SUPERNOVA REMNANT W49B AND ITS ENVIRONMENT

The 1420 MHz radio continuum and H i emission data come from the Very Large Array Galactic Plane Survey (VGPS; Stil et al. 2006). The continuum data has a spatial resolution of 1' and a noise of 0.3 K. The H i spectral line images have a resolution of 1' × 1' × 1.56 km s⁻¹ and a noise of 2 K per channel. We also use the 20 cm continuum data from the Multi-Array Galactic Plane Imaging Survey (MAGPIS) which have a resolution of ∼6'' (Helfand et al. 2006). The ¹³CO(J = 1–0) spectral line data is from the BU-FCRAO Galactic Ring Survey (Jackson et al. 2006). The data has an angular and spectral resolution of 46'' and 0.21 km s⁻¹ with a noise of about 0.13 K, respectively. We get the CO(J = 3–2) data from the first release of the JCMT CO High-Resolution Survey (Dempsey et al. 2013). The data is smoothed to a velocity resolution of 1 km s⁻¹ and a spatial resolution of 166. The noise in the region of W49B is about 0.44 K.

The Spitzer Infrared Spectrograph (IRS; Houck et al. 2004) observation of W49B was taken on 2005 April 24 using the staring mode (Program ID: 3483, PI: J. Rho). The short–low slit (SL, 5.2–14.5 μm) and long–low slit (LL, 14.0–38.0 μm) are used in the observation. The bad and rogue pixels of the co-added BCD frames are corrected with the IRSCLEAN tool provided by the Spitzer Science Center. Since W49B is extended, only different orders are used to remove the unrelated diffuse emission in the direction of the SNR. After background subtraction, an optimal spectrum is extracted by the SPICE software. The spectrum extraction region is selected to meet criterion of the brightest overlapped region between SL and LL. Error in the final extraction is of the order of 10% which are mainly caused by the background subtraction (Andersen et al. 2011).

The Herschel observation were taken on 2010 October 24 as part of the Herschel Infrared Galactic Plane Survey (Hi-GAL). The PACS (Poglitsch et al. 2010) and SPIRE (Griffin et al. 2009) are used under the parallel mode. The covered bands are 70 μm, 160 μm, 250 μm, 350 μm, and 500 μm. Each band has a measured beam of ∼97 × ∼107, ∼132 × ∼139, ∼228 × ∼239, ∼293 × ∼313, and ∼411 × ∼438, respectively (Traficante et al. 2011). The PACS bands (70 μm and 160 μm) have been processed using the standard pipeline. Since the SPIRE data have a stripe problem, we reprocess them by combining the scripts "Photometer Map Merging" with "Baseline Removal and Destriper" to merge the two direction scan data to a single map. We also remove the SPIRE observation baseline and do a correction for the relative gain of the bolometer. The process uses scripts within the software package Herschel Interactive Processing Environment (version 11.0.1) and calibration tree 11.0, which was the most up-to-date version at the time.

The 2.12 μm data is from the "UKIRT Widefield Infrared Survey" for H₂ (Froebrich et al. 2011). Other infrared continuum images of WISE 12 μm (Wright et al. 2010), Spitzer 24 μm (Carey et al. 2009), MSX 12.1 μm, 14.7 μm, and 21.3 μm (Price et al. 2001) are collected directly from the NASA/IPAC Infrared Science Archive.

The left panel of Figure 1 displays the 1420 MHz continuum image around W49B. We build the H i absorption spectrum by the revised methods presented by Tian et al. (2007) and Tian & Leahy (2008). The methods minimize the possibility of a false absorption spectrum due to the potential H i distribution difference in two lines of sight. In the right panel, the marked white boxes are used to extract the spectra. Regions between the white boxes and yellow boxes are the background. For W49B, we select two bright regions to analyze its H i absorption spectrum.

Zoom In Zoom Out Reset image size

Figure 2 shows the H i absorption spectra of W49B, W49A, and three nearby H ii regions. The ¹³CO(J = 1–0) spectral lines in each panel are extracted from the same regions as H i spectra. There are three main H i absorption features in the spectra, which are at ∼10 km s⁻¹, the ∼40 km s⁻¹, and the ∼60 km s⁻¹, respectively. For H ii regions G42.43–0.26 and G43.19–0.53, radio recombination lines are detected at 62.7 km s⁻¹ and 55.0 km s⁻¹ (Lockman 1989). Since H i absorptions appear up to ∼70 km s⁻¹, the two H ii regions are located at the far side distances, i.e., ∼7.8 kpc and ∼8.4 kpc, respectively (using the Galactic rotation curve model with R_☉ = 8.5 kpc and V_☉ = 220 km s⁻¹). Therefore, the ∼10 km s⁻¹ absorption of G42.43–0.26 and G43.19–0.53 likely originates from local H i clouds. H ii region G42.90+0.58 (Wood & Churchwell 1989) is located outside the solar circle because of its absorption at negative velocity. The trigonometric parallax measurement (Gwinn et al. 1992) gives W49A a distance of 11.4 kpc, which is consistent with the far side kinematic distance of ∼10 km s⁻¹. Compared with the absorption spectra of G42.43–0.26 and G43.19–0.53, stronger absorption at ∼10 km s⁻¹ is detected toward both W49A and G42.90+0.58. This is because the absorption at 10 km s⁻¹ of W49A and G42.90+0.58 arises from both local and Perseus spiral arms.

Zoom In Zoom Out Reset image size

The absorption spectrum of W49B displays similar features as G42.43–0.26 and G43.19–0.53 at the ∼10 km s⁻¹ and no absorption at ∼12 km s⁻¹ (also see the H i channel map in Figure 3), thus, reasonably, its absorption at 10 km s⁻¹ likely originates from the local H i gas. The lack of absorption from the Perseus spiral arm hints that its distance is likely closer than W49A. To confirm this, we calculate the optical depth, assuming W49B is located at the same distance as or behind W49A, i.e., 11.4 kpc. The brightness temperature of H i emission at a velocity of ∼12 km s⁻¹ in the direction of W49B is T_b = 70–80 K. Assuming that the background continuum emission is very low (T_bg = 0 K) and the spin temperature of H i cloud (T_s) ranges from 90 to 300 K. The Radiative transfer equation gives the H i optical depth ${\tau }_{{\rm H\,{\scriptsize {\rm I}}}} = - \ln (1 - ({{{T_b}}}/{{{T_s}}}))$, thus we derive a ${\tau }_{{\rm H\,{\scriptsize {\rm I}}}}$ of 0.3–2.2, which is above the detection sensitivity of 0.1. Since no absorption at ∼12 km s⁻¹ is detected, W49B should be closer than W49A. Further analysis will be presented in the Section 4.

Zoom In Zoom Out Reset image size

Figure 4 shows the IRS map of W49B. The short boxes represent the SL slits and the long boxes are for the LL slits. To correct the extinction, we use the dust model of Draine (2003) with R_V = 3.1. The selected hydrogen column density of 4.8 × 10²² cm⁻² is the average of the results of several X-ray observations (e.g., Hwang et al. 2000, Miceli et al. 2006, Keohane et al. 2007). The final spectrum is presented in Figure 5.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We calculate the flux of each spectral line through a gauss fitting before and after extinction correction. The results are listed in Table 1. As displayed in Figure 5 and Table 1, pure rotational transition lines of H₂ (0,0) S(0)–S(7) are detected. Since these bright lines usually appear in the shocked region, this supports that W49B is interacting with molecular clouds at its southwest boundary. Besides the molecular lines, forbidden transition lines of atoms and ions are also detected, i.e., lines from S⁰, S^{2 +}, Ar⁺, Cl⁺, Fe⁺, Ne⁺, Ne^{2 +}, and Si⁺. Detecting spectral lines from different gas phase suggests W49B has a complex environment.

Figure 6 presents the middle and far infrared images (from WISE, Spitzer, and Herschel) and the 20 cm continuum image from MAGPIS. W49B's morphology in the infrared images with a wavelength of less or equal to 70 μm is similar with that in the radio image, which indicates that the infrared emissions are correlated with W49B. At 160 μm, unrelated foreground and background cold dust emissions begin to dominate the image, but weak features correlating with the radio emission can still be seen. For longer wavelengths, infrared emissions related with W49B entirely disappear in the images.

Zoom In Zoom Out Reset image size

We fit the spectral energy distribution (SED) to understand the property of dusts related to W49B. In Figure 6(c), we select the region within the polygon excluding the circle to extract the emission fluxes from 12 μm to 160 μm. Two rectangles are used to subtract the local background. Because W49B shows strong spectral lines (see Figure 5), which contribute important emissions to the middle infrared images, we correct the line contributions by convolving the IRS spectrum with the response curves of Spitzer and MSX. Because the synchrotron flux of W49B at 6 cm is about 38 Jy and the spectral index is about −0.48 (Moffett & Reynolds 1994), the contribution of electron synchrotron emission to the infrared flux is negligible (The contributed fluxes from 160 μm to 12.1 μm are 2.21, 1.49, 0.89, 0.84, 0.70, and 0.64 Jy). The fluxes at each wavelength are listed in Table 2. We fit a two-component modified blackbody to the dust SED. The function is

$$\begin{equation} {S_v} = \frac{{{M_h}}}{{{D^2}}} \times {\kappa _v} \times B(v,{T_h}) + \frac{{{M_w}}}{{{D^2}}} \times {\kappa _v} \times B(v,{T_w}), \end{equation} \tag{ 1 }$$

where M_h and M_w represent the masses of the hot and warm dust, B(v, T) is the Plank function, and κ_v is the dust absorption cross section per unit mass, which is from Draine (2003) with R_v = 3.1. Figure 7 displays the results. The hot dust has a temperature of 151 ± 20 K with a mass of 7.5 ± 6.6 × 10⁻⁴ M_☉ and the warm dust has a temperature of 45 ± 4 K with a mass of 6.4 ± 3.2 M_☉. The dust mass of W49B is calculated using a distance of 10 kpc (see Section 4 for distance).

Zoom In Zoom Out Reset image size

Previous H i absorption studies suggested W49B at a distance of ∼8.0 kpc (Moffett & Reynolds 1994) or ∼11.4 kpc (Brogan & Troland 2001). The distance of 8.0 kpc is based on two factors. First, H i absorptions are seen clearly up to a velocity of ∼70 km s⁻¹, which argues that W49B lies behind the tangent point at ∼6.2 kpc (assuming Galactic center distance R_☉ = 8.5 kpc and rotation velocity V_☉ = 220 km s⁻¹). Second, compared with W49A, W49B is absent H i absorptions at velocities from 7 to 14 km s⁻¹ and 50 to 55 km s⁻¹. The lack of absorption at ∼55 km s⁻¹ favors that W49B is at the distance of ∼8.0 kpc. We contrast the H i optical depth (${\tau }_{{\rm H\,{\scriptsize I}}}$) of W49B with W49A in Figure 8. The ${\tau }_{{\rm H\,{\scriptsize I}}}$ differences at velocities from 7 to 14 km s⁻¹ and 50 to 55 km s⁻¹ are confirmed. Beside these, there are two other ${\tau }_{{\rm H\,{\scriptsize I}}}$ differences at velocities of ∼40 km s⁻¹ (Figure 8(a)) and ∼65 km s⁻¹ (Figure 8(b)), which have similar intensities as the $\Delta {\tau }_{{\rm H\,{\scriptsize I}}}$ at ∼55 km s⁻¹. The H i absorption of different regions of W49B has an obvious difference at ∼65 km s⁻¹ (Figure 8(c)). Moreover, in the ${N_{{\rm H\,{\scriptsize I}}}}/{T_s}$ image toward both W49A and W49B, Brogan & Troland (2001) found that there is obvious change on size scales of about 1'. Therefore, the reason supporting the distance of ∼8.0 kpc for W49B is not sufficient.

Zoom In Zoom Out Reset image size

The evidence for distance of 11.4 kpc is weak. Figure 9(e) of Brogan & Troland (2001) shows the ${N_{{\rm H\,{\scriptsize I}}}}/{T_s}$ image of W49B integrated velocities from −4.8 to 12.6 km s⁻¹. They found distinct H i enhancement concentrated toward the south boundary of the SNR. Because the enhancement is coincident with the sharp edge in the 20 cm continuum image, they explained it as the result of interaction between W49B and its ambient medium. If this were true, W49B would be located at the far side distance of ∼5 km s⁻¹, i.e., ∼12 kpc, which is nearly the same distance as W49A. For the H i absorption differences at velocities ranging from 7 to 14 km s⁻¹ between W49A and W49B, Brogan & Troland (2001) suggested that it might be caused by the differences of H i kinematics, distribution, and temperature in the direction of W49B and W49A. However, as illustrated in Section 3.1, even with large scale variations of spin temperature, H i absorption at a velocity of ∼12 km s⁻¹ should be visible if W49B is at 11.4 kpc. There is other evidence supporting that W49B is closer than W49A. Figure 9 shows the 2.12 μm image, which traces the shocked H₂. The left panels of Figure 10 present the intensity maps of ¹³CO(J = 1–0) and CO(J = 3–2) integrated from a velocity of 1–15 km s⁻¹. The shocked H₂ can be clearly seen at the east, south, and west boundaries of W49B, indicating that molecular clouds surround W49B except in the northern part. This picture is also consistent with the radio continuum image (Figure 4 of Moffett & Reynolds 1994, Figure 2 of Lacey et al. 2001), which shows sharp boundaries at the east, south, and west, and diffuse emission at the north. Diffuse emission means long electron free path length, and thus low medium density. On the contrary, CO clouds at ∼10 km s⁻¹ only show strong emission to the north. W49B is likely not associated with the 10 km s⁻¹ CO clouds, and thus is not at the same distance of W49A.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Since the [Fe ii] 1.64 μm, the H₂ 2.12 μm morphology, and the estimated mass of the progenitor prefer W49B are likely in a bubble surrounded by molecular clouds, we check the distribution of CO emission in the direction of W49B to further constrain its distance. Only the ∼40 km s⁻¹ CO cloud has a bubble-like feature (also see Figure 1 of Chen et al. 2014). Because H i self-absorption is detected at velocity of 40 km s⁻¹ (Simon et al. 2001), a previous study suggested that the CO clouds are at the near side distance of 40 km s⁻¹. However, we find multi-velocity components in the 40 km s⁻¹ CO cloud. We suggest that the 40 km s⁻¹ CO cloud is composed of two components: the near side cloud and the far side cloud, which also causes additional H i self-absorption. W49B is likely surrounded by the far side CO cloud, and thus has a distance of ∼10 kpc, which is similar to Chen et al.'s (2014) suggestion.

4.2.1. Molecular Phase

Since H₂ (0,0) S(0)–S(7) lines are usually optically thin, the lines' intensity can be used directly to derive the column density of the upper state of the transitions. Therefore, the lines are a useful tool to estimate the temperature of the shocked H₂ (Hewitt et al. 2009). Figure 11 is the Boltzmann diagram of the shocked H₂. A striking zig–zag pattern is presented in the figure. This implies that the ortho- to para-state ratio (OPR) of the H₂ is out of equilibrium. Under the local thermodynamic equilibrium, a two temperature distributions model, including parameters N(H₂)_{w, h} (the column density of warm and hot H₂ components), T_{w, h} (the temperature), and OPR_{w, h} (the ortho- to para-state ratio), is used to fit the data. Since a free fitting will make the OPR_w become unreasonably small (less than 10⁻³⁸), we fix it to 0.01, 0.05, and 0.10. This leads to a slightly bigger χ² than the free fitting. Our result is presented in Figure 11 and Table 3. The three fitting lines almost completely overlap each other, indicating that they have similar χ². According to the work of Timmermann (1998), the conversion of ortho- to para-state in the shock begins at about 700 K and quickly reaches balance with the ratio of three with a temperature of 1300 K. The temperature of the hot H₂ is ∼1060 K with the OPR of ∼0.96, this might mean that the hot component is going through the transition from ortho to para.

Zoom In Zoom Out Reset image size

Table 3. Fitted Excitation Parameters to the Shocked H₂ Lines

Parameter	N(H2)w	Tw	N(H2)h	Th	OPRh	χ2
	(cm−2)	(K)	(cm−2)	(K)
OPRw = 0.01	8.86E20	246	1.38E20	1050	0.97	14.15
OPRw = 0.05	8.41E20	254	1.32E20	1063	0.96	14.56
OPRw = 0.10	6.91E20	266	1.27E20	1075	0.95	15.60

Download table as:  ASCIITypeset image

4.2.2. Ionic Phase

The abundant ionic spectral lines, especially the Fe⁺ and S^{2 +} lines, are perfectly diagnostic tools to ionic gas. The [S iii] 18.7 μm and 33.48 μm lines are produced by the fine-structure transition of ³P₂ to ³P₁ and ³P₁ to ³P₀. As showed in the left panel of Figure 12, since their upper levels have similar temperature, the line intensity ratio is only sensitive to electron density. The right panel of Figure 12 shows the density dependence on the ratio, computed by solving the rate equations for three level systems. Transition probabilities and collision strengths are from Mendoza & Zeippen (1982) and Galavis et al. (1995), respectively. The collisional deexcitation coefficient is calculated by equation (Osterbrock & Ferland 2006)

$$\begin{equation} {q_{21}} = \int \nolimits _0^\infty {u{\sigma _{21}}f(u)du} = \frac{{8.629 \times {{10}^{ - 6}}}}{{{T^{1/2}}}}\frac{{\Upsilon (1,2)}}{{{g_2}}}, \end{equation} \tag{ 2 }$$

where u is the electron velocity, σ₂₁ represents the cross section of deexciation, f(u) is the electron velocity distribution function, ϒ(1, 2) is the collision strength and g₂ represents the statistical weight of the upper level. The collisional excitation coefficient is derived by

$$\begin{equation} {q_{12}} = \frac{{{g_2}}}{{{g_1}}}e^{(h{v_{21}}/kT)}. \end{equation} \tag{ 3 }$$

The red line in the right panel of Figure 12 shows the ratio, which indicates an electron density of 500–700 cm⁻³. Diagnostics of the [Fe ii] lines are presented, following the work of Hewitt et al. (2009), that use the excitation rate equations of Rho et al. (2001). The line ratio of 17.9/5.35 μm is sensitive primarily to electron density and the ratio of 17.9/25.9 μm depends on both density and temperature. More details can be found in Hewitt et al. (2009). For W49B, the former ratio is 0.25 and the latter value is 0.51. The values prefer a density of 400–600 cm⁻³ and a temperature of ∼10⁴ K.

Zoom In Zoom Out Reset image size

Note that Keohane et al. (2007) suggested the electron temperature and electron density surrounding W49B range from (1000 K, 8000 cm⁻³) to (700 K, 1600 cm⁻³). These measurements are different from ours. The difference may be caused by two factors. First, Keohane et al. (2007) gave the parameters only using a single spectral line (i.e., [Fe ii] 1.64 μm) measurement based on some assumptions by considering a wide range of reasonable excitation conditions, though still not covering all possible conditions for SNR (see Tables 5 and 9 of Hewitt et al. 2009, and Table 3 of Neufeld et al. 2007). This might cause large uncertainty when comparing with our parameters, based on a pair of spectral lines, which are powerful diagnostic tools for the measurement. Second, we might observe parts of W49B that are different from theirs. As revealed by the Spitzer IRS spectrum, W49B has a very complex environment including all three gas phases (molecular, neutral, and ionic gas). In large scale, the temperature and density distribution may form a gradient descent from the inside to the outside due to the stellar wind before SN explosion and the interaction between the SNR and its surrounding cloud. If the IRS observation point is closer to the inner part compared with Keohane et al.'s (2007), the electron temperature and density should be higher and more tenuous.

To clarify this explanation, we must compare the spatial distribution of [S iii] 18.7 μm and 33.48 μm emission, [Fe ii] 5.34 μm, 17.9 μm, and 25.9 μm emission with [Fe ii] 1.64 μm emission. However, only the IRS staring mode is used to observe W49B, the comparison will not be possible until new data are obtained from IRS mapping mode observations. In addition, our electron temperature and density estimate for W49B are similar with SNR G349.7+0.2 (7000 K and 700 cm⁻³, Table 9 of Hewitt et al. 2009). The molecular temperature estimation is also typical (1000–2000 K for hot component, Table 3 of Neufeld et al. 2007, Table 5 of Hewitt et al. 2009).

Hollenbach & McKee (1989) calculated the low excitation lines' intensity in a J-shock with per-shock medium density of 10³–10⁶ cm⁻³. For W49B, we find a shock with a velocity from 60 to 100 km s⁻¹ in 10³–10⁴ cm ⁻³medium can reproduce the main feature of the measured ionic lines. The [Ne iii] to [Ne ii] line ratio is another indicator of J-shock velocity (Andersen et al. 2011; Hewitt et al. 2009). In Figure 13, we show the ratio in the W49B based on the model (Hartigan et al. 1987), thus we suggest a shock velocity of ∼90 km s⁻¹.

Zoom In Zoom Out Reset image size

4.3.1. A Comparison with Previous Results

4.3.2. The Origin of Dust

An SN explosion has been considered as the potential source to produce large amounts of dust in a short timescale, i.e., a few Myr. 2–4 M_☉ newly formed cold dust has been detected in the young Galactic SNR, Cas A, which has an age of 320 years (Dunne et al. 2003). For W49B, the observed infrared emission is not all from newly formed dust, because the total mass of the ejecta is only about 6 M_☉ (Miceli et al. 2008). We have derived that the total mass of dust associated with W49B is 6.4 ± 3.2 M_☉, any newly formed one must consume all of the ejecta to produce dust. Other non-SN origins need to be explored in order to explain the strong infrared emission of W49B.

W49B's infrared morphology follows the radio emission well, the hot component could be naturally explained by the swept up circumstellar and interstellar material. Assuming a gas-to-dust ratio of 125 (e.g., Draine 2003), the needed mass of swept up material is a few ∼10⁻²M_☉. However, this explanation will not work well for the warm component, because the swept up material will be ∼800 M_☉, which is not consistent with the young age of W49B.

Zhou et al. (2011) found that the interactions between W49B and the molecular cloud have a significant affect on the evolution of the SNR, e.g., the formation of overionized plasma due to the evaporation of the cloud. The work of Inoue et al. (2012) also showed that the densest cloud clumps can survive from the shock and remain inside the SNR. Both works indicate that the evaporation of dust from molecular clouds might have a big contribution to the far-infrared emission. Following the work of Lee et al. (2011), the expected infrared emission luminosity of an evaporating cloud in the hot gas of an SNR can be estimated by equation (Dwek 1981)

$$\begin{equation} {L}_{IR} \approx 200n_h^2{\left(\frac{{{R_c}}}{{\rm 1\,pc}}\right)^3}{\left(\frac{{{T_h}}}{{{{10}^7}\,{\rm K}}}\right)^{1.5}}{L_ \odot }, \end{equation} \tag{ 4 }$$

where R_c is the radius of the cloud, and n_h and T_h are the density and temperature of the hot gas. For W49B, n_h is about 5 cm⁻³ (Miceli et al. 2008). T_h ranges from 1.13 keV to 3.68 keV or 1.31 × 10⁷ K to 4.27 × 10⁷ K obtained from the X-ray spectral fitting of different regions of W49B and by different authors (Miceli et al. 2006, 2010; Keohane et al. 2007; Ozawa et al. 2009; Lopez et al. 2009, 2013a). We also estimated the T_h in an indirect way using equation (McKee & Hollenbach 1980)

$$\begin{equation} {T_h} = {\left(\frac{{{v_r}}}{{839\,{\rm km\,s}^{-1}}}\right)^2} \times {10^7}{\,\rm K}, \end{equation} \tag{ 5 }$$

where v_r is the forward shock velocity in the plasma. Keohane et al. (2007) estimated the v_r to be ∼1200 km s⁻¹ for W49B. Therefore, the T_h is derived to be ∼2.0 × 10⁷ K, which is consistent with the value from X-ray spectral analysis. If we use the value of 2.0 × 10⁷ K and assume a radius of $2.3_{ - 0.5}^{ + 0.3}$ pc for the evaporation cloud, the emission of evaporated dusts will have the same luminosity as the observation, i.e., 17.0 ± 8.5 × 10⁴L_☉ derived from equation, L = ∫κ_vB(v, T)M_dustdv. Higher T_h, such as 2.5 × 10⁷ K or 3.0 × 10⁷ K, will reduce the radius to $2.0_{ - 0.4}^{ + 0.3}$ pc or $1.9_{ - 0.4}^{ + 0.2}$ pc. W49B has a dense environment, thus the dust from evaporation clouds is a feasible way to explain the observed far-infrared emission.