NEW RADIO OBSERVATIONS OF ANOMALOUS MICROWAVE EMISSION IN THE H ii REGION RCW175

H ii regions play a crucial role in broadening our understanding of the mechanisms regulating the interstellar medium (ISM). The complex morphology of H ii regions is the result of an ensemble of factors, i.e., the evolutionary stage, the number and spectral type of the ionizing sources, and their interplay with the surrounding medium. RCW175 is a relatively well-studied, complex H ii region, and is approximately 10 × 15 arcmin² in size. The region was identified by Sharpless (1959) and has been observed at various wavelengths ever since. In the microwave range it was first observed by the Very Small Array (VSA) experiment (Todorovic et al. 2010), which found evidence of an anomalous microwave emission (AME) excess, and then by the Cosmic Background Imager (CBI) experiment, which confirmed the anomalous excess and a basic structure of at least two components at ∼30 GHz (Dickinson et al. 2009).

AME was first identified as diffuse emission in the 10–60 GHz range in excess of the expected level of microwave emission from free–free, synchrotron, and thermal dust emission by experiments such as the Cosmic Background Explorer-DMR (Kogut et al. 1996), the Saskatoon experiment (de Oliveira-Costa et al. 1997), Owens Valley Radio Observatory (Leitch et al. 1997), and the Tenerife experiment (de Oliveira-Costa et al. 1999). Several authors have found further statistical evidence of AME (see, e.g., Banday et al. 2003; Lagache 2003; de Oliveira-Costa et al. 2004; Finkbeiner et al. 2004; Davies et al. 2006; Hildebrandt et al. 2007; Lu et al. 2012; Miville-Deschênes et al. 2008), while it has also been detected directly in individual regions in a limited number of cases (see, e.g., Watson et al. 2005; Scaife et al. 2007; Casassus et al. 2008; AMI Consortium 2009; Murphy et al. 2010; Dickinson et al. 2010; Genova-Santos et al. 2011; Planck Collaboration 2011a, 2014a). Recent models predict that the AME is dominated by electric dipole emission from the smallest grains, possibly polycyclic aromatic hydrocarbons (PAHs; Draine & Lazarian 1998; Ysard et al. 2010; Hoang et al. 2010). However, other physical emission mechanisms, such as hot free–free emission (Leitch et al. 1997), hard synchrotron radiation (Bennett et al. 2003), or magnetic dipole emission (Draine & Lazarian 1999), cannot be completely ruled out.

To date there have been only a handful of polarization observations of individual AME regions, all of which show evidence of little or no polarization: to the level of a few percent (Rubiño-Martin et al. 2012; Battistelli et al. 2006; Mason et al. 2009; Dickinson et al. 2007; Casassus et al. 2008; Hoang et al. 2013; Lopez-Caraballo et al. 2011; Dickinson et al. 2011). AME polarization is able to differentiate between the different models since different models predict a different polarization percentage. This is the case, for instance, of magnetic dipole emission, which predicts a polarization fraction that can be of the order of 10% or more with frequency-dependent behavior. In contrast, Stokes I multi-band observations in the 5–30 GHz range are essential to separate the AME from the other types of ISM emission in conjunction with millimeter observations for thermal dust removal. High angular resolution measurements are fundamental to understanding the physics behind the AME and to limit source confusion (see, e.g., Battistelli et al. 2012). In this respect, though, it is remarkable that, up until now, no observation has had sufficient sensitivity, resolution, or frequency coverage to disentangle fully the numerous candidate mechanisms. Arcminute-level angular resolution observations are crucial to disentangling different contributions within the same Galactic region. These are starting to reveal cases in which the dust-to-radio correlation is observed to decrease when we go to finer angular scales (see, e.g., Vidal et al. 2011; Castellanos et al. 2011; Tibbs et al. 2013). Ysard et al. (2011) and others have suggested that this behavior could be due to the fact that whereas the infrared (IR) dust emission is proportional to the intensity of the radiation field and to the dust column density, it is not as straightforward for spinning dust emission since its excitation processes are more numerous and complex. Indeed, the rotation of interstellar grains is both excited and damped by the emission of IR photons (proportional to the radiation field intensity), collisions with neutrals (H i, H₂) and ions (H ii, C ii), plasma drag (H ii, C ii), photoelectric emission, formation of H₂ on the grain surface, and the emission of electric dipole radiation itself (Draine & Lazarian 1998).

A detailed multi-wavelength study of RCW175 has been performed by Tibbs et al. (2012) based on publicly available data. Tibbs et al. (2012) computed the star formation rate of the region, identified the position of young stellar object candidates, and estimated the total dust mass of RCW175. By combining high angular resolution Spitzer (Werner et al. 2004) and Herschel (Pilbratt et al. 2010) IR data with the dust model DustEM (Compiêgne et al. 2011), Tibbs et al. (2012) characterized the dust properties within the region. They cross-correlated the coarse angular resolution microwave data (e.g., CBI) with the derived dust property maps, and found that the AME is not strongly correlated with the smallest dust grain abundance but rather with the interstellar radiation field within the region. A missing key piece of information in the available data set are observations in the range 8–25 GHz, where the AME is expected to rise. These are essential for a detailed study of the physical characteristics of the AME, as well as for separating the various types of emission.

In this paper we present high angular resolution, multi-frequency, and polarization-sensitive observations of this source at 8.4, 13.5, and 21.5 GHz conducted with the Parkes Radio Telescope. These observations cover the gap in the spectral energy density (SED) of currently available data and can increase our understanding of the morphology of RCW175 at frequencies where the AME is a significant fraction of the total emission. This paper is structured as follows: in Section 2 we describe the observations we have conducted with the Parkes telescope; in Section 3 we discuss the SED derived from our new Parkes observations in addition to other data available in literature. In Section 4 we discuss the morphology of the region, including the spatial distribution of the AME; in Section 5 we present the results obtained from cross-correlating Herschel and Spitzer IR maps with microwave maps, while in Section 6 we present the results obtained from our polarization observations. In Section 7 we discuss the interpretation of the results and present our conclusions.

The observations were conducted at three different frequency bands with the Parkes Radio Telescope, NSW Australia, a 64 m telescope operated as a National Facility by ATNF-CASS, a division of CSIRO. Photometric observations at 8.4 GHz and 13.5 GHz are described in Sections 2.1 and 2.2, respectively, while the polarimetric observations at 21.5 GHz are described in Section 2.3.

2.1. 8.4 GHz Observations

The 8.4 GHz observations were conducted with the MARS receiver of the Parkes telescope on 2011 July 26 for 4 hr. The receiver is a circular polarization system with T_sys  ∼  30 K, a resolution of FWHM = 2.4 arcmin, and a bandwidth of 400 MHz centered at a frequency of 8.4 GHz. To detect the whole useful bandwidth, the backend Digital Filter Banks Mark 3 (DFB3) was used with a configuration bandwidth of 1024 MHz and 512 frequency channels each of width 2 MHz.

The correlator has full Stokes parameter capability, recording the two autocorrelation products RR*, LL* and the complex cross-product of the two circular polarizations RL* whose Real and Imaginary parts are the two Stokes parameters Q and U. The feed illuminates the primary mirror with an edge taper of 19 dB, the first side lobe amplitude is −30 dB relative to the antenna pattern peak, and the gain is 1.18 Jy K⁻¹.

The flux density scale was calibrated using the source PKS B1934-638 with an accuracy of 5% (Reynolds 1994). This source is compact compared to the beam size, and in the case of slightly extended sources, as for RCW175 at this resolution, some flux can be picked up by the first side lobes resulting in an error of the flux scale accuracy. To check this, we have compared the solid angle of the main beam and first sidelobes. Our calibration observations are not sensitive enough to measure the sidelobes, but their shape can be estimated in the case of a mirror illuminated by a Gaussian feed⁹ as a function of its edge taper and the central blockage of the focus cabin (Goldsmith 1987). Following Goldsmith (1987), we find that the first sidelobe stretches out to 11 arcmin (diameter), which covers most of our source in one dimension and exceeds it in the other, and its solid angle is 0.6% of the main beam. The effects on the flux accuracy are thus negligible. This confirms the usual assumption that it is safe to use a flux calibration performed using compact sources for objects extending up to a few beam-widths. The main beam efficiency is 92%, consistent with the results from Goldsmith (1987) for similar edge taper values.

The channels spanning the IF were then binned into twenty 20 MHz sub-bands for the subsequent map-making processing. A standard basket-weaving technique with orthogonal scan sets along R.A. and decl. spaced by 45 arcsec was used to observe an area of 30' × 30' centered on the source. The scan speed was 05 minute⁻¹ with a sampling time of 0.25 s. Map-making software based on the Emerson et al. (1988) Fourier algorithm was applied to make the maps (Carretti et al. 2010). This technique effectively reduces 1/f noise, and removes stripes and features different in the two orthogonal sets of scans.

The 20 sub-band maps were then binned together into one map for the analysis (see Figure 1). The final rms on the map is 22 mJy beam⁻¹ on beam-size scales. This is higher than the expected sensitivity (some 0.7 mJy beam⁻¹). However, RCW175 is close to the Galactic plane at (l, b) = (291, −07), and such an excess rms signal might be due to diffuse Galactic emission. To investigate this option, we have used the map at 1.4 GHz and resolution of 14.5 arcmin of CHIPASS (Calabretta et al. 2014), where we find that an rms signal of rms$_{1.4\ \rm GHz}^{14\prime 5} =700$ mK in the area around RCW175. The behavior of the rms signal with the angular scale is described by rms $\propto \sqrt{\ell (\ell +1)C_\ell }$, where C_ℓ is the angular power spectrum and the multipole ℓ relates to the angular scale θ as θ  ∼  180°/ℓ. The angular power spectrum follows a power-law C_ℓ ∝ ℓ^−1.7 near the Galactic plane (Bruscoli et al. 2002). In the 1.4–8.4 GHz range the free–free emission is the leading contribution close to the Galactic plane with a brightness temperature that goes as $T_b^{\rm ff}\,{\propto}\,\nu ^{-2.1}$ as a function of the frequency ν. Rescaling the CHIPASS rms for the frequency and angular spectrum previously described, and converting the temperature rms for the calculated gain, we estimate a Galactic rms signal of rms$_{8.4\ \rm GHz}^{2\prime 4} =25$ mJy beam⁻¹ at the frequency and resolution of our observations, consistent with the rms signal we detect around the source.

Zoom In Zoom Out Reset image size

2.2. 13.5 GHz Observations

The 13.5 GHz observations were conducted with the Ku-band receiver of the Parkes telescope on 2011 August 31 and 2011 September 1 for a total of 6 hr. The receiver is a linear polarization system with T_sys  ∼  150 K, a resolution of FWHM = 1.7 arcmin, and a bandwidth of 700 MHz centered at a frequency of 13.55 GHz. To detect the whole useful bandwidth, the backend Digital Filter Banks Mark 3 (DFB3) was used with a configuration bandwidth of 1024 MHz and 512 frequency channels (2 MHz each). Only the two autocorrelation products were used (XX*, YY*) for Stokes I measurements. The feed illuminates the primary mirror with an edge taper of 18 dB, the first sidelobe amplitude is −27 dB relative to the antenna pattern peak, and the gain is 1.55 Jy K⁻¹.

The flux density scale was calibrated using the source PKS B1934-638 with an accuracy of 5% (Reynolds 1994). As for the 8.4 GHz data set, we could not measure the secondary lobes, and to estimate the error in the flux scale accuracy for our extended source, we estimated the antenna pattern following Goldsmith (1987). In this case, to cover most of the source, the first two sidelobes have to be considered that stretch out to 13 arcmin (diameter). We found that their combined solid angle is 1.9% of the main beam with marginal effects on the flux scale accuracy. The main beam efficiency is 90%.

The flux calibration also accounts for the atmospheric opacity attenuation. The opacity, when observing the calibrator, was 0.085 dB at the observing Elevation (EL); that way the flux calibration applies a constant correction for 0.085 dB attenuation. During the observations the opacity ranged from 0.064 to 0.120 dB (including EL effects) for a maximum variation compared to the constant opacity correction of 0.035 dB (0.8%), with marginal effects on the flux density scale accuracy. The opacity at the observing EL was computed from the zenithal opacity correcting for EL effects (1/cos(EL)), while zenithal opacity was computed from atmospheric parameters (temperature, pressure, and relative humidity).

The channels spanning the IF were then binned into seven 100 MHz sub-bands for the subsequent map-making processing (see Figure 2). A standard basket-weaving technique with orthogonal scan sets along R.A. and decl., spaced by 30 arcsec, was used to observe an area of 30' × 30' centered at the source. The scan speed and sampling time were the same as those for the 8.4 GHz observations. The same map-making software was applied to make the maps, and finally the seven sub-band maps were binned together into one. The final rms on the map is 18 mJy beam⁻¹ on beam-size scales, larger than the expected sensitivity (4 mJy beam⁻¹). Following the procedure described in Section 2.1, we estimate the Galactic emission contribution at rms$_{13.5\ \rm GHz}^{1\prime 7} =13$ mJy beam⁻¹ at the frequency and resolution of the Ku-band observations, consistent with the observed rms.

Zoom In Zoom Out Reset image size

2.3. 21.5 GHz Observations

3.1. Radio and Microwave Ancillary Data

Publicly available ancillary data have been analyzed and combined with our Parkes observations to derive the SED of RCW175. At 1.4 GHz we have considered the NRAO VLA Sky Survey (NVSS; Condon et al. 1998) data, characterized by a relatively high angular resolution (FWHM = 0.75 arcmin), although they lack sensitivity on scales larger than 15 arcmin. Effelsberg 1.4 GHz data by Reich et al. (1990a) are also available,¹⁰ but these need to be re-analyzed using the Stockert 1.4 GHz data (Reich 1982) to add missing large-scale emission to the Effelsberg data. This results, for the present analysis, in an underestimation of the integrated flux estimate form the source, and thus it has not been included in our analysis. We have thus analyzed and used, to constrain the integrated 1.4 GHz emission, the Green Bank 300 ft data by Altenhoff et al. (1970), which are affected by a larger uncertainty but better match our angular resolution and map size (FWHM ≈ 10 arcmin). Data from the Effelsberg 2.7 GHz (Reich et al. 1990b) and Parkes 5 GHz (Haynes et al. 1978) surveys match the angular resolution needed for the present analysis, and thus we have included the flux density estimated by Tibbs et al. (2012) for these surveys.

In addition to these data, we performed an integrated flux density analysis on the GBT data (8.35 GHz and 14.5 GHz) by Langston et al. (2000), as well as the Nobeyama 10 GHz data by Handa et al. (1987), which have been considered in the analysis of the integrated flux density for the RCW175 H ii region. These data, however, are affected by significant systematics on angular scales that are important for our study. For instance, the Nobeyama data present a non-constant offset that is probably due to an inaccurate removal of the Galactic plane emission. For this reason, they are not included in our analysis. At microwave frequencies, as well as for our new Parkes data, we have also used the CBI 31 GHz (Dickinson et al. 2009) and VSA 33 GHz (Todorovic et al. 2010) flux density estimates. A summary of the data set used in this work is provided in Table 1.

Table 1. Integrated Fluxes and Instrumental Properties Used for the RCW175 Analysis and Spectral Energy Density Fit

Frequency	Telescope/	Ang. Res.	Flux Density	Calibration	Reference
(GHz)	Experiment	FWHM	(Jy)	Uncertainty
1.4	Green Bank 300 ft	94×104	6.0 ± 1.8	Included	Altenhoff et al. (1970)
2.7	Effelsberb 100 m	43	5.7 ± 0.9	Included	Reich et al. (1990b)
5.0	Parkes 64 m	41	4.0 ± 0.8	Included	Haynes et al. (1978)
8.4	Parkes 64 m	24	4.86 ± 0.51	5%	This work
13.5	Parkes 64 m	17	5.71 ± 0.62	5%	This work
31	CBI	43×40	5.97 ± 0.30	1.3%	Dickinson et al. (2009)
33	VSA	131×100	7.8 ± 1.6	≲ 2%	Todorovic et al. (2010)
100	Planck HFI	959	3.2 ± 2.7	0.5%	Planck Collaboration (2011b)
143	Planck HFI	718	3.1 ± 3.5	0.5%	Planck Collaboration (2011b)
217	Planck HFI	481	23 ± 22	0.5%	Planck Collaboration (2011b)
353	Planck HFI	470	81 ± 92	1.2%	Planck Collaboration (2011b)
600	Herschel SPIRE-500	363	650 ± 130	7%	SPIRE Obs. Man. (2011)
857	Herschel SPIRE-350	249	1580 ± 320	7%	SPIRE Obs. Man. (2011)
1200	Herschel SPIRE-250	182	4110 ± 830	7%	SPIRE Obs. Man. (2011)
1875	Herschel PACS-160	1035	9660 ± 1940	7%	Balog et al. (2014)
3000	IRAS IRIS-100	43	10300 ± 1800	13.5%	Miville-Deschênes et al. (2005)
4286	Herschel PACS-70	80	9600 ± 1900	7%	Balog et al. (2014)
5000	IRAS IRIS-60	40	6740 ± 1050	10.4%	Miville-Deschênes et al. (2005)
12670	Spitzer MIPS-24	60		10%	Carey et al. (2009)

Notes. The fluxes have been estimated through aperture photometry of the available data (see text for details). Uncertainties are obtained by quadrature summation of the statistical and systematic/calibration uncertainties.

Download table as:  ASCIITypeset image

3.2. Millimetric, Sub-millimetric, and IR Ancillary Data

3.3. Integrated Emission Spectrum and Fit

The integrated emission from the RCW175 H ii region, presented in Figure 3, was obtained using aperture photometry with an annular radius of 12 arcmin and an estimate of background in the annular region from the 12–17 arcmin range. Uncertainties were estimated from the map fluctuations outside the source. This procedure allows us to neglect structures larger than ∼15 arcmin. Using the data set described above, we have fitted a model that includes the four emission mechanisms that we have assumed to be dominant in the microwave range for RCW175: synchrotron, free–free, spinning dust, and thermal dust emission.

Zoom In Zoom Out Reset image size

The thermal dust was fitted to a minimum wavelength of 60 μm, using the IR, sub-mm, and mm data and a sum of two modified blackbodies, S_ν ∝ ν^βBB(ν, T_dust), as in Tibbs et al. (2012). In this fit, upper limits or tentative detections derived in the Planck 143, 217, and 353 GHz bands (Planck Collaboration 2011b) have also been included. The Planck 100 GHz data point was instead included in the "low frequency" fit, in order to account for the impact of the thermal dust Rayleigh–Jeans tail. Planck data at frequencies above 353 GHz were not considered, as we rely on the higher spatial resolution SPIRE data. From the fit, we retrieve results consistent with Tibbs et al. (2012), i.e., a cold dust population with temperature $T_{\rm c}=24.3^{+2.4}_{-3.2}$ K, and a warm dust population with temperature $T_{\rm w}=50.8^{+1.5}_{-1.4}$ K, having assumed a dust spectral emissivity index β = 2 for both populations. The interstellar radiation field (ISRF) plays an important role in heating and exciting the smallest dust grains and PAHs. This can be quantified though the parameter χ_ERF, which can be estimated as χ_ERF = (T_c/17.5K)^{(4 + β)}. We find χ_ERF = 7, which is indicative of a radiation field intensity a few times the ISRF in the solar neighborhood (Mathis et al. 1983; Ali-Haimoud et al. 2009; Planck Collaboration 2014a). The dust temperatures are slightly higher with respect to most of the AME sources, which are found to be associated mostly with colder regions (Planck Collaboration 2014a); however, they follow the same trend as found by Tibbs et al. (2011) for the Perseus complex, where the strength of the ISRF was found to be of key importance for AME regions.

The low-frequency data (<8 GHz) were initially fitted with a single power law (i.e., S_ν ∝ ν^α), which resulted in an average spectral index of $\alpha =-0.28^{+0.09}_{-0.06}$. Later, a fit was performed by accounting for a free–free component plus a synchrotron component, with intensity and spectral index left free to vary ($S_{\nu }\,{\propto}\, A_{\rm ff}\nu ^{\alpha _{\rm ff}} + A_{\rm sy}\nu ^{\alpha _{\rm sy}}$). The results show that, below 8 GHz, the emission is dominated by free–free emission with spectral index α_ff = −0.18 ± 0.07, with an additional sub-dominant contribution due to a steeper power-law with spectral index $\alpha _{\rm sy}=-0.7^{+0.5}_{-0.2}$, consistent with synchrotron emission. At 1 GHz, the fraction of free–free emission with respect to synchrotron emission is A_ff/A_sy = 1.8, clearly increasing with frequency and indeed confirming the hypothesis of synchrotron contamination along the line of sight of RCW175 as previously found by Tibbs et al. (2012). To fit the spinning dust component, we used the spdust.2.01¹¹ code (Ali-Haimoud et al. 2009; Silsbee et al. 2011) to compute different emission spectra accounting for the radiation field intensity and gas temperature of this region. The hydrogen number density and gas temperature have been chosen to be consistent with standard values associated with molecular clouds (MC) and the warm ionized medium (WIM; Draine & Lazarian 1998; Dickinson 2012). For the WIM component we have assumed hydrogen number densities $n_{\rm H}^{\rm WIM}=26.3$/cm³ derived from the 5 GHz integrated flux of RCW175 and assuming an electron temperature, T_e = 5800 K, following the analysis presented in Paladini et al. (2004) and Tibbs et al. (2012). As for the MC component, we built a grid of spectra with $n_{\rm H}^{\rm MC}$ varying from $n_{\rm H}^{\rm MC}=100$/cm³ to $n_{\rm H}^{\rm MC}= 1000$/cm³, with a stepsize of 25/cm³, and performed a best fit over $n_{\rm H}^{\rm MC}$. The best fit was obtained for $n_{\rm H}^{\rm MC}=150$/cm³, corresponding to a minimum χ²/dof = 0.78. The major ion fractions, [H+] and [C+], are estimated as described in Ysard et al. (2011). Fits have been performed using MPFIT (Markwardt 2009) with physically reasonable priors, such as ensuring non-negative values. Attention has also been paid to ensure that the fitted parameters avoid hitting the prior limits. The parameter uncertainties were determined through Monte Carlo analysis by randomizing the data points by their 1σ uncertainty and evaluating the scatter as the final uncertainty. In Table 2 we collect the parameters resulting from the aforementioned fits.

Table 2. Best-fit Parameters Obtained form the SED Fit of the Integrated Emission of the RCW175 H ii Region

Parameter	Value	Notes
Tc	$24.3^{+2.4}_{-3.2}\, {\rm K}$	Sum of two modified black-bodies
Tw	$50.8^{+1.7}_{-1.8}\, {\rm K}$	Sum of two modified black-bodies
β	2	Fixed
χERF	7	Derived from Tc
α	$-0.28^{+0.09}_{-0.06}$	Single power-law
Aff	$4.8^{+0.9}_{-1.0}\, {\rm Jy}$	Sum of two power-laws
αff	−0.18 ± 0.07	Sum of two power-laws
Asy	$2.5^{+0.7}_{-0.7}\, {\rm Jy}$	Sum of two power-laws
αsy	$-0.7^{+0.5}_{-0.2}$	Sum of two power-laws
Te	5800 K	Derived from 5 GHz integrated flux
$n_{{\rm H}}^{{\rm WIM}}$	26.3/cm3	Derived from 5 GHz integrated flux
$N_{{\rm H}}^{{\rm WIM}}$	$(0.057^{+0.004}_{-0.022})\times 10^{22} H/{\rm cm}^{2}$
$n_{{\rm H}}^{{\rm MC}}$	150/cm3	Range =[100/cm3–1000/cm3]; step =25/cm3
$N_{{\rm H}}^{{\rm MC}}$	$(3.8^{+0.4}_{-0.8})\times 10^{22} H/{\rm cm}^{2}$

Notes. The two dust temperatures, T_c and T_w, were obtained from the fit of the sum of two modified blackbodies. The dust emissivity β has been taken equal to 2. The ISRF has been parameterized through χ_ERF. The low frequency fit was either performed through a single power law with spectral index α, or through the sum of two power laws of spectral indices α_ff and α_sy, and normalization A_ff and A_sy, respectively. The spinning dust models are constrained with the hydrogen number densities n_H and the column density N_H for an MC and a WIM component.

Download table as:  ASCIITypeset image

3.4. SED Interpretation

The morphology of the RCW175 H ii region has been studied by Dickinson et al. (2009) and others, who have identified two sub-regions within RCW175 (G29.0−0.6 and G29.1−0.7) and built the spectrum of the whole region, as well as the spectrum of its brightest constituent (G29.0−0.6). Tibbs et al. (2012) performed a detailed multi-wavelength study of RCW175, and described G29.0−0.6 as the brighter, more dusty component, with G29.1−0.7 being described as the more diffuse, more evolved component. So far, the only high angular resolution (that is better than 2 arcmin) maps of RCW175 have been the 1.4 GHz NVSS maps (i.e., NVSS data at 1.4 GHz by Condon et al. 1998) at radio frequencies and, in the infrared, the Spitzer and Herschel maps. Our new microwave maps allow a more detailed study of the different spatial components of this H ii region also in this frequency domain. In our maps, we can clearly identify the brighter and unresolved G29.0−0.6 region toward the west. In addition, we resolve the more diffuse G29.1 region into two separate components, i.e., G29.1−0.7 toward the east, and G29.1−0.6 toward the north (see Figure 4). A similar structure can be seen in the 1.4 GHz NVSS data (Condon et al. 1998) and in the 10 GHz Nobeyama map (Handa et al. 1987). As mentioned, with both of these data sets, we could not extract information about the larger angular scales within the region. Despite the limitations, we could probe the self-similarity of the spectral behavior of each region.

Zoom In Zoom Out Reset image size

Relying on the quality and on the good (or common) systematic effect control of the Parkes maps, as well as on the self-similarity of the SED of the different constituents of the region studied, we have attempted to perform a simple component separation analysis.

To do this, we first smoothed the 13.5 GHz map to the same angular resolution as the 8.4 GHz map (i.e., FWHM = 2.4 arcmin). We then extrapolated the 8.4 GHz map to 13.5 GHz, assuming an average spectral index power law of α = −0.28 (as derived in Section 3), and subtracted it from the actual 13.5 GHz map, producing a map of the AME at 13.5 GHz (see Figure 5). For this analysis we have assumed that the 8.4 GHz map is not contaminated by AME, resulting in a bias in the extracted map, which should thus be treated as an underestimate of the AME in the region. The difference map clearly reveals residual emission that is mainly concentrated in the compact component, G29.0−0.6, plus an additional diffuse component surrounding the G29.1 complex. In Figure 5 we use a dashed line to illustrate the level corresponding to 1σ uncertainty. The internal part thus exceeds the 1σ uncertainty where we have included a 5% error arising from the modeling uncertainty. We have also included in the uncertainty the propagated error arising from the fit affecting the estimation of the flux. We emphasize that this is one of the first attempts to perform a detailed component separation at microwave wavelengths on a single galactic source of AME. This approach will likely represent a pathfinder for detailed studies of the physics of the individual ISM components.

Zoom In Zoom Out Reset image size

A cross-correlation analysis between radio-frequency and IR data can improve our understanding of the physical components of the region. In Figures 6 and 7 we show the correlation analysis we performed using the Ku-band 13.5 GHz data and the SPIRE, PACS, and MIPS IR data, which span the wavelength range from 500 μm to 24 μm. To this end, we have convolved all the maps to the 13.5 GHz data angular resolution. On the left, we show the IR maps and, as contours, the map at 13.5 GHz; on the right, we provide the corresponding scatter plot, i.e., the IR map versus the microwave map, where different regions in the map are marked with different colors, following the same color scheme defined in Figure 4. Red crosses are used to denote the compact and bright G29.0−0.6 region, blue crosses identify the G29.1−0.6 central region, and black crosses refer to G29.1−0.7.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Noteworthy, the combination of sub-mm and IR bands is typically used to trace the three major populations of dust grains found in the ISM. In the reference framework of dust emission in the diffuse ISM (T_d ∼ 17 K), the SPIRE (500 μm, 350 μm, and 250 μm) and PACS 160 μm bands trace thermal dust emission from Big Grains, while the PACS 70 μm and MIPS 24 μm bands trace emission from very small grains (VSGs). This picture still holds (apart from the 24 μm emission that we discuss below) in the case of H ii regions.

From the 13.5 GHz versus SPIRE-500 μm correlation, we note an absence of correlation for G29.1−0.6 and G29.1−0.7, and a mild correlation, although with a large scatter, for G29.0−0.6. A more detailed analysis reveals a shell-like structure from which the IR emission probably originates, while the microwave emission appears to generate from the bulk of G29.0−06. A similar behavior is highlighted by the correlation with the 350 μm, 250 μm, 160 μm maps, and especially with the 70 μm map.

The 13.5 GHz versus 24 μm data correlation is, at first glance, more pronounced. However, this behavior is not straightforward to interpret. As discussed by Paladini et al. (2012), the 24 μm emission arising from the interior of H ii regions, contrary to what happens in the diffuse ISM, is not necessarily associated with VSG emission, but rather with BG replenishment, as proposed by Everett et al. (2010) for the case of wind-blown bubbles. Therefore, for 24 μm, the correlation in Figure 7 is not a priori an indication that the AME carriers are small dust grains.

In addition to the IR maps discussed above, we incorporated the dust parameter maps derived by Tibbs et al. (2012), using the DustEM dust model (Compiêgne et al. 2011), in the correlation analysis. These parameter maps include important physical quantities such as: (1) Y_VSG, the abundance of VSGs relative to BGs; (2) Y_PAH, the abundance of PAHs relative to BGs; (3) χ_ERF, the strength of the exciting radiation field, as parameterized in Tibbs et al. (2012); (4) N_H, the hydrogen column density; and (5) T_EQ, the dust equilibrium temperature derived as the median temperature of the BGs.

In Figures 8 and 9, we compare the RCW175 13.5 GHz map with the parameter maps derived from DustEM: on the left, we provide the parameter maps and, as contours, the microwave map, with all the maps at a common angular resolution; on the right, we show the corresponding scatter plot, i.e., 13.5 GHz data versus parameter map, where, once again, different colors denote different regions of the map.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We have calculated the Pearson correlation coefficients, ρ, for the 13.5 GHz map and the various IR or DustEM parameter maps, together with the probability to exceed (PTE) that correlation value in the case of null correlation. In doing so, we have assumed that the variable $t= \rho \sqrt{(}(n-2)/(1-\rho)^2)$ follows a t-Student distribution, with n − 2 dof, n being the number of data points.¹² The result of this correlation measure is reported in Table 3 for the correlations between the microwave versus IR maps, and in Table 4 for the correlations between the microwave versus DustEM parameter maps. When PTE <0.002, we also provide the slope of the linear fit to the microwave map versus parameter map. Microwave versus IR dust emissivities, in terms of microwave brightness temperature relative to 100 μm brightness, are sometimes used to highlight common behavior of AME regions (see, e.g., Dickinson 2012). Comparison with our correlations is not straightforward although we can see a similar qualitative behavior.

For the DustEM analysis, both the plots and Pearson coefficients reveal a clear correlation between the microwave emission in G29.0−0.6 and the dust parameters (Y_VSG, Y_PAH, X_ERF, and T_EQ). The presence of highly embedded ultracompact H ii regions, which could induce the observed correlations in this source component, has been ruled out by the analysis carried out by Tibbs et al. (2012). In this light, the above correlations indicate that the 13.5 GHz emission observed in G29.0−0.6 could indeed be ascribed to AME. In the case of the G29.1 system (i.e., G29.1−0.6 and G29.1−0.7), no significant correlation is found. On the contrary, we find some indication of an anti-correlation between the microwave emission and Y_PAH. This result is in apparent contradiction with the one for G29.0−0.6. Without additional information, we can only speculate that, in this case, the 13.5 GHz emission may have an origin other than dipole emission from PAHs or VSGs.

Interestingly, our analysis highlights, for both G29.0−0.6 and the G29.1 complex, also an anti-correlation between the spatial distribution of AME and N_H. Similar behavior has been observed by Lagache (2003) and Vidal et al. (2011). We can interpret this result as evidence that the bulk of AME is not generated in the PDR, where the peak of N_H is reached along the line of sight, rather in the ionized interior of the H ii region (in G29.0−0.6) or in the surrounding diffuse medium (i.e., G29.1−0.1 and G29.1−0.7). In these regions, the abundance of PAHs and VSGs will be lower with respect to the abundance in the PDR, especially in the interior of G29.0−0.6, due to radiation pressure drift, as investigated by Draine (2011). However, our result suggests that, despite a decreased abundance, mechanisms such as ion collisions and plasma drag may supply the necessary momentum to the grains to allow them to spin, and hence emit efficiently at microwave wavelengths.

As described in Section 2, we have performed polarization observations of RCW175 at 21.5 GHz using one-dimensional scans through the center of the region. Our polarization measurements stand in the low signal-to-noise regime (typically PI/σ  <  3). In this case, it is well known (see, e.g., Wardle & Kronberg 1974; Simmons & Stewart 1985) that the determination of reliable polarized fluxes is hindered by the fact that the posterior distribution of the polarized intensity, PI, does not follow a normal (Gaussian) distribution. Also, PI is a quantity that must always be positive, which introduces a bias into any estimate. For any true PI₀ we would expect to measure on average a polarization PI  >  PI₀. In order to de-bias the measured polarized fluxes, thus obtaining an estimate of the underlying real polarization level PI₀, we follow the Bayesian approach described in Vaillancourt (2006), consisting of integrating the posterior probability density function over the parameter space of the true polarization. For PI/σ < 5 we then integrate the analytical posterior shown in Vaillancourt (2006), to get the most likely de-biased polarized intensity and the associated confidence intervals at the 68.3% level, setting upper limits when ${PI}/\sigma \,{<}\,\sqrt{2}$. When PI/σ  ⩾  5, we just obtain the polarized intensity as $\sqrt{PI^2-\sigma ^2}$, following Wardle & Kronberg (1974). Vaillancourt (2006) focuses on the de-biasing of the polarized intensity, while no analytical solution is given for the posterior of the polarization fraction. Then, in order to de-bias the polarization fraction we numerically evaluate the corresponding posterior through Monte Carlo simulations, and obtain estimates of PI/I or upper limits in an equivalent way as it is done for PI.

The de-biased estimates for the polarized signals are shown in Figure 10. In the top plot we show the polarization intensity in units of mJy beam⁻¹, where the receiver beam at 21.5 GHz has an FWHM of 67 arcsec. Overplotted is the intensity measured at the same frequencies. The points in the plot are separated by one beam and the errors are quoted at 2σ with systematic uncertainty (which includes the calibration, the atmospheric opacity, and the residual instrumental polarization) linearly added in the dotted error bars. In the lower plot we show polarization fraction. Also in this second plot, the dotted error bars include the statistical and the systematic uncertainties, in this case being quantified to the level of 0.3%.

Zoom In Zoom Out Reset image size

The observed source appears to be weakly polarized to the level of a few percent, especially in the center of our scans. Limiting our analysis to the core of the source (i.e., for galactic latitude 2896 < GL < 2920), we find a weighted average polarized signal of 2.2 ± 0.2(rand.) ± 0.3(sys.)% (68% CL), where statistical uncertainties are the result of the weighted average within the core of the source and the systematic uncertainty is the residual instrumental polarization after correction.

Considering the relatively steep spectrum of the integrated low frequency emission of RCW175, it is plausible that some synchrotron contamination is present along the line of sight toward the source accounting for the relatively low level of polarization. We have modeled the presence of synchrotron radiation through the low frequency fit performed in Section 3. Assuming an average 30% polarization only for the synchrotron component, we expect up to 10% polarization of the total emission at 21.5 GHz. Also, weak polarization could be due to free–free emission in particular specific conditions of optically thick medium toward the edges of the clouds (Keating 1998). On the other hand, our result is compatible with the low level of polarization expected from electric dipole emission (i.e., ≲ 1% at the frequencies of our observations; Hoang et al. 2013). Multi-frequency polarization measurements would be required to discriminate whether the polarized emission comes from the synchrotron or AME component.

We have observed the H ii region RCW175 with the Parkes telescope at 8.4, 13.5, and 21.5 GHz. These observations confirm and strengthen the evidence for AME spatially correlated with thermal dust emission. In particular, approximately half of the RCW175 flux density in the range 15–30 GHz appears to be due to AME, as previously reported by Dickinson et al. (2009). The SED fit suggests that multiple spinning dust components are present in the region, i.e., one associated with diffuse ionized gas, and one associated with dense molecular gas.

Combining the new Parkes observations with ancillary data, we generated an anomalous emission map at 13.5 GHz for the RCW175 complex. The map shows that the microwave excess is characterized by both a diffuse component and a compact structure that overlaps with the location of G29.0−0.6. This corroborates the hypothesis that AME regions have a spatial extension not coincident to the bulk of microwave emission in a given region, as noted, based on statistical arguments, by Planck Collaboration (2014a).

The cross-correlation analysis of the microwave map with both the IR maps and the DustEM parameter maps obtained for the whole region by Tibbs et al. (2012) highlights a good degree of correlation mainly for G29.0−0.6. In this component, the investigation on small angular scales reveals an overall decrease of the microwave–IR correlation, with the exception of the 24 μm map, for which the correlation is more pronounced with respect to other wavelengths. No strong correlation is found for the other H ii region components (i.e., G29.1−0.6 and G29.1−0.7). In G29.0−0.6, the correlation with short wavelength IR data, with PAH abundance, Y_PAH, with the radiation field strength, χ_ERF, and with dust temperature, T_EQ, might indicate that the AME carriers are small dust grains. However, we emphasize that the origin of the 24 μm emission arising from the interior of H ii regions is still debated.

The observed anti-correlation of N_H with the microwave map, already observed by Lagache (2003) and Vidal et al. (2011), suggests that the bulk of AME is likely emitted in low-density regions, i.e., in our case the interior of G29.0−0.6 and the diffuse medium surrounding the G29.1 complex. Additionally, it is important to keep in mind that, if the intensity of the excitation field increases, the ion fraction equally increases, leading to a higher degree of excitation for PAHs and VSGs, hence to an effective anti-correlation between microwave emission and N_H. In the G29.1 complex, the prevalent diffuse emission and the mild anti-correlation between Y_PAH and the 13.5 GHz map seem to suggest a different origin for the observed microwave excess. We emphasize that part of the 13.5 GHz, microwave excess could be due to a rising of the spectral indices between 1.4 GHz and 5 GHz, compatible with the presence of strong stellar winds (Scaife 2012; Paladini et al. 2014) from massive OB stars. Polarization measurements at 21.5 GHz show a low level of average polarized signal toward the center of the source of 2.2 ± 0.2(rand.) ± 0.3(sys.)% (68% CL) compatible with the low level of polarization of the AME or with a residual astrophysical contamination due to synchrotron or free–free from the observed region. This aspect of the analysis demands further investigations in the future.

We acknowledge the use of data from the Planck and Herschel ESA missions. Some of the results are based on observations made with the Spitzer Space Telescope, which is operated by the Jet Propulsion Laboratory, California Institute of Technology under a contract with NASA. We acknowledge the use of the HEALPix package. We acknowledge the logistic support provided by Parkes operators. The Parkes radio telescope is part of the Australia Telescope National Facility, which is funded by the Commonwealth of Australia for operation as a National Facility managed by CSIRO. We thank the referee for constructive comments that improved the paper.