9 ± 7.6 mJy, respectively.

The G28.20−0.04 MSFR hosts two radio continuum components: N and S. Both components were detected in previous radio continuum (0.7 cm–6 cm, e.g., Kurtz et al. 1994; Walsh et al. 1998; Sewiło et al. 2004a, 2008) and RRL (H92α and H53α; Sewiło et al. 2004a, 2008) observations. The N component has been identified as the HC H ii region candidate by Sewiło et al. (2004a) based on the 3.6 cm continuum and H92α RRL observations. The H92α lines reported in Paper I have FWHMs of 74 ± 3 and 35 ± 1 km s⁻¹ for components N and S, respectively. Higher-frequency RRLs from G28.20 N (H53α, Sewiło et al. 2008; H30α, Keto et al. 2008) are significantly narrower than the H92α line, with line widths of 39.8 ± 1.7 km s⁻¹ and 20.9 ± 0.6 km s⁻¹, respectively.

In Figure 3, we present a 2 cm and a new 7 mm continuum image of G28.20, along with the corresponding H76α and H53α line profiles. Both RRLs are narrower than the H92α line reported in Paper I for both the N and S components. The H76α and H53α FWHMs are 55 ± 3 and 38 ± 1 km s⁻¹ for component N, and 24 ± 4 km and 19 ± 2 s⁻¹ for component S, respectively. The significant increase in line width with increasing quantum number for G28.20 N indicates that pressure broadening is an important (possibly dominant) line broadening mechanism for the H92α line detected from this source.

Although not evident in Figure 3, we detect marginal evidence for 7 mm continuum emission at the southeast border of G28.20 N. The emission is slightly more evident in the 6 cm CORNISH image (C. Purcell 2010, private communication). We do not report parameters for this marginal detection, but note the possible presence of a non-thermal source at this position. In addition, our continuum images—at both 7 mm and 2 cm—show evidence for a new source, with a rising spectral index, just to the east of G28.20 S. We refer to this source as G28.20 S1 (see Figure 3(a)) and report approximate values for its parameters in Table 3.

However, Sewiło et al. (2008) showed that gas dynamics also contributes to line broadening in G28.20 N. The morphology and velocity structure of G28.20 N was resolved in their high resolution (∼015) 7 mm continuum and H53α VLA observations. The source appears as a shell-like structure with an inner radius of 1100 AU and an outer radius of 2500 AU. They detected a velocity gradient of 10³ km s⁻¹ pc⁻¹ along the minor axis of the continuum source, from which they inferred a rotating torus around an ∼30 M_☉ central object. Based on line broadening and splitting along the major axis, they also suggested the presence of an outflow.

4.2.1. Molecular Gas Kinematics

4.2.2. Maser Emission

4.2.3. Dust Emission

Four radio continuum components (A–D) and two infrared sources (E and F; Campbell et al. 2000) have been identified in the G34.26 MSFR. Component C is a prototypical cometary UC H ii region with a compact head toward the east and a diffuse tail to the west (e.g., Reid & Ho 1985; Garay et al. 1986; Sewiło et al. 2004a). Components A and B, located to the east of component C, are HC H ii region candidates. A more evolved component (D), located southeast of components A–C, is thought to be an H ii shell with a diameter ∼1', expanding within an ambient density gradient (Fey et al. 1994).

We detected H53α emission toward G34.26 components A, B, and C. The line parameters are presented in Tables 7 and 8; line profiles and the 7 mm continuum image are shown in Figure 5.

The H76α (Paper I) and H53α (this paper) lines from G34.26 A are narrow, with FWHMs of ∼22 km s⁻¹, consistent with a combination of thermal and turbulent broadening. The H53α linewidth of component C is 50.2 ± 6.4 km s⁻¹, in good agreement with the H76α linewidth of 55.5 ± 1.9 km s⁻¹ reported by Garay et al. (1986). These authors demonstrate that this relatively broad linewidth probably results from microturbulent motions having a velocity of ∼30 km s⁻¹. Component B has a broad H76α line of FWHM 48.4 ± 4.4 km s⁻¹ (Paper I); however the H53α line is ∼15 km s⁻¹ narrower (see Table 8) indicating that the H76α line has a substantial contribution from pressure broadening. Also notable is the higher H53α line center velocity for component B; the velocity is more than 15 km s⁻¹ higher than for either component A or C. It is unclear if this velocity difference is important to understand the nature of component B. Hatchell et al. (2001) considered component B as a strong candidate for the driving source of the SiO outflow; if this is the case, there may be a dynamic explanation for the higher H53α velocity.

If we assume no contribution to the component A and B line widths from large scale motions and a non-thermal (turbulent) contribution of 5 km s⁻¹ (FWHM; Cesaroni et al. 1991; Plume et al. 1997; Sewiło et al. 2008), then we can determine an upper limit for pressure broadening from the H76α line. The electron temperatures for the A and B components are 5700 K and 5300 K, respectively (Paper I). Thus, the thermal plus turbulent contribution to the line width is ∼17 and ∼16 km s⁻¹ for components A and B, respectively. The resulting upper limits for the pressure broadening contribution to the H76α linewidth (calculated as described in Sewiło et al. 2008) are ∼10 km s⁻¹ and ∼43 km s⁻¹ for the A and B components, respectively.

Components A and B meet the criteria for HC H ii regions (Paper I; Avalos et al. 2006; this paper). The diameters, electron densities, and EMs derived in Paper I based on high-resolution (∼05) 2 cm observations are 0.008 pc and 0.006 pc (assuming a distance of 3.7 kpc); 1.4 × 10⁵ cm⁻³ and 2.2 × 10⁵ cm⁻³; 2.2 × 10⁸ pc cm⁻⁶ and 4.3 × 10⁸ pc cm⁻⁶, respectively. However, both components A and B have a relatively low turnover frequency of ∼ 10 GHz (Avalos et al. 2006). The physical parameters derived in this paper (see Table 5) for component A agree with the Paper I results. No physical parameters for component B can be derived because this source is partially blended with component C at the resolution of 15.

Avalos et al. (2009) resolved G34.26 A and B with 005 resolution VLA observations and detected possible limb brightening in both sources. They used spherical shell models with power-law density gradients (n_e∝r^−α) to explain their observations. They showed that the 7 mm intensity profiles, radio continuum spectra, and angular sizes of both G34.26 A and B can be reproduced by a shell of inner and outer radii ∼400 AU and ∼1000 AU, respectively, and a density gradient α ∼ 0.3–1.0. This physical picture is quite similar to what has been proposed for G24.78+0.08 A1 (Beltrán et al. 2007) and G28.20−0.04 N (Sewiło et al. 2008).

4.3.1. Maser Emission

4.3.2. Dust Emission

G45.07 consists of a bright NE component (S_2cm = 620 mJy) and a much fainter SW component (S_2cm = 40 mJy), separated by ∼6'' (see Figure 6). The SW component was first detected by Garay et al. (1986) at 2 cm and designated G45.07 A. Both NE and SW components are unresolved in our radio continuum maps. High resolution (HPBW ∼ 01) observations by Turner & Matthews (1984) of the NE component revealed a clumpy ring with a gap to the west which they interpreted as a shell with an inner cavity produced by a stellar wind from a hot central star. However, Garay et al. (1986) suggested that the ring-like structure may be the ionized inner wall of an accretion disk. They further suggest that the ring is expanding at ∼10 km s⁻¹ (owing to a hot stellar wind) and is inclined by ∼75° to the line of sight. The H76α RRL integrated over the whole NE component is broad with an FWHM of 48.1 ± 0.9 km s⁻¹ (Garay et al. 1986). Garay et al. (1985) also detected broad H66α emission with an FWHM of 42.3 ± 2.3 km s⁻¹. Keto et al. (2008) reported a detection of the H30α line toward the NE component with FWHM of 33.2 ± 4.2 km s⁻¹.

We detected H92α emission toward both the NE and SW components of G45.07; however, we were not able to determine reliable line parameters and physical properties of the sources based on these data. The H92α line observed toward the SW component is very narrow relative to the channel width (16.9 km s⁻¹) preventing a reliable Gaussian analysis. We estimate that the peak line-to-continuum ratio of this line is ≳0.07. For the NE component, the H92α line has a very poor signal-to-noise ratio, likely the result of the high optical depth which attenuates the line. The peak line-to-continuum ratio for this line is ∼0.03 and the deconvolved FWHM of 40 ± 9 km s⁻¹.

Hunter et al. (1997) reported a molecular outflow centered on G45.07 NE based on observations at millimeter and submillimeter wavelengths and in the CO(6–5) and CS(2–1) molecular lines. Infall was also detected by redshifted CS(2–1) absorption, confirming the young age of this source. The outflow axis is aligned spatially and kinematically with the H₂O maser spots in G45.07 NE (Hofner & Churchwell 1996), suggesting that the masers form in warm and dense shocked gas at the inner edges of the outflow lobes (Hunter et al. 1997).

Several masers have been detected in the NE component of G45.07, while no maser emission is associated with the SW component (see Figure 6(b)). Hofner & Churchwell (1996) detected four clusters of water masers in the area, three of which are closely associated with the centimeter continuum emission from the NE component. The fourth group is located about 2'' north of the NE component and is not coincident (in high resolution images) with any radio continuum emission. The H₂O masers are intermixed with 1665 and 1667 MHz OH masers at these two locations (Garay et al. 1985; Argon et al. 2000). A 44 GHz CH₃OH maser was detected ∼6'' north of the NE component by Kurtz et al. (2004). Single-dish observations by Menten (1991) and Caswell et al. (1995) revealed methanol 6.67 GHz masers.

4.4.1. Mid-infrared Emission

High-resolution observations (from 11 to 17; De Buizer et al. 2003, 2005; Kraemer et al. 2003) at mid-IR wavelengths from 10.5 μm to 20.6 μm revealed three mid-IR components associated with the G45.07 MSFR, designated KJK 1, 2, and 3 by Kraemer et al. (2003). The brightest component, KJK 1, is coincident with G45.07 NE and contributes roughly half of the total (KJK 1–3) flux at 12.5 μm and 20.6 μm (Kraemer et al. 2003). The second brightest component, KJK 2, lies 2'' north of KJK 1, coincident with the water maser group 2'' north of the NE component (Hofner & Churchwell 1996, see discussion above); this source is a good candidate for an HMC. Based on the mid-IR luminosities, De Buizer et al. (2003) estimated that the putative HMC in G45.07 would contain an O-type star if it is heated by a single star. KJK 2 may drive the outflow reported by Hunter et al. (1997). The third mid-IR component, KJK 3, corresponds to the SW component of G45.07.

GLIMPSE 3.6 and 4.5 μm IRAC images of the G45.07 MSFR are presented in Figure 10. At IRAC bands, G45.07 is a bright source. The KJK 1 and 2 sources are not well resolved in these images but are clearly apparent in Figure 10. KJK 3 is very faint at shorter IRAC bands, but is bright at 8 μm where PAH emission dominates.

Zoom In Zoom Out Reset image size

The GLIMPSE Catalog and Archive sources shown in Figure 10 reveal the positions of (proto)stars that may be responsible for exciting PAH emission. In general there is a good correspondence between the distribution of dust and gas in G45.07; however the dust emission extends beyond the region of ionized gas emission at 3.6 cm because the PAH contribution to this band traces the PDR that surrounds the nebula. The correspondence in projected emission does not necessarily imply that ionized gas and dust are well mixed. Additionally, the VLA filters out extended emission and much of the extended IR emission is instrumental (sidelobes of the Spitzer point-spread function), making the interpretation of the data more difficult.

We detected RRLs from all four sources in our sample (see Table 2 for details on the observed transitions). Figure 11 shows the relation between the widths (FWHMs) of the RRLs and frequency for these sources. Regions classified as HC H ii candidates and UC H ii regions are plotted separately in Figures 11(a) and (b), respectively. The dashed vertical lines indicate the rest frequencies of the observed transitions.

Zoom In Zoom Out Reset image size

The change of observed Gaussian line widths with frequency (or principal quantum number) presented in Figure 11 clearly illustrates the effects of pressure broadening based on our data. The limited sensitivity that is typical of very high spatial resolution RRL observations rarely permits the Voigt profile to be seen (Roelfsema & Goss 1992). Very broad bandwidths and good signal-to-noise ratios in the weak RRL wings are a prerequisite to fit Voigt profiles; our data are not adequate for such a fitting procedure.

Figure 11(a) shows that the lines from HC H ii regions (except G34.26 A), particularly the lower-frequency transitions, are too broad to be the result of a combination of thermal and turbulent broadening only. If an electron temperature of 10,000 K and turbulent velocity of 5 km s⁻¹ are assumed, the contribution from thermal and turbulent motions to the line widths would be only ∼22 km s⁻¹. The change in line width with transition apparent in Figure 11(a) strongly suggests that pressure broadening is a dominant broadening mechanism. Owing to the rapid change of pressure broadening with principal quantum number n (FWHM of pressure broadening ∝n^7.4; Brocklehurst & Seaton 1972), the lower-frequency lines are affected much more strongly by pressure broadening than the higher-frequency lines. At high frequencies, pressure broadening becomes negligible, thus line widths should be dominated by thermal, turbulent, and bulk motions. Even though pressure broadening depends so strongly on n, accurate adherence to the n^7.4 dependence is not expected because of other dependences on electron density and electron temperature, which vary within H ii regions (e.g., Roelfsema & Goss 1992).

Pressure broadening is less strongly dependent on electron density than on the principal quantum number (FWHM ∝ n_{e, true}, where n_{e, true} is the local true electron density; Brocklehurst & Seaton 1972). The true electron density traced by line widths is higher than the rms electron density measured from the continuum data. The difference depends on the filling factor f of the H ii region (f = n²_e,rms/n²_e,true). The filling factor is the fraction of the total volume that is occupied by dense gas; thus it is a measure of clumpiness in the H ii region. The higher electron density results in more pressure broadening, thus this effect may explain the observed increase in line width for the relatively low rms electron density UC H ii regions G10.96 E and G28.20 S (see Figure 11(b)).

Sources G28.20 N, G28.20 S, G34.26 B, and G34.26 C have been observed in high frequency RRLs: H30α for G28.20 N and H53α for the remaining three sources. By comparing these high frequency lines with lower frequency (and hence pressure-broadened) lines, we can determine n_{e, true}. Comparing these densities with the rms values from the continuum observations then provides an estimate of the filling factor. Sources showing the steepest slopes in the FWHM versus frequency plot of Figure 11 will have n_{e, true} much greater than n_{e, rms}, and hence have the smallest filling factors.

The largest change of linewidth with frequency of the RRL is obtained for G28.20 N. Sewiło et al. (2008) obtained an n_{e, true} more than an order of magnitude higher than n_{e, rms} and a filling factor of 6 × 10⁻⁴ for this source. G28.20 S, on the other hand, shows only a 4 km s⁻¹ change in linewidth between the H76α and H53α lines. We determine an n_{e, true} of (1.2  ±  0.7) × 10⁵ cm⁻³ compared to the continuum value of n_{e, rms} = 8 × 10³ cm⁻³, thus yielding a filling factor of 5 × 10⁻³.

For G34.26 components B and C we calculate true electron densities of (4.2 ± 0.9) × 10⁵ cm⁻³ and (4 ± 2) × 10⁵ cm⁻³, respectively, resulting in filling factors of 0.3 and 0.06.

We note that the true electron densities and hence the filling factors are calculated under the assumption that the H53α linewidth has no contribution from pressure broadening. If this assumption is not true, then the values we report should be considered as limits; the true electron densities will be higher and the true filling factors will be smaller.

The broad linewidths (FWHM ⩾ 40 km s⁻¹) detected toward the HC H ii candidates (Figure 11(a)) probably have a contribution from bulk motions: G28.28 N has been related to rotation and possible outflow; G34.26 B to an outflow; and G45.07 NE to expansion (see Section 4.4). All these phenomena may contribute to line broadening. For the UC H ii region G34.26 C (Figure 11(b)), Garay et al. (1986) proposed that the relatively broad H76α line results from microturbulent motions of ∼30 km s⁻¹ (see Section 4.3).

The continuum observations of MSFRs G10.96 and G28.20 at two wavelengths (Table 3) allow a determination of the spectral index for the individual sources in these regions, provided that the reliable flux density measurements can be made (G10.96 W, G28.20 N). For consistency, the two continuum images for each region (1.3 cm and 2 cm for G10.96; 7 mm and 2 cm for G28.20) were convolved to a common resolution before measuring the flux densities. For G10.96 W, the 1.3 cm and 2 cm flux densities of 296 ± 15 mJy and 288 ± 15 mJy, respectively, give a spectral index of +(0.07 ± 0.17), indicating that this source is in the optically thin regime. The 7 mm and 2 cm flux densities of G28.20 N are 723 ± 37 mJy and 490 ± 25 mJy, respectively, indicating a spectral index of +(0.4 ± 0.1). This value suggests that the Hii region has a turnover wavelength in the 2 cm–7 mm range. The flux density distribution for G28.20 N, based on high-resolution interferometric observations covering the wavelength range from 6 cm to 1.3 mm (Table 9), is shown in Figure 12. The plot shows the spectrum becoming optically thin in the 20–50 GHz range.

Zoom In Zoom Out Reset image size

We found that G28.20 N is very similar to the well-studied HC H ii region G24.78+0.08 A1 (Beltrán et al. 2007) in terms of physical properties, kinematics (rotation, infall, outflow), morphology (shell-like), and association with masers. The flux density distributions of these sources are similar in terms of the shape and the turn-over frequency (Beltrán et al. 2007; Figure 12). Neither of these HC H ii regions has a rising spectrum to millimeter wavelengths (as do G9.62+0.19 E and G75.78+0.34 (H₂O); see Franco et al. 2000). The shell-like morphology appears to be a common feature of the HC H ii regions. Besides G28.20 N and G24.78+0.08 A1, the A and B components in G34.26 and also G45.07 NE, show a shell morphology in high resolution images reported by Avalos et al. (2009) and Turner & Matthews (1984), respectively.

We investigated the distribution of maser emission in MSFRs from our sample. We found that maser emission is only detected toward sources with broad RRLs (typically the most compact and dense sources in the UC/HC H ii region complexes): G10.96 W, G28.20 N, G34.26 B and C, and G45.07 NE.

In Figure 4 we show the maser positions for G28.20 N overlaid on the continuum and thermal molecular line emission. The OH masers are distributed linearly over a region ∼14 (or ∼0.04 pc; Figure 4(a)) with a velocity gradient of ∼150 km s⁻¹ pc⁻¹. The velocities increase from ∼94 km s⁻¹ at the northwestern end of the linear feature to ∼100 km s⁻¹ at the southeastern end. The small cluster of maser spots north of the southeastern end has a velocity of ∼95 km s⁻¹. The 1.3 cm image (Figure 4(c)), with 20× higher angular resolution than the 7 mm image (Figure 4(a)), shows the OH masers situated along the southwestern boundary of the continuum source, at the edge of the CH₃CN emission (Figure 4(d)). The location of the OH masers suggests that they arise within a molecular shell surrounding the ionized gas.

The OH maser velocities in G28.20 N roughly agree with the CH₃CN velocity of ∼95–96 km s⁻¹; however, they are redshifted relative to the H ii region velocity of ∼93 km s⁻¹ estimated by the systemic velocity of the highest frequency RRL observed toward G28.20 N (H30α at 1.3 mm; Keto et al. 2008; Welch & Marr 1987). A similar redshift is seen for CH₃OH masers, detected over the velocity range 94–105 km s⁻¹. These redshifts for OH and CH₃OH masers, together with the fact that they appear projected against the optically thick continuum, suggest that they arise on the near side of G28.20 N, thus supporting the scenario that the masers are located in a collapsing molecular envelope surrounding the H ii region. A similar scenario was proposed for W3(OH) (Reid et al. 1980) and several other UC H ii regions (including G34.26 C and G45.07 NE; Garay et al. 1985) based on the maser and radio continuum emission in those regions. This interpretation can also apply to G10.96 W from our sample, where the two CH₃OH maser spots are near the continuum emission peak and have velocities (∼25 km s⁻¹; Walsh et al. 1998) redshifted by ∼14 km s⁻¹ relative to the H66α velocity (see Table 7).

The OH maser velocities in G28.20 N appear to be inconsistent with the velocity gradient reported for the H53α line by Sewiło et al. (2008). In particular, the H53α velocity gradient, when extrapolated to the position of the arc of maser emission, gives velocities of 115–120 km s⁻¹, significantly higher than the OH maser velocities of 93–101 km s⁻¹. The differing direction and magnitude of the H53α and the OH velocity gradients suggests that dynamics of the ionized and molecular gas trace different phenomena.

The orientation of the H₂O masers (SE–NW) in G28.20 N is consistent with the outflow suggested by Sewiło et al. (2008) and Qin et al. (2008). However, the velocity structure of the masers is ambiguous and does not provide strong evidence for the outflow.

In G45.07 NE, the H₂O maser spots are aligned spatially and kinematically with the outflow axis detected in CO(6–5) and CS(2–1) molecular lines (Hofner & Churchwell 1996; see Section 4.4), suggesting that the masers form in warm and dense shocked gas at the inner edges of the outflow lobes (Hunter et al. 1997).

We find that all the radio components have point-source counterparts in the GLIMPSE Archive except G34.26 A and B, and G45.07 SW. A faint mid-IR source was detected at the position of G45.07 SW; however, it was not included in the GLIMPSE catalog due to the close proximity of a much brighter source (G45.07 NE) that prevented an accurate photometric measurement. G34.26 A does not have a clear counterpart in the IRAC image; this source was detected in the mid-IR by earlier studies, unlike component B which is undetected in the mid-IR.

In Table 10, we list GLIMPSE counterparts to the radio sources (the closest neighbors). The distance between the mid-IR and radio positions ranges from 02 for G45.07 NE to 24 for G28.20 S (see Table 10). The proximity of the mid-IR sources to the radio peaks of HC H ii regions suggests that these sources probably are the ionizing (proto)stars.

The relatively low density of stars nearby to G10.96 E and G28.20 relative to more distant parts of these fields suggests extinction of background stars in these regions. The majority of IR sources shown in Figures 7–10 are probably background stars; however, we expect to find a number of young stellar objects (YSOs) in these MSFRs. The YSOs can be identified by their mid-IR colors and by fitting their spectral energy distributions with YSO models. This would be particularly interesting for G28.20, which is located in an IRDC (see Figure 8).

The Spitzer images of the sources from our sample suggest that generally the large-scale spatial distribution of gas and dust in these regions are similar; G34.26 C being the principal exception in our sample. Higher resolution infrared observations would be worthwhile, however, to separate diffuse emission from the mid-IR emission from outflows and disks (e.g., NGC 7538 IRS 1; De Buizer & Minier 2005).