VISIR/VLT AND VLA JOINT IMAGING ANALYSIS OF THE CIRCUMSTELLAR NEBULA AROUND IRAS 18576+0341

To map the dust distribution in the LBVN associated with IRAS 18576+0341, we have obtained high angular resolution and high sensitivity N- and Q-band images with VISIR (Lagage et al. 2004), the VLT imager for the mid-IR mounted at the Cassegrain focus of the VLT Unit 3 telescope (MELIPAN). The observations were performed on 2007 July 19 and 26 through the PAH2 filters, centered on known polycyclic aromatic hydrocarbon (PAH) features ($\lambda _{c}=11.26\, \mu \rm m, \Delta \lambda =0.59\, \mu \rm m$) and adjacent continuum (PAH2₂, $\lambda _{c}=11.88\, \mu \rm m$, $\Delta \lambda =0.37\, \mu \rm m$), and in Q1 ($\lambda _{c}=17.65\, \mu \rm m$, $\Delta \lambda =0.83\, \mu \rm m$). We also used the Ne ii filter ($\lambda _{c}=12.80\, \mu \rm m$, $\Delta \lambda =0.21\, \mu \rm m$) and its adjacent continuum (NeII₂ $\lambda _{c}=13.03 \, \mu \rm m$, $\Delta \lambda =0.22 \, \mu \rm m$) as a tracer of ionized gas to compare with the radio images. All the observations were performed under very good and stable weather, with an optical seeing of about 07. The target was observed at air masses ranging between 1.2 and 1.4. We used a fixed pixel scale of 0075, resulting in an FOV of 192 × 192. The standard chopping/nodding technique was adopted for subtraction of the sky background as well as the telescope's thermal emission; secondary mirror chopping was performed in the north–south direction, with a chop throw of 9'', and telescope nodding was applied in the opposite direction with equal amplitude. To further improve the image quality, a random jitter pattern with a maximum throw of 3'' was superimposed on the nodding sequence. A complete log of observations is summarized in Table 1. Data were reduced following the standard VISIR pipeline (version 1.7.0), consisting of co-adding of frames after flat fielding and removal of bad pixels. The chopped and nodded images were then combined to make one image in each filter. Three mid-IR standards were observed just before and after the target acquisition to flux calibrate the data. Such observations were also used to determine the FWHM of standard star images and thus to derive the actual angular resolution of our final maps. Values of the FWHM for each filter are reported in Table 1.

We observed IRAS 18576+0341 on 2004 October 12 at 4.8 GHz (6 cm) and 1.4 GHz (20 cm), with a bandwidth of 100 MHz using the VLA in A configuration, providing a typical beam size of ≈036 and ≈13, respectively. For all the observed bands, 1824+107 was used as phase calibrator and the flux density scale was determined by observing 3C 48.

The data processing was performed using the standard programs of the NRAO Astronomical Image Processing System (AIPS). We compared the images made from the A configuration data with our earlier images obtained using the C configuration (Umana et al. 2005) to confirm that the long-term variability was not present in our data.

After the calibration process, the A configuration uv data sets were combined with the earlier observations performed at the same frequencies using the VLA in C configuration and presented in Umana et al. (2005) and images have been produced at both wavelengths. This allowed us to obtain high-resolution maps of the entire nebula and, at the same time, fully recover all the extended emission. The mapping process was performed by using AIPS task IMAGR with a variety of weighting schemes and the dirty maps were CLEANed down as close as possible to the theoretical noise. The final uniform-weighted (ROBUST −5) maps have a synthesized beam of 038 × 036 and 139 × 127 and the rms noise of 0.04 mJy beam⁻¹ and 0.1 mJy beam⁻¹ at 6 cm and 20 cm, respectively. The noise level (rms) in the maps was estimated by analyzing an area on the map (using task IMEAN), whose dimension is of the order of more than 100 θ²_syn away from the phase center and free from evident radio sources.

All five VISIR images, shown in Figure 1, have a similar morphology, with an extended circumstellar envelope surrounding a central source that is clearly visible until almost 13 μm. In Figure 1, we show the images obtained in the filters PAH2₂, NeII₂, and Q1. From all the filters, we derived a shell diameter of approximately 7'', corresponding to a size of 0.35 pc at a distance of 10 kpc (Clark et al. 2009).

Zoom In Zoom Out Reset image size

We performed aperture photometry of the central object in each image by using the IDL procedure ATV, while the nebula contribution at each wavelength has been determined by integrating the flux emitted in a selected area of 9'' × 9''. The photometric results are summarized in Table 2. On the basis of their Infrared Space Observatory (ISO) SWS spectrum, Hrivnak et al. (2000) pointed out a strong difference between the IRAS and the simulated ISO photometry, concluding that the dusty envelope around IRAS 18576+0341 should have been very extended with some of the flux observed by IRAS falling outside of the largest ISO aperture (⩾30''). However, we fully recover the IRAS 12 μm flux density (F$_{60\, \mu \rm m}$ = 58.48 mJy), implying that, at least at 12 μm, all of the emission is from the compact 7'' source.

Table 2. Observed Mid-infrared Flux

Filter	F$_{\rm {star}}$	F$_{\rm {neb}}$
(μm)	(Jy)	(Jy)
11.26	0.12 ± 0.02	41.2 ± 0.2
11.88	0.15 ± 0.02	51.9 ± 0.2
12.80	0.19 ± 0.03	78.8 ± 0.1
13.04	0.17 ± 0.03	74.7 ± 0.1
17.65	...	224.7 ± 0.1

Download table as:  ASCIITypeset image

With respect to the previous mid-infrared images (Ueta et al. 2001) that discovered the presence of a roughly circularly symmetric nebula around IRAS 18576+0341, our higher resolution and sensitivity allow us to probe finer details of the dust distribution. The brightness of the shell varies around its circumference, with the major contribution to the flux coming from the north-west part of the nebula. The condensations, detected by Ueta et al. (2001) as an arc located in the north side of the shell, now show up as a clumpy structure with two major peaks located at 50° in the east-north direction and about 2'' from the central star. It appears that the northern and the southern emission peaks reported by Ueta et al. (2001) were an effect of the limited sensitivity and spatial resolution of their maps. From our VISIR maps, no hints for the claimed optically thin edge-on dust torus surrounding the central object are evident but only a suggestion of a possible spherical inner shell whose edges are strongly structured.

It is possible to isolate the PAH feature at $11.26\, \rm {\mu m}$ by subtracting the image obtained at the adjacent continuum (PAH2₂) from the PAH2 image. We have performed this subtraction after registering the images on the central object and normalizing them to the continuum as derived from a fit to the ISO spectrum (Sloan et al. 2003). The difference image is shown in Figure 2. The $11.26\, \rm {\mu m}$ PAH emission appears to be located in the inner shell with an higher concentration in the northwest part of the ring where the most of the thermal dust continuum emission originates.

Zoom In Zoom Out Reset image size

The [Ne ii] emission, which should trace the ionized gas, can be isolated following the same procedure, i.e., by subtracting the adjacent continuum image at $13.03\,\rm {\mu m}$ (NeII₂) from the $12.80\,\rm {\mu m}$ image. The striking result, shown in Figure 3, demonstrates that the [Ne ii] emission has a very different distribution compared to the dust and PAH tracers. As [Ne ii] emission is an ionized gas tracer this result suggests that significant difference exists between the dusty and ionized stellar ejecta (Section 3.2).

Zoom In Zoom Out Reset image size

Assuming that the extinction is constant across the images and that the flux that we have mapped is due to optically thin thermal emission from dust grains, we may use the NeII₂ and Q1 images to produce a ratio map. In this case, the observed intensity is given by

$$\begin{equation} I_{\nu }\approx B_{\nu }(T)\tau _{\nu }, \end{equation} \tag{ 1 }$$

and using the Wien approximation (hν ≫ kT) for the blackbody expression B_ν(T), the ratio of the intensities at two different frequencies ν₁ and ν₂ is given by

$$\begin{equation} \frac{ I_{\nu _1} }{I_{\nu _2} }=e^{-h(\nu _1-\nu _2)/kT} \left(\frac{ \lambda _{2}}{\lambda _{1}}\right)^{3}\frac{\tau _{\lambda _1} }{\tau _{\lambda _2}}. \end{equation} \tag{ 2 }$$

The optical depth τ_λ depends on the chemical composition and on the size a of the grains. There is an open controversy on the nature of dust grains present in the nebula surrounding IRAS 18576+0341 (Ueta et al. 2001; Hrivnak et al. 2000), indicating that it is possibly a mixed-chemistry object where features due to silicates and those related to carbon (PAH) coexist. In this paper, we follow Umana et al. (2010) and adopt for λ₁ = 11.88 and $\lambda _2=17.65\,\rm {\mu m}$, $\tau _{\lambda _1}/\tau _{\lambda _2}=1.05$ in the case of graphite and 1.27 in the case of astronomical silicates.

For each pixel of the map, we derive the dust temperature inverting Equation (2):

$$\begin{equation} T\approx 1.44\times 10^4 \frac{\big(\lambda _2^{-1} - \lambda _1^{-1}\big)}{\ln \left[ \frac{ I_{\nu _1}}{I_{\nu _2}} \frac{\tau _{\lambda _2} }{\tau _{\lambda _1}} \left(\frac{\lambda _1}{\lambda _2}\right)^{3} \right] }. \end{equation} \tag{ 3 }$$

The dust temperature has been computed only at those pixels whose brightness is greater than $0.06\,\rm {Jy\, arcsec^{-2}}$, corresponding to 3σ in the maps. In this derivation, we assume a constant temperature along the line of sight for each pixel.

The resulting dust temperature map is shown in the left panel of Figure 4. The temperature map can be used to examine dust temperature gradients across the nebula. We see that the color temperature values are confined to a relatively narrow range, from ∼130 to 160 K in the case of graphite grains. Slightly lower values (∼10%) are obtained if instead a mixture of silicate grains is assumed.

Zoom In Zoom Out Reset image size

The higher temperatures are reached at a distance of about 2'' from the center. To more clearly see the temperature distribution as a function of radius, in the right panel of Figure 4 we show radial cuts of the temperature map along diameters obtained at steps of 10°.

To evaluate the mass of the dust in the envelope, an optical depth map at $\lambda =17.65\,\rm {\mu m}$ is required. This can be obtained by inverting Equation (1), using the value of the temperature found in each point of the map and the observed intensity. If the medium is homogeneous, the optical depth is given by τ = k_dust ρ_dust l, where ρ_dust and l are the density of the medium and the thickness of the envelope and k_dust is the absorption coefficient per mass unit.

The mass of the nebula sampled by each pixel of the optical depth map is given by Δm = τ/k × ΔA, where the pixel area ΔA is $1.25 \times 10^{32}\,\rm {cm^{2}}$, with a pixel size of 0075 and assuming a distance of 10 kpc (Clark et al. 2009). A value of k_dust = 253 and $1082\,\rm {cm^2\,g^{-1}}$ at $\lambda =17 \,\rm {\mu m}$ has been adopted in the case of graphite and silicates, respectively (Umana et al. 2010). Integrating over all the pixels with brightness higher than 3σ, a total dust mass of $8.6\times 10^{31}\rm {g}$ (4.3 × 10⁻³ M_☉) is derived in the case of graphite, and $1.3\times 10^{32}\,\rm {g}$ (6.5 × 10⁻³ M_☉) in the case of silicates. Ueta et al. (2001) provide an estimate of total dust mass of ∼0.1 M_☉ assuming a toroidal morphology for the entire dust emitting region and a radius of ∼14''. This value is compatible with our results if scaled down to the extent of the mid-infrared emitting dust shell that is resolved in our images. Because the mid-IR observations we present only trace the cooler dust component in the nebula, our dust mass estimates should be considered as lower limits to the total dust mass in the LBVN.

The dust column density along the line of sight ρ_dust l (Figure 5, left panel) can be used to infer the density ρ_dust inside the nebula. Although its distribution is clumpy, as already noted, with a higher concentration of dust in the northeast part, a roughly spherical symmetry around the central star can be assumed. Cuts of the column density map along diameters, obtained at steps of 10° averaged together, are shown as a thin line in the right panel of Figure 5. We modeled the density distribution assuming a radial dependence. For this purpose, we built a three-dimensional matrix, sampled with a step corresponding to the pixel size of the VISIR maps (1.12 × 10¹⁶ cm) in all the three spatial dimensions. We found that the density must increase as a function of radius from the center, although it is not possible to discriminate among several radial dependences. As an example, the radial column density profile obtained assuming ρ_dust ∝ r², with a smooth at the edge, is shown as a thick line in Figure 5, right panel.

Zoom In Zoom Out Reset image size

This analysis shows that in the inner part of the nebula the density is low and it increases strongly with the distance from the star. This could indicate the following.

• 1.  
  An episode of strong mass loss at a previous time.
• 2.  
  The dust is processed by the UV radiation field of the star in the inner part of the nebula.
• 3.  
  A combination of the two.

The maximum of ρ_dust is at 3'' from the center, corresponding to a linear distance of 0.15 pc; assuming the same expansion velocity $v_{\rm {e}}\approx 70 \;\rm {km\,s^{-1}}$ measured from the Brγ line (Clark et al. 2009), the mass-loss episode should have occurred about 2000 years ago.

The continuum emission associated with IRAS 18576+0341 at 5 GHz and 1.4 GHz is shown in Figure 6. The spatial resolution achieved in the final image is about 13 in the L Band and 04 in the C Band. The images confirm the structure of the source as detected at higher frequencies (Umana et al. 2005): a compact, slightly resolved core, most likely associated with the stellar wind from the central object, surrounded by an extended ionized nebula with a remarkably similar shape and extension at all wavelengths. As expected, the combined effect of reduced spatial resolution and of intrinsic contribution at low frequency prevent us from detecting the core component and discerning it from the extended nebula in the 20 cm map; in contrast, the higher spatial resolution in the 6 cm map allows us to clearly resolve the compact source.

Zoom In Zoom Out Reset image size

The total radio flux densities associated with the entire source have been derived by summing the contribution of all the pixels contained in the radio source (AIPS task IMSTAT), while the flux density of the central object has been obtained by fitting a two-dimensional Gaussian brightness distribution to the map (AIPS task JMFIT). The results are reported in Table 3.

Table 3. Observed Radio Flux

Filter	F$_{\rm {star}}$	F$_{\rm {neb}}$
(cm)	(mJy)	(mJy)
6	1.3 ± 0.04	103 ± 2
20	⋅⋅⋅	92 ± 2

Download table as:  ASCIITypeset image

To better separate the core component from the remaining diffuse emission, the flux density from the compact core at 6 cm has been determined by fitting a two-dimensional Gaussian brightness distribution to the map obtained with the A configuration data alone. The resulting flux density of 1.3 ± 0.1 is in accordance with the flux density extrapolated from measurements at higher frequencies assuming a stellar wind spectrum, as reported in Umana et al. (2005).

The final images in Figure 6 show that the radio emission from the circumstellar ionized nebula is asymmetric with respect to the central core, extending out in the northwest direction, with the brightest radio region located about 36 to the north of the central core in the 6 cm map. The diffuse emission is remarkably similar to that detected at all the other observing frequencies, with similar extent of about 7''. The southern component, which contributes about 25% of the total mid-infrared emission, is not present in any of the radio maps. A certain degree of asymmetry is also present in the radio emission of other LBVNs, such as HR Car (White 2000) and Pistol Star (Lang et al. 2005, 1999). This may indicate that extremely asymmetric mass loss can be quite common in LBVs.

The overall morphology of the radio nebula agrees very closely with the spatial distribution of the [Ne ii] line emission. The contours of the [Ne ii] map, superimposed on a gray-scale image of the 6 cm data, is shown in Figure 3. It is evident how the emission in the [Ne ii] line follows quite faithfully that of the radio continuum. Such resemblance between the two maps is unsurprising, as both are tracers of the thermal ionized gas.

To search for the existence of denser regions with different opacity through the nebula, we computed the spectral index map by comparing the 6 cm image with the 0.7 cm one obtained in the first epoch. The images have comparable spatial resolution. We computed the spectral index, defined by the relation S(ν) ∝ ν^α, using the maps produced with the same beam and cell size for each pixel with a flux density of at least three times the rms noise at both the wavelengths. The resulting map, shown in Figure 7, shows evidence of the stellar wind component, characterized by a spectral index of α ≈ 0.8, and the extended halo having a much flatter radio spectrum (α ≈ −0.1), which is typical of free–free thermal emission. However, the image shows some patchiness probably associated with local density clumps. We note the presence of a second component with a spectral index of α ≈ 0.9 located about 15 to the south of the central core, in the direction of the southern infrared emission peak in the dust nebula. Such a component, which is clearly visible in the radio map reported in Figure 1(b) in Umana et al. (2005), is present probably due to the presence of a dust condensation whose thermal emission contributes to the flux observed at 7 mm.

Zoom In Zoom Out Reset image size