ALMA CO Observations of the Mixed-morphology Supernova Remnant W49B: Efficient Production of Recombining Plasma and Hadronic Gamma Rays via Shock–Cloud Interactions

Observations of ¹²CO(J = 2–1) and ¹³CO(J = 2–1) line emission were conducted using ALMA ACA Band 6 (211–275 GHz) as a Cycle 6 project (proposal No. 2018.1.01780.S). We used the mosaic observation mode with 10–12 antennas of a 7 m array and four antennas of a 12 m total-power (TP) array. The observed areas were $5\buildrel{\,\prime}\over{.} 1\times 2\buildrel{\,\prime}\over{.} 7$ rectangular regions centered at (α_J2000, δ_J2000) = (19^h11^m0900, $+9^\circ 06^{\prime} 24\buildrel{\prime\prime}\over{.} 8$) and (19^h11^m0744, $+9^\circ 05^{\prime} 56\buildrel{\prime\prime}\over{.} 2$). The actual observed area is shown in Figure 1. The combined baseline length of the 7 m array data is 8.85 to 48.95 m, corresponding to u–v distances of 6.8 to 37.6 kλ at 230.538 GHz. Two quasars, J1924−2914 and J1751+0939, were observed as bandpass and flux calibrators. We also observed four other quasars, J1907+0127, J1922+1530, J1938+0448, and J1851+0035, as phase calibrators. We performed data reduction using the Common Astronomy Software Application (CASA; McMullin et al. 2007) package version 5.5.0. We utilized the "tclean" task with a multiscale deconvolver and natural weighting. The emission mask was also selected using the auto-multithresh procedure (Kepley et al. 2020). We combined the cleaned 7 m array data and the calibrated TP array data using the "feather" task to recover the missing flux and diffuse emission. The beam size of the feathered data is 823 × 477 with a position angle of −75.31° for the ¹²CO(J = 2–1) data, and 828 × 504 with a position angle of −7937 for the ¹³CO(J = 2–1) data. The typical noise fluctuations are ∼0.065 K for the ¹²CO(J = 2–1) data and ∼0.055 K for the ¹³CO(J = 2–1) data at the velocity resolution of 0.4 km s⁻¹.

We also used archival data sets of ¹²CO(J = 1–0) and ¹²CO(J = 3–2) line emission for estimating physical properties of molecular clouds. The ¹²CO(J = 1–0) data are from the FOREST Unbiased Galactic Plane Imaging Survey with the Nobeyama 45 m telescope (FUGIN; Umemoto et al. 2017), and the ¹²CO(J = 3–2) data are from the CO High-resolution Survey (Dempsey et al. 2013), obtained with the James Clerk Maxwell Telescope (JCMT). The angular resolution is ∼20'' for the ¹²CO(J = 1–0) data and ∼166 for the ¹²CO(J = 3–2) data. The velocity resolutions of the ¹²CO(J = 1–0) and ¹²CO(J = 3–2) data are 1.3 and 1.0 km s⁻¹, respectively. To improve the signal-to-noise ratio of the ¹²CO(J = 3–2) data, we combined four spatial pixels and the rebinned pixel size was 12''. The typical noise fluctuations are ∼1.4 K for the ¹²CO(J = 1–0) data and ∼0.18 K for the ¹²CO(J = 3–2) data at each velocity resolution.

The radio continuum data at the 20 cm wavelength are from MAGPIS (Helfand et al. 2006), obtained with the VLA and the Effelsberg 100 m telescope. The angular resolution is ∼6'', which is compatible with the ALMA ACA resolution. The typical noise fluctuations are ∼1–2 mJy.

We utilized archival X-ray data obtained by Chandra (the observation IDs are 117, 13440, and 13441), which have been published in numerous papers (e.g., Kawasaki et al. 2005; Lopez et al. 2009a, 2009b, 2011, 2013a, 2013b; Yang et al. 2009; Koo et al. 2016; Zhou & Vink 2018). The X-ray data sets were taken with the Advanced CCD Imaging Spectrometer S-array (ACIS-S3). We used the Chandra Interactive Analysis of Observations (CIAO; Fruscione et al. 2006) software version 4.12 with CALDB 4.9.1 (Graessle et al. 2007) for data reduction and imaging. After reprocessing for all data sets using the "chandra_repro" procedure, we created an energy-filtered, exposure-corrected image using the "fluximage" procedure in the energy bands of 0.5–7.0 keV (broad band; see Figure 1), 0.5–1.2 keV (soft band), 1.2–2.0 keV (medium band), 2.0–7.0 keV (hard band), and 4.2–5.5 keV (continuum band). Because the soft- and medium-band images are heavily affected by interstellar absorption (e.g., Zhou & Vink 2018), in this paper we focus on the X-ray images at energies greater than 2.0 keV. We also created an exposure-corrected, continuum-subtracted image of Fe Heα line emission (6.4–6.9 keV) following the method presented by Lopez et al. (2013b). The typical angular resolution of the Chandra images is ∼05.

Figure 1 shows the Chandra broadband X-ray image of W49B superposed on the VLA radio continuum contours at a 20 cm wavelength. As presented in previous studies, a barrel-shaped radio continuum shell with several co-axis filaments and center-filled X-rays are seen (e.g., Keohane et al. 2007; Lopez et al. 2009a, 2011). The X-ray elongated feature brighter than ∼5 × 10⁻⁷ photons cm⁻² s⁻¹ is roughly consistent with the spatial distribution of Fe Heα emission (see Figure 3 in Lopez et al. 2013b). We note that the overall distributions of X-rays and the radio continuum are quite different between the northeastern and southwestern halves: the shell boundaries of the northeastern halves roughly coincide with each other, whereas the southwestern shell of X-rays is significantly deformed compared to that of the radio continuum. In particular, the X-rays are dim around positions of (α_J2000, δ_J2000) ∼ (19^h11^m000, $+09^\circ 06^{\prime} 00^{\prime\prime}$), (19^h11^m030, $+09^\circ 05^{\prime} 00^{\prime\prime}$), and (19^h11^m080, $+09^\circ 06^{\prime} 06^{\prime\prime}$): the first two correspond to the bright peaks of the radio continuum and the other is partially surrounded by radio filaments. This trend is also seen in the 4.2–5.5 keV band image, which is mostly free from interstellar absorption as well as line emission.

Figure 2 shows integrated intensity maps of ¹²CO(J = 2–1) and ¹³CO(J = 2–1) for three velocity ranges of 1–15 km s⁻¹ (hereafter the "10 km s⁻¹ cloud"), 38–47 km s⁻¹ (hereafter the "40 km s⁻¹ cloud"), and 57–67 km s⁻¹ (hereafter the "60 km s⁻¹ cloud") as previously mentioned in several papers (e.g., Zhu et al. 2014; Kilpatrick et al. 2016; Lee et al. 2020). The kinematic distance of the molecular cloud is ∼11 kpc for the 10 km s⁻¹ cloud, ∼9 kpc for the 40 km s⁻¹ cloud, ⁸ and ∼7 kpc for the 60 km s⁻¹ cloud (Sofue et al. 2021). Note that there are no other CO clouds within the velocity range of −15.0 to 92.6 km s⁻¹, and hence we focus on the three molecular clouds in the present paper.

Zoom In Zoom Out Reset image size

In the 10 km s⁻¹ cloud (Figures 2(a) and (b)), there is an intensity gradient increasing from southeast to northwest. The radio continuum shows fairly good spatial correspondence to molecular clouds in the northern inward protrusion and along the north, northwest, and southwest rims. On the other hand, both the ¹²CO and ¹³CO emission lines are faint in the southeastern shell, where shocked H₂ emission is strongly detected (e.g., Keohane et al. 2007). Dense clouds traced by ¹³CO emission are located not only outside the shell boundary, but also inside the radio continuum shell.

In the 40 km s⁻¹ cloud (Figures 2(c) and (d)), the ¹²CO emission has a relatively uniform distribution compared to that of the 10 km s⁻¹ cloud, but the ¹³CO emission shows an intensity gradient increasing from northwest to southeast. We note that the radio-brightest shell in the west shows a lack of both ¹²CO and ¹³CO emission lines. The bright ¹²CO emission and ¹³CO clumps are located both toward the southwest of the SNR and inside the southeastern shell. The southwestern CO clumps seem to be located along the sharp edge of the radio shell, whereas the CO clumps inside the SNR show no significant spatial correlations with the radio shell morphology. At this spatial coverage, we could not find the bubble-like CO structure toward W49B mentioned by Zhu et al. (2014).

In the 60 km s⁻¹ cloud (Figures 2(e) and (f)), there are dense clouds across the SNR from northeast to southwest with two bright CO peaks at (α_J2000, δ_J2000) ∼ (19^h11^m170, $+09^\circ 07^{\prime} 30^{\prime\prime}$) and (19^h10^m575, $+09^\circ 05^{\prime} 07^{\prime\prime}$). The former contains two H ii regions cataloged by Urquhart et al. (2009), whereas the latter does not have any cataloged objects. The diffuse CO emission inside the SNR appears to be spatially anticorrelated with the radio continuum contours.

To derive the masses of the three molecular clouds, we used the following equations:

$\begin{eqnarray}&&M={m}_{{\rm{H}}}\mu {\rm{\Omega }}{D}^{2}\sum _{i}{N}_{i}({{\rm{H}}}_{2}),\end{eqnarray} \tag{ 1 }$

$\begin{eqnarray}&&N({{\rm{H}}}_{2})=X\cdot W(\mathrm{CO}),\end{eqnarray} \tag{ 2 }$

where m_H is the mass of hydrogen, μ = 2.8 is the mean molecular weight, Ω is the solid angle of each pixel, D is the distance to W49B, N_i (H₂) is the column density of molecular hydrogen for each pixel, X is the CO-to-H₂ conversion factor of 2 × 10²⁰ cm⁻² (K km s⁻¹)⁻¹ (Bertsch et al. 1993), and W(CO) is the velocity-integrated intensity of the ¹²CO(J = 1–0) emission line obtained from the FUGIN data (Umemoto et al. 2017). We estimated the mass of the molecular cloud inside the radio shell to be ∼4.1 × 10⁴ M_☉ for the 60 km s⁻¹ cloud and ∼2.7 × 10⁴ M_☉ for the other two clouds, where we adopted a shell radius of 25 centered at (α_J2000, δ_J2000) = (19^h11^m0734, $+09^\circ 06^{\prime} 01\buildrel{\prime\prime}\over{.} 1$). These values are roughly consistent with previously derived cloud masses using the ¹³CO(J = 1–0) emission and the ¹³CO-to-H₂ conversion factor (H.E.S.S. Collaboration et al. 2018).

Figure 3 shows the velocity channel maps for each molecular cloud superposed on the radio continuum contours. We find that the dense ¹³CO clumps at the velocity range of 3.8 to 9.4 km s⁻¹ are superposed nicely not only on the northern and southern shells, but also on radio filaments inside the SNR at positions of (α_J2000, δ_J2000) ∼ (19^h11^m064, $+09^\circ 07^{\prime} 03^{\prime\prime}$) and (19^h11^m055, $+09^\circ 05^{\prime} 56^{\prime\prime}$). We also find that both the ¹²CO and ¹³CO clouds at the velocity range of 9.4–12.2 km s⁻¹ show global anticorrelation with the radio shell. In addition, ¹²CO emission at the velocity range of 12.2–15.0 km s⁻¹ shows a good spatial correspondence to the western shell especially for the sharp edge of the southwestern rim. This velocity range is consistent with that inferred from Kilpatrick et al. (2016). The southwestern CO clumps at 42.0–44.0 km s⁻¹ seem to be located along the sharp edge of the radio shell especially prominent in the ¹³CO line emission. Moreover, the ¹³CO clouds at 38.0–40.0 km s⁻¹ show good spatial correspondence to the southeastern half of the radio continuum shell. Furthermore, the 63–65 km s^{−1 12}CO map shows a lack of CO emission along the eastern shell and shows bright CO emission in the gaps between the western and northern radio contours. Although the other CO clouds also appear to be overlapped with the radio continuum shell and filaments, their spatial correspondence is not clear.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 4 shows the position–velocity diagrams for the three molecular clouds. The velocity distributions of ¹²CO and ¹³CO emission are similar to each other except for a velocity at ∼40 km s⁻¹, suggesting that the ¹²CO emission at ∼40 km s⁻¹ is subject to self-absorption due to an optically thick component. In fact, the ¹²CO emission at ∼40 km s⁻¹ shows a dip-like feature, whereas the ¹³CO spectrum at the same velocity has strong line emission. We also find incomplete and complete cavity-like structures in the 10 and 40 km s⁻¹ clouds, respectively (see dashed curves in Figures 4(a)–(d)). The incomplete cavity (or arc-like distribution) in the 10 km s⁻¹ cloud lies at 5–11 km s⁻¹ with a velocity dispersion of a few kilometers per second. On the other hand, the complete cavity in the 40 km s⁻¹ cloud is clearly seen especially for ¹²CO emission, whose velocity range is 41–47 km s⁻¹. Note that the spatial extents of these cavities are roughly consistent with the diameter of the radio continuum shell. By contrast, there is no clear evidence for such cavity-like structures in the position–velocity diagram of the 60 km s⁻¹ cloud (see Figures 4(e) and (f)).

Zoom In Zoom Out Reset image size

Figure 5 shows the intensity ratio maps of ¹²CO(J = 3–2)/¹³CO(J = 2–1) (hereafter R_{12CO32/13CO21}) toward the three molecular clouds. A higher value of R_{12CO32/13CO21} tends to be observed in diffuse warm gas thermalized by supernova shocks and/or stellar radiation assuming that the abundance ratio of ¹²CO/¹³CO is constant within a molecular cloud complex (see Bieging & Peters 2011; Dell'Ova et al. 2020). We find that high intensity ratios of R_{12CO32/13CO21} ∼ 10 are distributed toward the shell of W49B in the 10 km s⁻¹ cloud. The southwestern edge of the shell (hereafter the SW-edge) shows the highest value of R_{12CO32/13CO21} ∼ 20, where the radio continuum shell is strongly deformed. On the other hand, the 40 km s⁻¹ cloud shows no significant enhancement of R_{12CO32/13CO21} toward the SNR shell. The southeastern shell in the 60 km s⁻¹ cloud also shows higher values of R_{12CO32/13CO21}, while regions with high intensity ratios continuously extend beyond the radio shell boundary and may not be related to W49B.

Zoom In Zoom Out Reset image size

To reveal the physical conditions for each cloud in detail, we performed large velocity gradient (LVG) analysis (e.g., Goldreich & Kwan 1974; Scoville & Solomon 1974). LVG analysis can calculate the radiative transfer of molecular line emission, assuming a spherically isotropic cloud with a uniform photon escape probability, temperature, and radial velocity gradient of dv/dr. Here, dv is the half-width half-maximum of the CO line profiles and dr is the cloud radius. We selected six individual CO peaks for each cloud, which are significantly detected in both the ¹²CO and ¹³CO emission lines. Four of them appear to be located along the radio continuum shell or filaments (hereafter referred to as "shell clouds"), and the others were selected as reference, which are located outside of the shell (hereafter referred to as "reference clouds"). The CO spectra toward each position are shown in Figure 6. We adopted dv/dr = 2.5 km s⁻¹ pc⁻¹ for the 10 km s⁻¹ SW-shell and the 40 km s⁻¹ SE-shell; 1.5 km s⁻¹ pc⁻¹ for the 60 km s⁻¹ SW-edge, NW-shell, and Reference-SE; and 1.0 km s⁻¹ pc⁻¹ for the others. We also utilized an abundance ratio of [¹²CO/H₂] = 5 × 10⁻⁵ (Blake et al. 1987) and an isotope abundance ratio of [¹²CO/¹³CO] = 49 (Langer & Penzias 1990).

Zoom In Zoom Out Reset image size

Figure 7 shows the LVG results on the number density of molecular hydrogen, n(H₂), and the kinematic temperature, T_kin, toward the six positions for each cloud. The best-fit values of n(H₂) and T_kin are summarized in Table 1. In the 10 km s⁻¹ cloud, we find that the shell clouds show T_kin ∼ 20–60 K, which are significantly higher than that of the reference clouds (T_kin = 15 K). By contrast, all shell clouds in both the 40 km s⁻¹ and 60 km s⁻¹ components show T_kin ∼ 10 K, which is roughly consistent with their reference clouds except for Reference-SE in the 60 km s⁻¹ cloud (T_kin = 21 K). We also note that there is no relation between the number density of molecular hydrogen and the kinetic temperature for each cloud.

Zoom In Zoom Out Reset image size

Table 1. Results of LVG Analysis at the 10 km s⁻¹, 40 km s⁻¹, and 60 km s⁻¹ Clouds

	12CO		13CO
Name	J = 3–2	J = 2–1	J = 2–1	n(H2)	Tkin
	(K)	(K)	(K)	(×103 cm−3)	(K)
(1)	(2)	(3)	(4)	(5)	(6)
10 km s−1 Cloud
SW-edge	1.02	1.35	0.14	$0.{83}_{-0.05}^{+0.10}$	${60}_{-26}^{+47}$
SW-shell	3.12	3.97	1.12	$2.{45}_{-0.16}^{+0.50}$	${18}_{-3}^{+7}$
Inner-filament	2.06	3.25	0.37	$0.{78}_{-0.02}^{+0.03}$	${23}_{-5}^{+9}$
NE-shell	2.16	2.96	0.60	${1.26}_{-0.09}^{+0.15}$	${20}_{-4}^{+8}$
Reference-SW	3.02	4.03	1.43	$1.{86}_{-0.08}^{+0.14}$	${15}_{-2}^{+3}$
Reference-NW	2.49	3.19	1.41	$2.{29}_{-0.20}^{+0.46}$	${15}_{-3}^{+6}$
40 km s−1 Cloud
SE-shell	1.81	3.32	0.68	$1.{78}_{-0.00}^{+0.08}$	${10}_{-2}^{+1}$
SW-edge	1.78	3.13	2.40	$1.{26}_{-0.03}^{+0.03}$	${9}_{-1}^{+2}$
Inner-filament	0.64	1.91	0.22	$0.{69}_{-0.00}^{+0.02}$	${8}_{-2}^{+3}$
N-shell	0.93	5.22	0.23	$0.{76}_{-0.02}^{+0.05}$	${13}_{-3}^{+6}$
Reference-SW	1.65	2.83	0.77	$1.{32}_{-0.09}^{+0.09}$	${10}_{-2}^{+2}$
Reference-N	1.05	2.40	0.42	$0.{93}_{-0.00}^{+0.03}$	${9}_{-2}^{+2}$
60 km s−1 Cloud
SW-edge	2.78	4.61	1.62	$2{.04}_{-0.04}^{+0.05}$	${9}_{-1}^{+1}$
SW-shell	2.09	3.66	1.19	$1.{48}_{-0.03}^{+0.03}$	${9}_{-1}^{+1}$
Inner-filament	5.22	7.93	2.82	$1.{66}_{-0.04}^{+0.08}$	${11}_{-1}^{+1}$
NW-shell	3.98	6.13	1.31	$1.{51}_{-0.03}^{+0.04}$	${14}_{-2}^{+1}$
Reference-SW	1.76	3.64	0.42	${0.91}_{-0.00}^{+0.02}$	${12}_{-1}^{+3}$
Reference-SE	5.20	7.15	1.33	$1.{17}_{-0.02}^{+0.06}$	${21}_{-2}^{+3}$

Note. Column (1): Region name for each cloud. Columns (2)–(3): Radiation temperature for each line emission derived by least-squares fitting using a single Gaussian function. Column (4): Number density of molecular hydrogen. Column (5): Kinetic temperature.

Download table as:  ASCIITypeset image

Previous studies have proposed three candidates of molecular clouds that are interacting with the SNR W49B: namely a 10 km s⁻¹ cloud, a 40 km s⁻¹ cloud, and a 60 km s⁻¹ cloud (Zhu et al. 2014; Kilpatrick et al. 2016; Lee et al. 2020). Their claim is mainly based on three elements: (1) a line-broadening feature of CO emission at ∼10 km s⁻¹, (2) a wind-bubble-like morphology of a CO cloud at ∼40 km s⁻¹, and (3) a central velocity of shocked H₂ line emission at ∼64 km s⁻¹. In this section, we discuss which cloud is the most likely to be associated with W49B in terms of spatial distributions, the presence of expanding gas motion, and the physical conditions of the CO clouds.

4.1.1. Spatial Distributions of CO Clouds

4.1.2. Shock- and Wind-induced Expanding Gas Motion

4.1.3. Kinetic Temperature of Molecular Clouds

4.1.4. Final Decision and Consistency with Previous Studies

To reveal a physical relation between the 10 km s⁻¹ cloud and X-ray radiation, we here compare the CO distributions with Chandra X-ray images. ⁹ Figure 8(a) shows the ALMA ACA ¹²CO(J = 2–1) integrated intensity overlaid with the Chandra X-ray contours at the energy band of 2–7 keV. In the integration velocity range of 12.2–15.0 km s⁻¹, the CO clouds are perfectly located along the zigzag pattern of the western X-ray shell, indicating that shock ionization occurred. Note that spatial correspondence is also seen in the X-ray image at 4.2–5.5 keV, which is mostly free from interstellar absorption. We thus suggest that the shockwave was strongly decelerated and deformed in the western shell along the dense clouds, whereas the eastern shell was freely expanded with a smooth shape of the forward shock. This also indicates that the shock velocity of the eastern shell is faster than that of the western shell. Further proper-motion studies might be able to reveal the velocity difference in the east–west direction.

Zoom In Zoom Out Reset image size

Figure 8(b) shows an overlay map of the ¹³CO(J = 2–1) intensity image and the continuum-subtracted Fe Heα emission in white contours. To compare the Fe-rich ejecta with the total amount of dense clouds, we use ¹³CO with the whole velocity range of 1.0–15.0 km s⁻¹. Although the elongated structure of Fe-rich ejecta is believed to be related to a bipolar/jet-driven Type Ib/Ic explosion (Keohane et al. 2007; Lopez et al. 2013a), the Fe-rich ejecta is mainly located in the void of dense molecular clouds. Moreover, the Fe-rich ejecta is almost completely surrounded by dense molecular clumps. We argue that this situation is consistent with the supernova explosion inside a barrel-shaped cavity that was proposed by Zhou & Vink (2018). The authors revealed that an enhancement of the cool plasma component along the Fe-rich ejecta (or in the void of dense clouds) was observed by spatially resolved X-ray spectroscopy (see Figure 4 in Zhou & Vink 2018). Following the proposed scenario, the forward shock was freely expanded in the low-density medium at the beginning, and then it suddenly encountered the dense gaseous materials traced by ¹³CO line emission and/or the cool plasma component. Since shock–cloud interaction generates multiple-reflected (or inward) shocks, the Fe-rich ejecta is efficiently heated up at higher densities toward the center of the SNR (see also Sano et al. 2019b). The X-Ray Imaging and Spectroscopy Mission (XRISM; XRISM Science Team 2020) will provide us with further understanding of shock interactions through a detailed spatial comparison between X-ray-derived ionic properties and CO clouds.

It is a long-standing question how recombining (overionized) plasma is formed in SNRs since its discovery in 2002 (IC443; Kawasaki et al. 2002). Subsequent detailed X-ray spectroscopic observations have revealed that nearly 20 SNRs show an overionized state (e.g., W49B, Ozawa et al. 2009; G359.1−0.5, Ohnishi et al. 2011; W28, Sawada & Koyama 2012; W44, Uchida et al. 2012; G346.6−0.2, Yamauchi et al. 2013; 3C 391, Ergin et al. 2014; CTB 37A, Yamauchi et al. 2014; G290.1−0.8, Kamitsukasa et al. 2015; LMC N49, Uchida et al. 2015; Kes 17, Washino et al. 2016; G166.0+4.3, Matsumura et al. 2017a; 3C400.2, Ergin et al. 2017; LMC N132D, Bamba et al. 2018; HB21, Suzuki et al. 2018; CTB1, Katsuragawa et al. 2018; Sagittarius A East, Ono et al. 2019; and G189.6+3.3, Yamauchi et al. 2020; see also a review by Yamaguchi 2020). However, the physical origin of recombining plasmas is still under debate.

Since recombining plasma is characterized by higher ionization temperature kT_i compared to the electron temperature kT_e, rapid electron cooling or an increasing ionization state is needed to produce the plasma state. Three scenarios have been proposed to explain the origin of recombining plasmas in SNRs, called the adiabatic cooling, thermal conduction, and photoionization scenarios. In the adiabatic cooling (a.k.a. rarefaction) scenario, rapid electron cooling occurs when the shockwaves break out from a dense ISM (e.g., CSM) into a much less dense medium (e.g., Itoh & Masai 1989; Masai 1994; Yamaguchi et al. 2018). In the thermal conduction scenario, such rapid electron cooling is caused by interactions between the shockwaves and cold dense clouds through thermal conduction (e.g., Kawasaki et al. 2002; Matsumura et al. 2017a, 2017b; Okon et al. 2018, 2020). On the other hand, the photoionization scenario proposes that external X-ray radiation or low-energy CRs increase the ionization state via photoionization (e.g., Nakashima et al. 2013; Ono et al. 2019; Hirayama et al. 2019). Because photoionization can be seen in limited environments such as near the Galactic center or an SNR with strong Fe i Kα emission, the adiabatic cooling and thermal conduction scenarios are thought to be the formation mechanisms of recombining plasmas in most SNRs.

The origin of recombining plasma in W49B has been discussed in the past decade. The thermal conduction scenario was initially proposed by Kawasaki et al. (2005), whereas the adiabatic cooling scenario is more highly favored in subsequent studies (Miceli et al. 2010; Lopez et al. 2013a; Zhou et al. 2011; Yamaguchi et al. 2018) because the recombining plasma in W49B shows a positive correlation between the ionization timescale n_e t and kT_e. Further, there is no correlation between the plasma condition and ambient clouds traced by near-infrared emission (Yamaguchi et al. 2018). This trend is in contrast to what is observed in W44 (see also Okon et al. 2020 and a review in Yamaguchi 2020). On the other hand, most recent X-ray studies have presented the observation that the X-ray spectra from W49B are reproduced by two ejecta components (low- and high-temperature plasma). The authors have proposed thermal conduction scenarios especially for the high-temperature recombining plasma in W49B, considering the conduction timescale (Sun & Chen 2020; Holland-Ashford et al. 2020). Note that Holland-Ashford et al. (2020) argued that thermal conduction is a possible origin of recombining plasma in the eastern regions of W49B because dense molecular clouds are thought to be associated with the southwestern shell (Keohane et al. 2007; Zhu et al. 2014). In this section, we argue that the origin of the high-temperature recombining plasma in W49B can be understood to be the thermal conduction scenario considering the CO-traced interacting molecular clouds in W49B.

Figure 9(a) shows the ¹²CO(J = 2–1) integrated intensity map of the 10 km s⁻¹ cloud superposed on the NuSTAR Fe Heα flux contours (Yamaguchi et al. 2018). The 12 1' × 1' boxes indicate the regions used for spatially resolved spectral analysis using NuSTAR by Yamaguchi et al. (2018). The authors fitted X-ray spectra above 3 keV using a single-temperature model, because the energy band is dominated by the high-temperature plasma. We note that the CO integrated intensities are significantly changed region to region, which shows an intensity gradient from the southeast to the northwest as mentioned in Section 3.

Zoom In Zoom Out Reset image size

Figure 9(b) shows a scatter plot between kT_e of the high-temperature recombining plasma (Yamaguchi et al. 2018) and the peak integrated intensity of the ¹²CO(J = 2–1) line emission for each box. We find a clear negative correlation between the two. More precisely, the kT_e values in the high-temperature plasma are increasing from the western (cloud-rich) regions to the eastern (cloud-poor) regions. This is consistent with the thermal conduction scenario: rapid electron cooling occurred in cold/dense cloud-rich regions. Note that this finding does not rule out the adiabatic cooling scenario in W49B. In fact, the X-ray spectra from W49B are reproduced by two ejecta components: the low-temperature recombining plasma favors the adiabatic cooling scenario whereas the high-temperature component is likely produced by thermal conduction (Holland-Ashford et al. 2020; Sun & Chen 2020). In other words, both the thermal conduction and adiabatic cooling processes coexist in W49B.

Finally, we discuss the reason why our conclusion—a thermal conduction origin for the high-temperature plasma—is different from those of some previous studies. One of the most important issues is the previous evaluation of the ISM interacting with W49B. Almost all previous studies have used the shocked H₂ distribution as the bulk mass of the ISM. However, as discussed in Section 4.1.4, the shocked H₂ mass is only ∼2% of the CO-traced molecular cloud mass. Because the shocked H₂ map is bright in the southeast, most researchers believe that the southeast shell is interacting with dense molecular clouds and the ISM mass of the east is higher than that of the west. Some previous studies have therefore concluded that the lower plasma temperature in the west was caused by the adiabatic cooling process (e.g., Holland-Ashford et al. 2020; Yamaguchi et al. 2018). By contrast, it is noteworthy that Zhou & Vink (2018) suggested that molecular cloud density is higher in the west than in the east, which is compatible with our ALMA results. We also note there are different interpretations for the n_e t variation in the high-temperature plasma. Yamaguchi et al. (2018) used n_e t as a proxy for electron density, assuming uniform time since heating and a uniform initial temperature. The former contrasts with the results of Holland-Ashford et al. (2020) and Zhou & Vink (2018), who found higher recombination ages in the east than in the west. The positive correlation between n_e t and kT_e in W49B may have to be reconsidered. In any case, we emphasize that proper evaluation of the ISM surrounding an SNR is essential to understanding the origin of recombining plasma correctly. Further detailed comparative studies of CO-based molecular cloud properties and X-ray spectroscopic results are needed to better understand the origin of recombining plasma in SNRs.

It is a hundred-year problem how CRs, mainly comprising relativistic protons, are accelerated in interstellar space. SNRs are believed to be acceleration sites for Galactic CRs below ∼3 PeV through diffusive shock acceleration (e.g., Bell 1978; Blandford & Ostriker 1978). A conventional value of the total energy of CRs accelerated in an SNR is thought to be ∼10⁴⁹–10⁵⁰ erg, corresponding to ∼1%–10% of the typical kinematic energy released by a supernova (10⁵¹ erg; e.g., Gabici 2013; Leahy et al. 2019). One of the foremost challenges is to validate these predictions experimentally.

Gamma-ray and radio-line observations hold a key to understanding the acceleration of CRs in SNRs. Gamma rays from SNRs are produced by two different mechanisms: hadronic and leptonic processes (e.g., Aharonian et al. 1994; Drury et al. 1994). For the hadronic process, CR proton–interstellar proton interaction creates a neutral pion that quickly decays into two gamma-ray photons (hadronic gamma rays). For the leptonic scenario, a CR electron energizes a low-energy photon into gamma-ray energy via inverse Compton scattering, in addition to producing gamma rays through nonthermal bremsstrahlung (leptonic gamma rays). To confirm the acceleration of relativistic protons, the main components of CRs, it is crucial to detect the characteristic spectral feature of hadronic gamma rays with a cutoff at a few GeVs known as a pion-decay bump (e.g., Giuliani et al. 2011; Ackermann et al. 2013). In addition, a good spatial correspondence between gamma rays and interstellar protons provides alternative support for CR proton acceleration (e.g., Fukui et al. 2003, 2012, 2017; Aharonian et al. 2008; Yoshiike et al. 2013; Sano et al. 2019a), because the hadronic gamma-ray flux is proportional to the total energy of CR protons and the number density of interstellar protons. Note that bulk components of interstellar protons consist of both neutral molecular and atomic hydrogen traced by CO and Hi radio lines, respectively. Adopting the number density of targeted interstellar protons, the total energy of CR protons is estimated to be W_p ∼ 10⁴⁸–10⁴⁹ erg toward a dozen gamma-ray SNRs. However, it is still under debate which parameters are important to understand the variety of observed (or in situ) W_p values. To better understand the origins of CR protons and the variety of W_p, we need more samples as well as detailed gamma-ray and radio-line studies for SNRs.

W49B is thought to be one of the CR proton accelerators because of the detection of hadron-dominant gamma rays with a pion-decay bump in it (H.E.S.S. Collaboration et al. 2018). In fact, the best-fit position of GeV gamma rays detected by the Fermi Large Area Telescope is the edge of the SE-shell, where the bright radio continuum, shocked H₂ emission, and dense molecular clouds are located. To obtain the total energy of CR protons in W49B, we first estimated the number density of interstellar protons interacting with the SNR. Using Equations (1) and (2) the averaged number density of interstellar protons in molecular form was estimated to be ∼650 ± 200 cm⁻³ assuming a shell radius of 8 pc and a thickness of 3 pc (e.g., Moffett & Reynolds 1994). The error was derived as the typical uncertainty of the CO-to-H₂ conversion factor of ∼30% (see Bolatto et al. 2013). Additionally, the interstellar protons in atomic form are neglectable in W49B because the derived column density of atomic hydrogen is significantly lower than that of molecular hydrogen (see Brogan & Troland 2001). A similar situation is also seen in other middle-aged SNRs (e.g., W44; Yoshiike et al. 2013). We therefore adopted a number density of interstellar protons n_p of ∼650 ± 200 cm⁻³.

According to H.E.S.S. Collaboration et al. (2018), the total energy of CR protons W_p is written as

$\begin{eqnarray}&&{W}_{{\rm{p}}}\sim 2.0\mbox{--}2.2\times {10}^{49}{\left({n}_{{\rm{p}}}/650\ {\mathrm{cm}}^{-3}\right)}^{-1}\ {\left(d/11\ \mathrm{kpc}\right)}^{-2}\ \ \mathrm{erg},\ \ \ \ \ \end{eqnarray} \tag{ 3 }$

where d is the distance to the SNR. Adopting n_p = 650 cm⁻³ and d = 11 kpc, we then obtained W_p ∼ 2 × 10⁴⁹ erg, corresponding to ∼2% of the typical kinematic energy released by a supernova explosion. Table 2 compares the physical properties of 11 gamma-ray SNRs including W49B. Here, all values of n_p and W_p were derived from CO/H i radio-line observations. We find that W_p in W49B is roughly consistent with that in other gamma-ray-bright SNRs located in our Galaxy or in the Large Magellanic Cloud. In addition, it is noteworthy that young SNRs RX J1713.7–3946, RX J0852.0−4622 (a.k.a. Vela Jr.), and RCW 86 as well as an evolved SNR IC 443 show the lowest values of W_p ∼ 10⁴⁸ erg, while the others hold higher values of W_p ∼ 10⁴⁹ erg.

Table 2. Comparison of Physical Properties of 11 Gamma-Ray SNRs

Name	Distance	Diameter	Age	np	Wp	References
	(kpc)	(pc)	(kyr)	(cm−3)	(1049 erg)
(1)	(2)	(3)	(4)	(5)	(6)	(7)
RX J1713.7−3946	1.0	18	1.6	130	${0.16}_{-0.08}^{+0.07}$	Fukui et al. (2012)
RX J0852.0−4622	0.75 a	24	1.7 a	100	${0.07}_{-0.02}^{+0.02}$	Fukui et al. (2017)
RCW 86	2.5	30	1.8	75	${0.11}_{-0.01}^{+0.01}$	Sano et al. (2019a)
HESS J1731−347	5.7	44	4.0	60	${0.66}_{-0.22}^{+0.22}$	Fukuda et al. (2014)
G39.2−0.3	6.2	14	${5.0}_{-2.0}^{+2.0}$ b	400	${3.2}_{-0.8}^{+1.1}$	de Oña Wilhelmi et al. (2020)
W49B	11.0	16	${6.0}_{-1.0}^{+1.0}$ c	650	${2.1}_{-0.6}^{+1.1}$	This work
Kes 79	5.5	16	${8.3}_{-0.5}^{+0.5}$	360	0.5	Kuriki et al. (2018)
W44	3.0 d	27	20.0 e	200	1.0	Yoshiike et al. (2013)
IC443	1.5 f	20	${25.0}_{-5.0}^{+5.0}$ g	680	0.09	S. Yoshiike et al. (2021, in preparation)
LMC N132D	50.0	25	${2.5}_{-0.2}^{+0.2}$ h	<2000	>0.5	Sano et al. (2020a)
LMC N63A	50.0	18	${3.5}_{-1.5}^{+1.5}$ i	190	${0.9}_{-0.6}^{+0.5}$	Sano et al. (2019b)

Notes. Column (1): Name of SNR. Column (2): Distance to SNR in units of kiloparsecs. Column (3): Diameter of SNR in units of parsecs. Column (4): Age of SNR in units of kiloyears. Column (5): Averaged number density of total interstellar protons n_p in units of cm⁻³. Column (6): Total energy of CR protons W_p in units of 10⁴⁹ erg. Column (7): References for CO/H i derived n_p and W_p for each SNR. Other specific references are also shown as follows:

^a Katsuda et al. (2008). ^b Su et al. (2011). ^c Zhou & Vink (2018). ^d Caswell et al. (1975). ^e Wolszczan et al. (1991). ^f Welsh & Sallmen (2003). ^g Lee et al. (2008); Olbert et al. (2001). ^h Law et al. (2020). ⁱ Hughes et al. (1998).

Download table as:  ASCIITypeset image

To better understand the trend, we plot the W_p values as a function of the age of the SNRs. Figure 10 shows a scatter plot between the ages of the SNRs and W_p. We find a positive correlation between the two parameters in the SNRs with a young age less than ∼6000 yr, suggesting that in situ values of W_p are strongly limited by a short duration time of acceleration, also known as age-limited acceleration (see Ohira et al. 2010). On the other hand, the SNRs with an older age more than ∼8000 yr show a steady decrease of W_p as they get older. This trend could be understood by considering the energy-dependent diffusion of CRs (e.g., Aharonian & Atoyan 1996; Gabici et al. 2007). In other words, the in situ values of W_p have been decreased due to CR escape from the SNR. In fact, hadron-dominant gamma rays have been detected in nearby giant molecular clouds of W44, suggesting that the molecular clouds are illuminated by CR protons that have escaped from W44 (e.g., Uchiyama et al. 2012; Peron et al. 2020). These authors have suggested the actual value of W_p including escaped CRs is ∼10⁵⁰ erg, corresponding to 10% of the typical kinematic energy released by a supernova explosion. In any case, W49B shows one of the highest in situ values of W_p in gamma-ray-bright SNRs, which implies that the escape (diffusion) of CRs is not significant at the moment. Further gamma-ray observations using the Cerenkov Telescope Array will unveil a transition phase from age-limited acceleration to the escape-dominant stage in detail.

Zoom In Zoom Out Reset image size