A&A 615, A150 (2018) Astronomy https://doi.org/10.1051/0004-6361/201731583 & © ESO 2018 Astrophysics

Asymmetric Type-Ia supernova origin of W49B as revealed from spatially resolved X-ray spectroscopic study Ping Zhou1,2 and Jacco Vink1,3

1 Anton Pannekoek Institute, University of Amsterdam, PO Box 94249, 1090 GE Amsterdam, The Netherlands e-mail: p.zhou,[email protected] 2 School of Astronomy and Space Science, Nanjing University, Nanjing 210023, PR China 3 GRAPPA, University of Amsterdam, PO Box 94249, 1090 GE Amsterdam, The Netherlands

Received 17 July 2017 / Accepted 23 March 2018

ABSTRACT

The origin of the asymmetric supernova remnant (SNR) W49B has been a matter of debate: is it produced by a rare jet-driven core- collapse (CC) supernova, or by a normal supernova that is strongly shaped by its dense environment? Aiming to uncover the explosion mechanism and origin of the asymmetric, centrally filled X-ray morphology of W49B, we have performed spatially resolved X-ray spectroscopy and a search for potential point sources. We report new candidate point sources inside W49B. The Chandra X-ray spectra from W49B are well-characterized by two-temperature gas components ( 0.27 keV + 0.6–2.2 keV). The hot component gas shows a large temperature gradient from the northeast to the southwest and is over-ionized∼ in most regions with recombination timescales 11 3 of 1–10 10 cm− s. The Fe element shows strong lateral distribution in the SNR east, while the distribution of Si, S, Ar, Ca is relatively× smooth and nearly axially symmetric. Asymmetric Type-Ia explosion of a Chandrasekhar-mass white dwarf (WD) well- explains the abundance ratios and metal distribution of W49B, whereas a jet-driven explosion and normal CC models fail to describe the abundance ratios and large masses of iron-group elements. A model based on a multi-spot ignition of the WD can explain the observed high MMn/MCr value (0.8–2.2). The bar-like morphology is mainly due to a density enhancement in the center, given the good spatial correlation between gas density and X-ray brightness. The recombination ages and the Sedov age consistently suggest a revised SNR age of 5–6 kyr. This study suggests that despite the presence of candidate point sources projected within the boundary of this SNR, W49B is likely a Type-Ia SNR, which suggests that Type-Ia supernovae can also result in mixed-morphology SNRs. Key words. ISM: individual objects: W49B – ISM: supernova remnants – nuclear reactions, nucleosynthesis, abundances – white dwarfs

1. Introduction and a shell-type morphology in the radio. Initially, it was noted that mixed-morphology SNRs show thermal emission in the The study of supernova remnants (SNRs) provides information interior originating from low-abundant hot gas (Rho & Petre about both the supernova explosions themselves, and about the 1998; Jones et al. 1998). W49B was included in the list, but environments in which the supernova explosions took place. appeared to have enhanced abundances. However, increasingly The environment often carries important information about the more of the originally mixed-morphology SNRs appeared to also supernova progenitor itself, such as whether it formed in a have enhanced abundances in their interiors (Lazendic & Slane star-forming region, and whether the progenitor shaped its own 2006). It is clear that W49B, together with some other metal- environment with a stellar wind. In particular, massive stars are rich cases such as Sgr A East (Sakano et al. 2004; Park et al. known to create large wind-blown bubbles of several tens of 2005) stand out. In most reviews of mixed-morphology SNRs, parsecs in size (Weaver et al. 1977; Chevalier 1999). W49B is listed as a mixed-morphology SNR (Lazendic & Slane The morphology and spectra of SNRs are determined by the 2006; Vink 2012; Zhang et al. 2015; Dubner & Giacani 2015), combined effects of both the intrinsic explosion properties and and the definition of mixed-morphology SNR is in those cases the ambient medium in which they involve. Unfortunately, how- based solely on different radio and X-ray morphology, and the ever, it is sometimes difficult to disentangle the effects of explo- fact that the X-ray emission is thermal in nature. It is thought that sion properties and the environment in which they occurred. mixed-morphology SNRs evolve in denser environments, and There are properties that can be firmly attributed to the explosion since massive stars are associated with molecular cloud environ- properties, but also properties that may be attributed to either the ments, it is usually assumed that these SNRs are remnants of CC explosion characteristics or to the environment. For example, for supernovae. Indeed, some of the mixed-morphology SNRs have young SNRs it is clear that the abundance pattern provides clear associated young pulsars, proving that these SNRs are indeed CC signatures of the type of explosion, with Type-Ia supernovae pro- SNRs (e.g., W44, IC 443, Wolszczan et al. 1991; Olbert et al. ducing more iron-group elements (IGEs), whereas core-collapse 2001). (CC) SNRs are more abundant in oxygen, neon, and magnesium Thermonuclear (or Type-Ia) supernova progenitors are (Hughes et al. 1995; Vink 2012). carbon–oxygen white dwarfs (WDs), which take a longer time Mixed-morphology SNRs are a special class of SNRs char- to evolve (>40 Myr), and, moreover, only explode if they accrete acterized by bright thermal X-ray emission from their center, sufficient matter from a companion star (the single-degenerate

Article published by EDP Sciences A150, page 1 of 14 A&A 615, A150 (2018) scenario; Whelan & Iben 1973), or merge with a companion Although the X-ray spectrum of W49B shows the SNR to be carbon–oxygen WD (the double-degenerate scenario; Webbink very iron-rich, it is usually assumed that it is a CC SNR, like 1984). By the time they explode, their ambient medium does most mixed-morphology SNRs, albeit a peculiar one. Hwang not necessarily contain any information anymore about their et al.(2000) expressed some doubts, suggesting that neither a progenitors. The exact origin of Type-Ia supernovae is still a CC origin, nor a Type-Ia origin could explain the measured source of debate (see reviews Branch et al. 1995; Hillebrandt & abundances. The brightness of the Fe–K lines, but also the Niemeyer 2000; Livio 2000; Wang & Han 2012; Maoz et al. peculiar, jet-like morphology of the ejecta, has been interpreted 2014, and references therein), but also the manner in which the as evidence that W49B is the result of a hypernova explosion WDs explode is uncertain, with models involving deflagration (Keohane et al. 2007). On the other hand, Miceli et al.(2006) (Nomoto et al. 1984), competing with so-called delayed det- compared the observed abundances with yields for hypernova onation (DDT) models (Khokhlov 1991). In general, Type-Ia and supernova nucleosynthesis and found better agreement for SNRs are often to be found in less dense regions of the Galaxy. the abundances of W49B with models with a normal explosion For example, SN 1006 is found high above the Galactic plane energy (1051 erg). More recently, Lopez et al.(2013b), assuming (b = 14.6◦, corresponding to 560 pc at a distance of 2.18 W49B to be a CC SNR, presented evidence that the supernova 0.08 kpc, Winkler et al. 2003). The∼ less disturbed media in which± produced a black hole rather than a neutron star (NS). The rea- they are often found may account for the generally more symmet- son is that they did not find evidence for a cooling NS, similar to ric morphology, as compared to CC SNRs (Lopez et al. 2011). the X-ray point source in Cas A (Tananbaum 1999). On the other hand, the more symmetric morphologies of Type- The study presented here was prompted by the many pecu- Ia SNRs may also be caused by intrinsically more symmetric liarities of W49B. Most notably, we were puzzled by the fact that explosions. black holes are thought to be the end products of the most mas- The idea that Type-Ia progenitors do not shape the super- sive stars (>25 M , e.g., Heger et al. 2003), but W49B seems nova environments has recently been challenged. For example, to be evolving in a cavity of only 5 pc radius (Keohane et al. it is clear that Kepler’s SNR (Vink 2016, for a review), a Type- 2007). In contrast, a progenitor more∼ massive than 25 M is Ia SNR (Kinugasa & Tsunemi 1999; Reynolds et al. 1994), is expected to create a cavity with a radius of at least 20 pc (Chen evolving inside a bow-shock-shaped high-density region caused et al. 2013). by the wind from a progenitor system (Chiotellis et al. 2012). In With our study we therefore tried to investigate, a) whether contrast, the likely Type-Ia SNR RCW 86 (see Gvaramadze et al. a cooling NS may, after all, be present, given that W49B pro- 2017, for a recent paper suggesting a CC origin) seems to evolve vides a spatially non-uniform X-ray background that could hide inside the low-density environment created by a powerful low- a point source, and that the interstellar absorption is relatively 22 2 density wind (Williams et al. 2011; Broersen et al. 2014). Type-Ia high (NH > 10 cm− ; and b) whether W49B is indeed a CC SNR Tycho is suggested to be overrunning a slowly expand- SNR or even a jet-driven CC SNR, as often assumed. ing molecular bubble created by its progenitor’s outflow (Zhou To answer these questions we reanalyzed the archival et al. 2016). The middle-aged SNR G299.2 2.9 is a Type-Ia SNR Chandra data, using a state-of-the-art adaptive binning method showing asymmetries in the ejecta distribution− due to an asym- for spatially resolved X-ray spectroscopy, and we made a new metric explosion and/or a nonuniform surrounding medium (Post search for X-ray point sources. We indeed found a few point et al. 2014; Park et al. 2007). Moreover, some Type-Ia supernovae sources, but our overall conclusion is that the X-ray spectra fit may explode with intrinsic asymmetries (Röpke et al. 2007; better with a Type-Ia origin for W49B, and in particular that the Maeda et al. 2010b), which has been used to interpret the spec- abundance pattern best fits the multi-point ignition DDT models tral evolution diversity observed in Type-Ia supernovae (Maeda of Seitenzahl et al.(2013b). et al. 2010a). Many of the above-mentioned issues of relating SNRs and their environments to the explosion types, come together in the 2. Data and analysis peculiar SNR W49B, which is suggested at a distance of 8– 2.1. Data 11.3 kpc (Radhakrishnan et al. 1972; Brogan & Troland 2001; Chen et al. 2014; Zhu et al. 2014). The X-ray emission from W49B was observed with Chandra in three epochs in 2000 W49B is dominated by emission from the center, which was ini- (obs. ID: 117; PI: Stephen Holt) and 2011 (obs. IDs: 13440 tially attributed to the presence of a pulsar wind nebula (Pye and 13441; PI: Laura Lopez), with exposures of 54, 158, and et al. 1984), but was not much later discredited by the fact 60 ks, respectively. We retrieve three sets of Chandra data, which that the X-ray spectra obtained by the EXOSAT satellite dis- covered the SNR with the backside-illuminated S3 chip in the played bright Fe–K lines (Smith et al. 1985). Due to the centrally faint mode. We use CIAO software (vers. 4.9 and CALDB enhanced X-ray morphology, it was listed in the first cata- vers. 4.7.7)1 to reduce the data, extract spectra, and detect the log of mixed-morphology SNRs, but with a metal-rich interior point-like sources. Xspec (vers. 12.9.0u)2 is used for spectral (Rho & Petre 1998). However, the brightness of the Fe–K lines analysis. and detailed spectroscopy with ASCA (Hwang et al. 2000) sug- gested that W49B is relatively young (1000–4000 yr) compared to most mixed-morphology SNRs, although it does share some 2.2. Spatial-spectral analysis and adaptive binning method characteristics with Sgr A East, a metal-rich, mixed-morphology To optimize the binning of the X-ray data for spatially resolved SNR (Maeda et al. 2002). A property that W49B shares with study, we employ a state-of-the-art adaptive spatial binning many other mixed-morphology SNRs is that the plasma appears method called the weighted Voronoi tessellations (WVT) bin- over-ionized (Kawasaki et al. 2005; Yamaguchi et al. 2009; ning algorithm (Diehl & Statler 2006), which is a generalization Miceli et al. 2010), rather than under-ionized, as in most SNRs. of Cappellari & Copin(2003) Voronoi binning algorithm. The Like many mixed-morphology SNRs, W49B is a GeV gamma- ray source, but it is also a TeV gamma-ray source, which is more 1 http://cxc.harvard.edu/ciao rare for this class of SNRs (H. E. S. S. Collaboration 2018). 2 https://heasarc.gsfc.nasa.gov/xanadu/xspec

A150, page 2 of 14 P. Zhou & J. Vink: Asymmetric Type-Ia supernova origin of W49B

reduced chi-squared NH (1E22 cm^-2) kT (keV) kT_i (keV) Tau (cm^-3 s)

0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 5 5.5 6 6.5 7 7.5 8 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0123456789 10 2E+11 4E+11 6E+11 8E+11 1E+12

Si S Ar Ca Fe

0123456789 10 0123456789 10 0123456789 10 0123456789 10 02468 10 12 14 16 18 20

Fig. 1. Distribution of the parameters fitted with the best-fit single component model, which, for every cell, was taken to be either the absorbed 2 2 “vrnei” or the absorbed “apec” model, whichever produced the smallest χν. First panel: χν overlaid with VLA 20 cm radio contours (Helfand et al. 2006) in white color.

X-ray eventsreduced taken chi-squared from the event file are adaptivelykTc (keV) binned with bekTh distinguished (keV) in 90% ofkT_i the (keV) extracted regions (basedTau (cm^-2 on s) the the WVT binning algorithm. F-test, 2σ level), and for the two-component case in Sect. 2.4, First, we use the observation with longest exposure (obs. the two groups of models are distinguishable in 80% of regions. ID: 13440) to generate the bins, with each bin containing about 3600 counts in 0.3–10 keV. Among the 238 detected bins, 177 2.3. Single thermal component are located inside the SNR, with mean counts per pixel (100) larger than 5. Subsequently, we extract spectra of the 177 regions We jointly fit the spectra in each bin with an absorbed CIE model from three Chandra data sets. The combined data provide 6000 (vapec) and an absorbed recombining plasma model (vrnei) in ∼ 2 counts in each spatial bin. We then perform spectral analysis on Xspec, and select the model with smaller χν as the best-fit model. the 177 bins associated with W49B by jointly fitting three groups The two plasma models use the atomic data in the ATOMDB of0.9 spectra1 1.1 1. from2 1.3 three1.4 1. observations.5 1.6 0 0.1 0.2 For0.3 each0.4 0. spatial5 0.6 0.7 bin0 0. and2 0.4 0. each6 0.8 1 1.2code1.4 1.631.version8 2 2.2 0 3.0.7.0.5 1 1. The5 2 2. Tuebingen–Boulder5 3 3.5 4 4.5 5 2E+11 4E+1 interstellar1 6E+11 8E+1 medium1 1E+12 observation, separateSi response matrix file and ancillaryS response (ISM)Ar absorption model tbabsCais used for calculation ofFe the file are generated. The background spectra are selected from a X-ray absorption due to the gas-phase ISM, the grain-phase ISM, source-free region to the northeast of the SNR, which is at a and the molecules in the ISM (Wilms et al. 2000). The vrnei similar Galactic latitude to the remnant. model describes a plasma that has cooled/heated rapidly from Since previous studies (Kawasaki et al. 2005; Miceli et al. an initial temperature kTi to a temperature of kT, whereas the 2006, 2010; Ozawa et al. 2009; Lopez et al. 2013a,b) indicate that ionization state lags behind, and is characterized by a recombi- the spectrum of W49B is best fit with either a collisional ioniza- nation/ionization timescale τr/τi applied to all ions. When the tion equilibrium (CIE) model or a recombining plasma model ions are in the recombining (overionized) state, kTi and (a non-equilibrium ionization model with over-ionization), or a the “ionization temperatures” of some ions kTz are larger than combination of the two models, we started our analysis by fit- the current electron temperature kT. The lower limit on kTi is set ting012345678 each spatial bin twice,9 10 012345678 once with a CIE and9 once10 012345678 with an to 2 keV9 to10 ensure012345678 that the recombining9 10 0246 model is8 used,10 12 where14 16 1 the8 20 over-ionization model; in the end we selected the model that fits high ionization temperatures of Ar and Ca (kTz = 2.2–2.7 keV; the spectra of a given spatial bin best with the smallest reduced Kawasaki et al. 2005) are also considered. We set the upper limit 2 chi-squared χν; see Sect. 2.3). of kTi to 10 keV, which is a very high value for the plasma in Because this single-component model did not always give an evolved SNR. We allow the abundances of Si, S, Ar, Ca, and satisfactory fits, we decided to also use a two-component model, Fe to vary and tie the abundance of Ni to Fe. Lower-mass ele- combining a relatively cool CIE component with a hotter com- ments, such as O, Ne, and Mg, have X-ray line emission below ponent, which could be either a CIE or an over-ionized model 1.8 keV, in the part of the spectrum that is heavily affected by (Sect. 2.4). The two component model gives better fits to the interstellar absorption. The abundances of these elements are spectra, but a problem is that some of the parameters are therefore unconstrained by the fits, and we fixed the abundances correlated, so some additional constraints had to be imposed. to their solar values. The solar abundances of Asplund et al. One should be aware that the recombining plasma model, (2009) are adopted in plasma emission and photoelectric absorp- 2 with two more free parameters, may provide slightly smaller χν tion models. Compared to the older widely used abundances than the CIE model even though sometimes the two models are obtained by Anders & Grevesse(1989), the O and Fe abundances not statistically distinguished. Nevertheless, we found that for the single-component case discussed in Sect. 2.3 the two models can 3 http://www.atomdb.org/

A150, page 3 of 14 reduced chi-squared NH (1E22 cm^-2) kT (keV) kT_i (keV) Tau (cm^-3 s)

0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 5 5.5 6 6.5 7 7.5 8 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0123456789 10 2E+11 4E+11 6E+11 8E+11 1E+12

Si S Ar Ca Fe

0123456789 10 0123456789 10 0123456789 10 0123456789 10 02468 10 12 14 16 18 20

A&A 615, A150 (2018)

reduced chi-squared kTc (keV) kTh (keV) kT_i (keV) Tau (cm^-2 s)

0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 2E+11 4E+11 6E+11 8E+11 1E+12

Si S Ar Ca Fe

0123456789 10 0123456789 10 0123456789 10 0123456789 10 02468 10 12 14 16 18 20

Fig. 2. Distribution of the parameters fitted with the best-fit double component model, which was either an absorbed “hot vrnei + cool apec” model, 2 or an absorbed “hot vapec + cool apec” model. First panel: χν overlaid with VLA 20 cm radio contours in white color. Last panel: dashed circle and the green cross sign denote the SNR sphere used for density calculation and the sphere center used for abundance – PA diagram in Fig. 10, respectively. P. Zhou & J. Vink: Asymmetric Type-Ia supernova origin of W49B

0.01 0.01 −1 −1 keV keV −1 −1

10−3 10−3 Counts s Counts s

10−4 10−4 4 4

2 2

0 0

−2 −2

(data−model)/error −4 (data−model)/error −4 1 2 5 1 2 5 Energy (keV) Energy (keV)

Fig. 3. Exemplified ACIS-S spectra from one bin in W49B fitted with one component (vapec; left2χ2/d.o. f. = 1.44/308 ) and two components Fig. 3. Exemplified ACIS-S spectra from one bin in W49B fitted with one component (vapec; left: χν/νd.o.f. = 1.44/308) and two components (apec (apec in blue dashed line plus vapec in black dotted line;2 right; χ2/d.o. f. = 1.15/307) models, respectively. The black, red, and green data in blue dashed line plus vapec in black dotted line; right: χν/d.o.f. =ν 1.15/307) models, respectively. The black, red, and green data correspond to correspondthe spectra to from the spectra the observations from the observations 13 440, 13 441, 13440, and 117, 13441 respectively. and 117, respectively.

14 3 in Asplund et al.(flux_h2009 /) flux decreased by 42% andH 32%, density respec- (cool) of 10 cm− s Hwhen density CIE (hot) is the best-fit model),recombination and the age abun- tively. Therefore, adopting different solar abundances(cm^-3) can result dances of Si, S, Ar,(cm^-3) Ca, and Fe. The figure illustrates(yr) a large in differences to the obtained abundances (especially for Fe) and variation of the gas properties inside W49B. The kTi values can absorption column density4. be constrained in only a small fraction of regions, while in the 2 The single thermal component model gives a χν between southwestern regions they run to the upper limit of 10 keV. 0.9 and 1.5 (degree of freedom (d.o.f.) = 222–365; mean 2 χν/d.o.f. = 1.16/291) across the remnant. Figure1 displays 2.4. Two thermal components the spatial distribution of the best-fit parameters, includ-

ing the foreground absorption NH, electron temperature kT, Previous studies suggest the existence of a cooler ISM compo- initial temperature kT (equal to kT when CIE is the best-fit nent ( 0.25 keV; Kawasaki et al. 2005) in addition to the hot, i ∼ model), recombination timescale τr (equal to the upper limit metal-rich component (Hwang et al. 2000; Miceli et al. 2006; 0.7 0.75 0.8 0.85 0.9 0.95 1 0 10 20 30 40 50 60 0Lopez 2 4 et al. 6 2013a 8). 10 Moreover, 12 0 the 2000 infrared 4000 observations 6000 8000 10000 reveal 3 4 For example, using the solar abundances of Anders & Grevesse that a large amount of dense gas ( 500 cm− and 800 M ) ∼ ∼ Fig.(1989 4. )Distribution result in of30% the smaller hot-to-totalNH and flux Feratio abundance (0.75–0.99), for hydrogen the spectra densitiesis shocked in the cold by and W49B hot phases, (Zhu respectively,et al. 2014), and which recombination implies that age of the theshowed hot components, in Fig.3 (for∼ overlaid both of thewith single green and contours two component the Chandra models). X-ray emissionSNR inin 0.3–10 such keV. an inhomogeneous The dashed circle environment denotes the SNR should sphere consist for the of density calculation. A150, page 4 of 14 F-test probability less than the typical value of 0.05 (2σ level), in the range 1–10 1011 cm 3 s. The recombining gas and CIE × − while for the other 8% bins near the western boundary the im- gas can be distinguished by comparing its kTi and kTh or by provement is less significant. checking τr values: the recombining gas shows kTi > kTh and 12 3 The X-ray flux of W49B is dominated by the hot component, τr . 10 cm− s. The CIE region is colored in white in the τr as indicated by the large fraction of the hot-component flux to panel of Figure 2 (upper-right). Our study confirms the existence the total flux in 0.5–8 keV (flux_h/flux=0.78 – 0.99; see the left of recombining plasma in the center and west as pointed out by panel of Figure 4). The X-ray contribution from the cool compo- previous studies (e.g., Miceli et al. 2010; Lopez et al. 2013a). We nent (1-flux_h/flux) is larger near the SNR shell than in the SNR note that the recombining plasma is also patchily distributed in interior. As a result, the single component fit with NH free gives the northern hemisphere of the remnant, but here the recombina- smaller NH values near the SNR edge to compensate the flux of tion timescales are generally much longer than in the southwest. the missing cool component (see the NH distribution in Figure 1).

Therefore, the variation of NH is likely to be much smaller than suggested by the single-component model. 2.5. Density and mass of the shocked gas A temperature gradient for the hot component gas (kTh) is re- W49B has a centrally bright X-ray morphology, indicating an vealed with an orientation northeast to southwest (see Figure 2). enhanced plasma density in the interior, unlike the shell-type The kTh is as high as 2.2 keV in the northeast but decreased SNRs in Sedov phase with a density enhancement at the shells to 0.7 keV in the southwest.∼ The kT shows some variation ∼ c (Borkowski et al. 2001). Since the SNR is highly structured with- across the SNR with a mean value of 0.27 keV. In a small frac- out a clear understanding of its three-dimensional density distri- tion of regions with N deviated largely from 8 1022 cm 2, kT H × − c bution, any complicate hypothesis on the gas distribution (e.g., can be affected by NH due to the degeneracy between them. For bar+shell, disk+shell) could differ from the real distribution and example, a region in the southwest shows large kTc = 0.7 keV, introduce unknown uncertainties. Therefore, we take the sim- which could be reduced to 0.18 keV if N is free ( 1023 cm 2). H ∼ − plest assumption that the plasma is uniform in each bin, so as The recombining plasma appears to be present throughout to provide a mean gas density of the X-ray-emitting gas. We the SNR, except in the southeast shell, and occupies about three note that the following results are based on this oversimplified quarters of the studied area. The recombination timescale varies assumption of the geometry.

Article number, page 5 of 15 P. Zhou & J. Vink: Asymmetric Type-Ia supernova origin of W49B

0.01 0.01 −1 −1 keV keV −1 −1

10−3 10−3 Counts s Counts s

10−4 10−4 4 4

2 2

0 0

−2 −2

(data−model)/error −4 (data−model)/error −4 1 2 5 1 2 5 Energy (keV) Energy (keV)

2 Fig. 3. Exemplified ACIS-S spectra from one bin in W49B fitted with one component (vapec; left χν/d.o. f. = 1.44/308 ) and two components 2 (apec in blue dashed line plus vapec in black dotted line; right; χν/d.o. f. = 1.15/307) models, respectively. The black, red, and green data correspond to the spectra from the observationsP. Zhou & 13440, J. Vink: 13441 Asymmetric and 117, Type-Iarespectively. supernova origin of W49B

flux_h / flux H density (cool) H density (hot) recombination age (cm^-3) (cm^-3) (yr)

0.7 0.75 0.8 0.85 0.9 0.95 1 0 10 20 30 40 50 60 0 2 4 6 8 10 12 0 2000 4000 6000 8000 10000

Fig. 4. Distribution of the hot-to-total flux ratio (0.75–0.99), hydrogen densities in the cold and hot phases, respectively, and recombination age of thethe hot hot components, components, overlaid with with green green contours the Chandra X-rayX-ray emission in 0.3–10 0.3–10 keV. The dashed circle denotes the SNR sphere for the density calculation. more than one density/temperature component. We, therefore The X-ray flux of W49B is dominated by the hot compo- F-test probability less than the typical value of 0.05 (2σ level), in the range 1–10 1011 cm 3 s. The recombining gas and CIE also apply a two-thermal-component model to fit the spectra, in nent, as indicated by the large− fraction of the hot-component while for the other 8% bins near the western boundary the im- gas can be distinguished× by comparing its kT and kT or by an attempt to check whether or not a cool component is needed flux to the total flux in 0.5–8 keV (flux_h/fluxi = 0.78–0.99;h see provement is less significant. checking τ values: the recombining gas shows kT > kT and across the SNR. The colder apec model with solar abundances the left panelr of Fig.4). The X-ray contribution fromi theh cool τ . 1012 cm 3 s. The CIE region is colored in white in the τ is addedThe X-ray to the flux absorbed of W49Bvrnei is dominated/vapec model by the and hot then component, used to componentr (1-flux_h/flux)− is larger near the SNR shell than inr panel of Figure 2 (upper-right). Our study confirms the existence fitas theindicated 177 groups by the of large spectra. fraction The of soft the X-ray hot-component photons . flux2 keV to the SNR interior. As a result, the single component fit with of recombining plasma in the center and west as pointed out by (dominatedthe total flux by in the 0.5–8 cooler keV component) (flux_h/flux= suffer0.78 – heavy 0.99; absorption, see the left N free gives smaller N values near the SNR edge to com- previousH studies (e.g., MiceliH et al. 2010; Lopez et al. 2013a). We whichpanel of results Figure in 4). some The X-ray degeneracy contribution between from the the cool cool compo- pensate the flux of the missing cool component (see the N note that the recombining plasma is also patchily distributed inH nent (1-flux_htemperature/flux)kT is largerand N nearin the the SNR spectral shell fit. than We in therefore the SNR distribution in Fig.1). Therefore, the variation of N is likely c H the northern hemisphere of the remnant, but here the recombina-H assumeinterior. that As a the result, foreground the single absorption component of W49B fit with isN notH free changed gives to be much smaller than suggested by the single-component tion timescales are generally much longer than in the southwest. acrosssmaller theNH smallvalues angular near the extent SNR edgeof the to SNR compensate (<50) and the fix flux the of model. the missing cool component22 2 (see the NH distribution in Figure 1). NH value to 8 10 cm− . This value is close to the best- A temperature gradient for the hot component gas (kTh) is Therefore, the variation× of NH is likely to be much smaller than fit NH for the global spectra, and is also similar to the mean 2.5.revealed Density with and an orientation mass of the northeast shocked to gas southwest (see Fig.2). Nsuggestedvalue ifbyN theis single-component allowed to vary model. for the 177 spectra. The The kT is as high as 2.2 keV in the northeast but decreased to H H h ∼ upperA temperaturelimit for kTc gradientis set to for 0.7 the keV hot to component ensure that gas it ( iskT smallerh) is re- W49B0.7 keV has in a the centrally southwest. bright The X-raykTc shows morphology, some variation indicating across an ∼ thanvealedkT withh. an orientation northeast to southwest (see Figure 2). enhancedthe SNR with plasma a mean density value in of the 0.27 interior, keV. In unlike a small the fraction shell-type of 22 2 TheFigurekTh is2 asshows high the as detailed2.2 keV distribution in the northeast of the gas but properties decreased SNRsregions in with SedovNH phasedeviated with largely a density from enhancement8 10 cm− at, kT thec can shells be ∼ 2 × into W49B0.7 keVbased in on the the southwest. two-component The kT model,c shows including some variationχ , the affected by N due to the degeneracy between them. For exam- ∼ ν (Borkowski etH al. 2001). Since the SNR is highly structured with- temperaturesacross the SNR of thewith colder a mean and value hotter of component0.27 keV. In (kT a smallc and frac-kTh, outple, a a clear region understanding in the southwest of its shows three-dimensional large kTc = 0 density.7 keV, distri-which 22 2 23 2 respectively),tion of regionskT withi, τrN,H anddeviated the abundances largely from of 8 Si,10 S, Ar,cm Ca,− , kT Fec bution,could be any reduced complicate to 0.18 hypothesis keV if NH is on free the ( gas10 distributioncm− ). (e.g., × ∼ elements.can be affected The Si by distributionNH due to the isdegeneracy smoother in between the two-thermal- them. For bar+Theshell, recombining disk+shell) plasma could di appearsffer from to the be real present distribution throughout and componentexample, a regionmodel. in Here the southwestthe upper limitshows of largekTi iskT setc = to0.7 5 keV, introducethe SNR, except unknown in the uncertainties. southeast shell, Therefore, and occupies we take about the three sim- 23 2 which appears could be to reduced be a limit to 0.18to the keV electron if NH temperaturesis free ( 10 foundcm− in). plestquarters assumption of the studied that the area. plasma The recombination is uniform in timescale each bin, varies so as ∼ 11 3 youngThe SNRs recombining (Vink 2012 plasma). We appears note that to the be proton present temperature throughout toin the provide range a 1– mean10 gas10 densitycm− s of. The the recombining X-ray-emitting gas gas.and CIE We × maythe SNR, be higher, except but in the the southeast ionization shell, properties and occupies of the plasmaabout three are notegas can that be the distinguished following results by comparing are based itsonkT thisi and oversimplifiedkTh or by determinedquarters of the by the studied electron area. temperatures. The recombination Moreover, timescale the adopted varies assumptionchecking τr ofvalues: the geometry. the recombining gas shows kTi > kTh and upper limit value is indicated by the best-fit kT values ( 5 keV) τ . 1012 cm 3 s. The CIE region is colored in white in the i ∼ r − in the SNR center. τr panel of Fig.2 (upper-right). Our studyArticle confirms number, page the 5 exis- of 15 Although the overall temperature and abundance patterns tence of recombining plasma in the center and west as pointed obtained with single and double component models appear to out by previous studies (e.g., Miceli et al. 2010; Lopez et al. be similar, the two-thermal-component model better describes 2013a). We note that the recombining plasma is also patchily the spectra than the model with only one thermal component: distributed in the northern hemisphere of the remnant, but here 2 it results in smaller χν values (mean value of 1.06 and d.o.f. the recombination timescales are generally much longer than in of 290), as well as smaller residuals in the spectral fit for the the southwest. Si-XIII line emission for the range 1.8–2.2 keV (Fig.3). The cooler component mainly affects the soft spectra .2 keV and the 2.5. Density and mass of the shocked gas best-fit values of Si (see Fig.3 for an example of the spectra fitted with one component and two components models, respec- W49B has a centrally bright X-ray morphology, indicating an tively). It is statistically meaningful to add the extra thermal enhanced plasma density in the interior, unlike the shell-type component to the single-thermal-component model to improve SNRs in Sedov phase with a density enhancement at the shells the spectral fit of majority regions according to an F-test analysis (Borkowski et al. 2001). Since the SNR is highly structured with- 7 (mean null hypothesis probability 10− , adopting a mean d.o.f. out a clear understanding of its three-dimensional density distri- 2 of 291 and χν = 1.16 in the single thermal model). For 92% of bution, any complicate hypothesis on the gas distribution (e.g., the bins, the two-component model significantly improves the bar + shell, disk + shell) could differ from the real distribution fit with an F-test probability less than the typical value of 0.05 and introduce unknown uncertainties. Therefore, we take the (2σ level), while for the other 8% bins near the western boundary simplest assumption that the plasma is uniform in each bin, so the improvement is less significant. as to provide a mean gas density of the X-ray-emitting gas. We

A150, page 5 of 14 A&A proofs: A&Amanuscript 615, A150 no. 31583_corr_proof (2018)

note that the following results are based on this oversimplified We estimate the mean density nH for a given bin usingassumption the normalization of the geometry. parameter in Xspec (norm = 14We estimate2 the mean density nH for a given bin 10− /(4πd ) nenHdV, where d is the distance, ne and nH are14 −1 using the normalization parameter in Xspec (norm = 10− / the electron and H densities in the volume V; n =1.2n for 0.1 2 e H keV fully(4πd ionized) nenRH plasmadV, where withd solaris the or enhanced distance, metalne and abundancesnH are the −1 inelectron W49B) and and H an densities assumed in prism the volume geometryV; forne = each 1.2n bin.H for Each fully ionizedR plasma with solar or enhanced metal abundances in prism has an area of the region and a depth across the SNR Counts s W49B) and an assumed prism geometry for each bin. Each l(r) = 2 R2 r2, where the radius of the SNR is R = 2.2, 0.01 prism has ans − area of the region and a depth acrosss the SNR0 corresponding2 to2 6.4 pc at a distance of 10 kpc, and r is its pro- jectionl(r) = distance2 pRs tor , the where assumed the radius SNR of center. the SNR The isSNRRs = circle2.02, de- cor- responding to− 6.4 pc at a distance of 10 kpc, and r is its projection 5 noted in Figurep 4 encloses all regions, except a bin in the south- west,distance where to the its densityassumed and SNR mass center. are The not calculated.SNR circle Thedenoted two- in temperatureFig.4 encloses gas allis assumed regions, to except fill the a bin whole in the volume southwest, ( f + f where= 1) 0 its density and mass are not calculated. The two-temperaturec h and in pressure balance (ncTc = nhTh), where f and n are the fill- inggas factor is assumed and hydrogen to fill thedensity, whole respectively, volume ( f andc + thefh = subscripts1) and in (data−model)/error = 5 6 7 “c”pressure and “h” balance denote ( thencTc parametersnhTh), for where the coolf and andn hotare phases, the fill- Energy (keV) respectively.ing factor and hydrogen density, respectively, and the subscripts Fig. 5. The global spectra of W49B in the 4.3–8.0 keV range fitted with “c”The and distributions “h” denote the of the parameters densities for are the displayed cool and in hot Figure phases, 4. Fig. 5. Global spectra of W49B in the 4.3–8.0 keV range fitted with aa recombining recombining plasma plasma model modelvvvvrneirnei. The. The black, black, red, red, and and green green lines lines Therespectively. brightness of the X-ray emission in 0.3–10 keV is overlaid The distributions of the densities are displayed in Fig.4. The correspondcorrespond to to the the spectra spectra from from observations observations 13440, 13 440, 13441 13 and 441, 117, and re- 117, with contours for comparison purposes. The hydrogen density spectively. brightness of the X-ray emission in 0.3–10 keV is overlaid with respectively. nc in the cooler component gas is clearly enhanced along the “bar-like”contours forfeature comparison across the purposes. SNR center The hydrogen and the shells density inn thec in the cooler component gas is clearly enhanced along the “bar- that the hot component consists of almost pure ejecta material, eastern and western sides; it displays a good spatial correlation proton interactions. The high average charge of the atoms and like” feature across the SNR center and the shells in the eastern with almost no hydrogen. In that case, the bremsstrahlung con- with the X-ray brightness. An enhancement of the density in the enhanced ratio of electrons to atom results then in a higher and western sides; it displays a good spatial correlation with tinuum is also dominated by electron-metal interactions rather those regions is also present in the hot component. The mean hy- emissivity per atom, which results in a lower mass for the hot the X-ray brightness. An enhancement of the density in those3 than electron–proton interactions. The high average charge of the drogen density of the colder and hotter components are 24 cm− component (Vink 2012, section 6.1 and 10.3). Abundance ra- regions is3 also present in the hot component. The mean hydro- atoms and the enhanced ratio of electrons to atom results then in and 5 cm− , respectively, and the mean filling factor of the hot-3 tios are not very sensitive to whether the gas is metal-pure, or gen density of the colder and hotter components are 24 cm− a higher emissivity per atom, which results in a lower mass for ter phase is3 40%. The total masses of the X-ray-emitting gas are hydrogen-rich with enhanced metal abundances. Currently with and 5 cm+−41, respectively,2.5 and the mean filling factor of the2.5 hot- the hot component (Vink 2012, Sects. 6.1 and 10.3). Abundance Mc = 484 d M for the cool phase and Mh = 52 8d M CCD X-ray spectroscopy we cannot distinguish between a pure ter phase is34 40%.10 The total masses of the X-ray-emitting10 gas are ratios are not very sensitive to whether the gas is metal-pure, for the hot− phase, where d = d/(10 kpc) is the distance± scaled metal plasma and a hydrogen-rich plasma with enhanced abun- M = 484+41d2.5 M for the10 cool phase and M = 52 8d2.5 M or hydrogen-rich with enhanced metal abundances. Currently to 10c kpc. 34 10 h 10 dances. However, in the future, high-resolution imaging spec- for the hot− phase, where d = d/(10 kpc) is the distance± scaled with CCD X-ray spectroscopy we cannot distinguish between a The mass uncertainties10 are calculated by using the 90% un- troscopy with the XARM (X-ray Astronomy Recovery Mission) to 10 kpc. pure metal plasma and a hydrogen-rich plasma with enhanced certainties of the normalization parameters and the filling fac- and/or Athena could perhaps distinguish these two cases spectro- The mass uncertainties are calculated by using the 90% abundances. However, in the future, high-resolution imaging tors of the 176 bins, where the uncertainties in the filling fac- scopically. Given the age of W49B (see also below) a mixture of uncertainties of the normalization parameters and the filling fac- spectroscopy with the XARM (X-ray Astronomy Recovery Mis- tors are incorporated into the overall error estimates, also tak- hydrogen-rich and metal-rich ejecta seems the most likely sce- tors of the 176 bins, where the uncertainties in the filling factors sion) and/or Athena could perhaps distinguish these two cases ing into account uncertainties in the temperatures and normal- nario. are incorporated into the overall error estimates, also taking spectroscopically. Given the age of W49B (see also below) a izations of the two components. If the errors are asymmetric, we into account uncertainties in the temperatures and normaliza- mixture of hydrogen-rich and metal-rich ejecta seems the most take the largest error. The systematic uncertainty is dominated tions of the two components. If the errors are asymmetric, we likely scenario. by assumptions about the volume V or the depth l of the X-ray 2.6. Abundance and distribution of the ejecta take the largest error. The systematic uncertainty is dominated emitting gas, since the gas mass in bin i weakly depends on them by assumptions about the volume V or the depth l of the X-ray The2.6. abundances Abundance and and distribution distribution of of heavy the ejecta elements have played (M(i) V(i)1/2 l(i)1/2 with given distance and norm). Since emitting∝ gas, since∝ the gas mass in bin i weakly depends on them an important role in probing the explosion mechanism of SNRs. mixed-morphology1/2 SNRs1/2 generally have a relatively smooth ra- The abundances and distribution of heavy elements+0.8 have played+1.1 (M(i) V(i) l(i) with given distance and norm). Since We obtained average abundances [Si]=3.4 0.7, [S]=4.9 1.0, dial density∝ distribution,∝ we assume the X-ray-emitting gas fills an important+1.2 role in probing+1.2 the explosion+1.2 mechanism− of SNRs.− mixed-morphology SNRs generally have a relatively smooth [Ar]=5.1 , [Ca]=5.2 , and [Fe]=5.7 . The+0. average8 abun-+1.1 the whole SNR from front to back. This introduces some uncer- We obtained1.0 average1.0 abundances [Si]1 =.1 3.4 , [S] = 4.9 , radial density distribution, we assume the X-ray-emitting gas fills dance values− here are− based on the weighted− sum0.7 (by the es-1.0 tainty, as the real depth, l(i), can be smaller. In the extreme case [Ar] = 5.1+1.2, [Ca] = 5.2+1.2, and [Fe] = 5.7+−1.2. The average− the whole SNR from front to back. This introduces some uncer- timated gas1 mass.0 for each sub-region)1.0 over all1 individual.1 sub- that the X-ray emission is only arising from a thin shell with a abundance− values here are− based on the weighted− sum (by the tainty, as the real depth,5 l(i), can be smaller.+18 In the extreme case regions, and are insensitive to the emission volume assumption thickness of 1/12RS, the masses are 202 14 M and 21 3M in estimated gas mass for each sub-region) over all individual sub- that the X-ray emission is only arising− from a thin shell± with a as long as the emission volume does not vary sharply at different the cool and hot components,5 respectively,+18 which puts very con- regions, and are insensitive to the emission volume assumption thickness of 1/12RS, the masses are 202 and 21 3 M in the regions. The abundance of Fe is greatly enhanced in the east of servative lower limits on the gas masses in14 W49B. We note that as long as the emission volume does not vary sharply at different cool and hot components, respectively, which− puts± very conser- the SNR, while the Si, S, Ar, Ca elements are highly enhanced the thin shell geometry does not agree with the overall, center- regions. The abundance of Fe is greatly enhanced in the east of vative lower limits on the gas masses in W49B. We note that the across the SNR, especially along the east and west (nearly axial filled, X-ray morphology of W49B. the SNR, while the Si, S, Ar, Ca elements are highly enhanced thin shell geometry does not agree with the overall, center-filled, symmetric distribution; see Section 3.2.2). M is much larger than what could have been produced by across the SNR, especially along the east and west (nearly axial X-rayc morphology of W49B. W49B is the first cosmic source in which Cr and possibly Mn the progenitor wind or the ejecta. Hence, the cool component symmetric distribution; see Sect. 3.2.2). Mc is much larger than what could have been produced by emission were found (Hwang et al. 2000). The global spectra of is dominated by the heated interstellar gas. The mass of the hot W49B is the first cosmic source in which Cr and possibly Mn the progenitor wind or the ejecta. Hence, the cool component is W49B in 4.3–8.0 keV show a clear Cr line at 5.6 keV and a Mn component is also too large to correspond to the total supernova emission were found (Hwang et al. 2000). The global spectra of dominated by the heated interstellar gas. The mass of the hot bump at 6.1 keV (see Figure 5). We fit the∼ global spectra with ejecta mass, which suggests that it consists of a mixture of ejecta W49B in 4.3–8.0 keV show a clear Cr line at 5.6 keV and a component is also too large to correspond to the total super- the vvrnei∼model and obtained [Cr]=6.6 0.8, [Mn]= 12.5 2.7 and circumstellar material. An alternative could be that the hot Mn bump at 6.1 keV (see Fig.5). We fit the global∼ spectra with nova ejecta mass, which suggests that it consists of a mixture (see Figure 5; χ2/d.o. f = 1.74/527; kT±= 1.48+0.07 keV, ±N is component consists of almost pure ejecta material, with almost the vvrnei model∼ ν and obtained [Cr] = 6.6 0.8, [Mn]0.01 = 12.5 H 2.7 of ejecta and circumstellar material. An alternative could be 22 2 − fixed to 8 10χ2/cm. −. ., kT= i . 4/.6 keV, [Fe]= =±.3.2+0.070.1; [Ni] cannot± no hydrogen. In that case, the bremsstrahlung continuum is also (see Fig.×5; ν d o f 1≥74 527; kT 1 48+1±.01.01 keV,11 NH is3 fixed be constrained and thus tied to [Fe]; τ = 6.0 − 10 cm s). dominated by electron-metal interactions rather than electron- to 8 1022 cm 2, kT 4.6 keV, [Fe]r = 3.2 0.06 .1; [Ni] cannot− be The thermal plasma− modeli constrains the− Mn× K line flux to 5 × ≥ ±+1.1 11 3 For a uniform density and a shock compression ratio of four, mass constrained5 and2 thus1 tied to [Fe]; τr = 6.0 . 10 cm− s). 5 1.6 10− cm− s− , similar to that obtained0 by6 × Miceli et al. conservationFor a uniform suggests density that and for a shock shell-type compression SNR the ratio shell of should four, havemass a The× thermal plasma model constrains the− Mn K line flux to conservation suggests that for shell-type SNR the2 shell should have3 a (2006) and Yang5 2 et al.1 (2013). The CIE model gives a slightly thickness of approximately one twelfth R: 4πR2 ∆R(4ρ0) = 4π/3R3 ρ0. 1.6 210− cm− s− , similar to that obtained by+4. 3Miceli et al. thickness of approximately one twelfth R: 4πR ∆R(4ρ0) = 4π/3R ρ0. larger×χν (1.78), with [Cr]=5.9 0.7, [Mn]= 13.9 4.1. Hereafter, ± − A150, page 6 of 14 Article number, page 6 of 15 P. Zhou & J. Vink: Asymmetric Type-Ia supernova origin of W49B

1.5 2424 2323 1.0E+32 7.0E+32 2.7E+34 1.0E+36 erg s-1 erg s-1 erg s-1 erg s-1 9:10:009:10:00 ) 1.0

s 1.5 keV 14

+f 0.7 keV

14 h 09:00 0.5 keV

09:00 )/(f s 0.5 -f h 08:00 7 08:00 13 22 7 6 0.3 keV 13 22 6 0.0 21 12 11 07:00 1021 12 11 07:00 20 10 -0.5 0.2 keV 20 19 5 Declination 06:00 18 4 19 5 Hardness ratio (f 0.1 keV Declination 06:00 3 18 4 -1.0 05:00 3 05:00 17 9 8 -1.5 04:00 -7 -6 -5 -4 -3 17 2 9 8 10 10 10 10 10 16 Photon flux (photon s-1 cm-2) 04:00 2 15 03:00 16 1 15 Fig. 7. Hardness ratios (see the definition in AppendixA)) and and photon photon 03:00 20 15 10 051 19:11:00 55 10:50 fluxes of the 24 point-like sources as a function of the blackbody tem- Right ascension perature and luminosity at a distance of 10 kpc. The curved lines from 32 32 1 36 36 1 1 20 15 10 05 19:11:00 55 10:50 thethe left left to to right right indicate indicate the the luminosity luminosity from from1010ergto s− 10to 10erg serg− . s An− . 22 22 2 2 Anabsorption absorption of N of =NH8= 810 10cm−cmis− assumedis assumed for allfor the all the sources. sources. 0510 Right15 ascension20 25 H × ×

05Fig. 6. Detected point-like10 sources in the15Chandra 0.7–520 keV raw image25 theWe detection derived method the blackbody and analysis temperatures are elaborated and in Appendix luminositiesA. 27 . of W49B. The color-bar shows the scale of the counts per pixel (0005). ofDetailed the sources information based the about photon these fluxes sourcesf and is summarizedhardness ratios in 28 The details of the sources in the ellipses are summarized in Table A.1. Fig. 6. Detected point-like sources in the Chandra 0.7–5 keV raw image (seeTable Fig. A.17.). The luminosities of these sources are in the range 29 ofThe W49B. red regions The color-bar denote the shows sources the with scale best-fit of the blackbody counts per luminosity pixel (000. 5 in). 32.1–36.2 1 32 34 2 1 10 erg s− at an assumed distance of 10 kpc. The details of 30 Thethe range details7. of0 the10 sources–2.7 in10 the ellipsesd10 erg s are− corresponding summarized in to Table the lumi- A.1. × × the detection method and analysis are elaborated in AppendixA. 31 Thenosity red range regions of denote NSs at the the sources age of with 5–6 kyr best-fit predicted blackbody with luminosity the minimal in 3. Discussion cooling paradigm (see32 Appendix34A 2and Fig.18 for details). Detailed information about these sources is summarized in 32 the range 7.0 10 –2.7 10 d10 erg s− corresponding to the lumi- nosity range of× NSs at the× age of 5–6 kyr predicted with the minimal TableHere A.1 we. revisit the progenitor problems and discuss the CC 33 cooling paradigm (see AppendixA and Fig.8 for details). (normal and energetic) and Type-Ia scenarios mainly based (2006) and Yang et al.(2013). The CIE model gives a slightly on three properties: metal abundances, metal distributions, and χ2 . . . +4.3 3. Discussion 34 larger ν (1.78), with [Cr] = 5 9 0 7, [Mn] = 13 9 4.1. Hereafter, environment. The metal abundances can be compared to predic- ± − 1 vv Thewe use thermal the best-fit plasma abundance model constrains results of the the Mnrnei K linemodel. flux The to 35 5 2 1 Heretions ofwe supernova revisit the nucleosynthesis progenitor problems models, and the discuss distribution the CC of 2 1residuals.6 10− atcm around− s− 6.6, similar keV are to likely that obtained due to the bygain Miceli shifts et for al. × (normalmetals reveals and energetic) explosion and (a)symmetries, Type-Ia scenarios and the densitymainly distri-based 36 3 (ACIS,2006) but and theYang shifts et al. are(2013 within). The typical CIE 0.3%model systematic gives a slightly uncer- 37 2 +4.3 onbution three provides properties: information metal abundances, on the circumstellar metal distributions, environment. and 4 largertaintiesχ in(1.78), the ACIS with [Cr] gain = (response5.9 0.7, from[Mn] = the13 CXC.9 . calibration Hereafter, ν 4.1 environment.We will also examine The metal the abundances SNR properties can be and compared discuss tothe predic- origin 38 6 ± − 5 vv wescientists) use the. best-fit The Fe abundance abundance results obtained of the by fittingrnei model. the global The 39 +1.2 tionsof the of bar-like supernova morphology nucleosynthesis according models, to the the spatially distribution resolved of 6 residualsspectra is at smaller around than 6.6 keV the mass-weighted are likely due to abundance the gain shifts (5.7 for), 1.1 metalsspectroscopic reveals analysis. explosion (a)symmetries, and the density distri- 40 − 7 ACIS,since the but Fe the element shifts are and within the plasma typical properties 0.3% systematic vary across uncer- the bution provides information on the circumstellar environment. 41 8 taintiesSNR (see in Fig. the2 ACIS). gain (response from the CXC calibration 42 6 We will also examine the SNR properties and discuss the origin 9 3.1. CC scenario and its problems scientists) . The Fe abundance obtained by fitting the global 43 +1.2 of the bar-like morphology according to the spatially resolved 10 spectra is smaller than the mass-weighted abundance (5.7 ), 1.1 spectroscopic3.1.1. Cavity analysis. 44 2.7. Point-like sources in the vicinity − 11 since the Fe element and the plasma properties vary across the 12 SNRW49B (see has Fig. been2). considered to host a black hole, as no W49B is suggested to be interacting with a molecular cavity 3.1. CC scenario and its problems 45 potential NS was detected down to an X-ray luminosity of surrounding the remnant (Chen et al. 2014; Zhu et al. 2014). 31 1 A massive star can evacuate a hot, low-density bubble with 2.7 10 erg s− , and a core collapse origin was assumed 46 13 2.7. Point-like sources in the vicinity 3.1.1. Cavity (Lopez× et al. 2013b). However, the remnant reveals a very its strong stellar wind during the main sequence stage. The 47 14 W49Bnonuniform has been X-ray considered brightness, to whichhost a providesblack hole, a spatially as no poten- var- W49Bslow and is densesuggested wind to during be interacting its later red with supergiant a molecular stage cavity will, 48 15 tialied NSbackground was detected for point-source down to an detection. X-ray luminosity After considering of 2.7 surroundinghowever, in the most remnant cases not (Chen reach et al. the 2014 main; Zhu sequence et al. 2014 cavity). 31 1 × 49 16 10the brighterg s background, and a core from collapse the SNR origin plasma, was assumed we detected (Lopez 24 Ashell. massive In the star molecular can evacuate environment a hot, low-density with a typical bubble pressure with its of − 5 3 strongp/k stellar10 cm− windK ( duringChevalier the 1999 main; Blitz sequence 1993), stage. the maximum The slow 50 17 etpoint-like al. 2013b sources). However, in, or the in remnant the vicinity reveals of, aW49B very nonuniform using the ∼ 51 18 X-rayMexican-Hat brightness, wavelet which detection provides method a spatially (wav variededetect backgroundin CIAO; andradius dense of the wind bubble duringRb itsis later determined red supergiant by the stageprogenitor will,how- mass ever,M : R inb most1.22 casesM /M not reach9.16 pc the (Chen main et sequence al. 2013). cavity This linearshell. 52 19 forsee point-sourceFig.6). The point detection. spread After function considering (PSF) of the chip bright and back- the ∗ ≈ ∗ − < / 53 20 groundvignetting from effect the SNRare taken plasma, into weaccount. detected The 24 applied point-like significance sources Inrelation the molecular is valid for environment stars with masses with a typical25 M , pressure while beyond of p k the 5 3 ∼ 6 54 21 in,threshold or in the of source vicinity identification of, W49B using (10− the) corresponds Mexican-Hat to one wavelet false 10masscm the− K bubble(Chevalier could 1999 be larger; Blitz due 1993 to), the the contribution maximum radius from ofthe the fast bubble wind inR Wolf–Rayetb is determined stage. by If the the progenitor molecular cavity mass M sur-: 55 22 detection inmethod the image. (wavedetect in CIAO; see Fig.6). The point ∗ Rroundingb 1.22 W49BM /M was9. created16 pc (Chen by the et al.main 2013 sequence). This linear wind rela- of a 56 23 spreadWe function derived (PSF) the blackbody of the chip temperatures and the vignetting and luminosities effect are ≈ ∗ − tionmassive is valid progenitor, for stars a withbubble masses with a<25 radiusM , of while6.04 would beyond sug- the 57 24 takenof the into sources account. based The the applied photon significance fluxes f and threshold hardness of source ratios 6 massgest a the progenitor bubble could mass beof larger13 M due(see to also the contributionChen et al. 2014 from; 58 25 identification(see Fig.7). The (10 luminosities− ) corresponds of these to one sources false are detection in the rangein the ∼ 32.1–36.2 1 theZhang fast et wind al. 2015 in Wolf–Rayet). A smaller stage. bubble If sizethe molecular ( 5 pc) is cavity also likely sur- 59 26 image.10 erg s− at an assumed distance of 10 kpc. The details of ∼ roundingas suggested W49B by Keohane was created et al. by(2007 the), main and sequence indicates awind progen- of a 60 66 http://cxc.harvard.edu/cal/summary/Calibration_ massiveitor mass progenitor,.13 M . aThis bubble is inconsistent with a radius with of 6 the.04 would idea that suggest the 61

Status_Report.html asupernova progenitor explosion mass of resulted13 M (see in the also creation Chen et of al. a2014 black; Zhang hole 62 ∼ Article number,A150, page 7 of 1414 A&A proofs:A&Amanuscript 615, A150 no. 31583_corr_proof(2018)

36 1 assumingmay have anbeen SNR born age with of a 6 high kyr) velocity if W49B (> was400 km caused s− assuming in a CC explosion.an SNR age Therefore, of 6 kyr) even if W49B if W49B was is caused a CC in SNR, a CC there explosion. is still 35 theTherefore, possibility even that if itW49B contains is a a CC cooling SNR, NS, there rather is still than the a possi-black hole,bility as that suggested it contains by Lopez a cooling et al. NS, (2013b). rather than a black hole, as suggested by Lopez et al.(2013b).

) 34 −1 3.1.3.3.1.3. Metal abundances and yields

(erg s 33 ∞ ForFor a a CC CC SNR SNR showing showing super-solar super-solar abundances, abundances, the the progenitor progenitor

Log L mass could be estimated by comparing the abundance ratios with 32 mass could be estimated by comparing the abundance ratios with thosethose predicted predicted by by the the nucleosynthesis nucleosynthesis models, models, since since the the nucle- nucle- osynthesisosynthesis yields yields of of CC CC SNe SNe are are related related to to the the progenitor progenitor mass, mass, 31 stellarstellar metallicity metallicity (solar (solar assumed assumed in in this this study), study), and and explosion explosion energyenergy (e.g., (e.g., NomotoNomoto et et al. al.2006). 2006). Figure Figure 9 showsshows the the compari- compar- 30 sonison of of the the abundances abundances of Si, of Ar,Si, Ca,Ar,Cr, Ca, Mn, Cr, and Mn, Fe and relative Fe relative to Ar 0 1 2 3 4 5 6 7 withto Ar the with predictions the predictions of the spherical of the spherical supernova supernova explosion explosion model Log t (yr) (Sukhboldmodel (Sukhbold et al. 2016), et al. hypernova2016), hypernova model (Nomoto model (Nomoto et al. 2006), et al. Fig. 8. 8.ThePredictions predictions of of NS NS cooling according according to to the the minimal minimal cooling cool- and2006 bipolar), and explosion bipolar explosion model (Maeda model & (Maeda Nomoto & 2003).Nomoto The 2003 ob-). paradigming paradigm (see Page (see etPage al. 2009). et al. The2009 light-). The and light- dark-gray and dark-gray lines represent lines servedThe observed abundance abundance ratios are ratios ascending are ascending with the with atomic the weight atomic therepresent cooling the curves cooling of NSs curves with of light- NSs and with heavy-element light- and heavy-element envelopes, re- fromweight Si fromto Mn. Si We to found Mn. We that found none of that the none available of the nucleosyn- available spectively.envelopes, Therespectively. curves for The intermediate-element curves for intermediate-element envelopes are envelopes between thesisnucleosynthesis models explain models the explain ascending the ascending abundance abundance ratios. The ratios. less them.are between Each envelope them. Each group envelope includes group 100 curves includes corresponding 100 curves to corre- four massiveThe less stars massive (12–13 starsM (12–13) provideM a) flatterprovide abundance a flatter abundance pattern as 3 3 1 1 choicessponding of toP four2 gaps, choices and of fiveP choices2 gaps, and of neutron five choicesS 0 and of neutron proton S 0 functionpattern as of function element mass,of element whereas mass, the whereas very massive the very progenitors massive 1 gaps.and proton The pinkS 0 gaps. belt indicates The pink the belt luminosity indicates the range luminosity of an NS range in an of age an andprogenitors energetic and models energetic produce models sharper produce descending sharper abundance descending ra- rangeNS in of an 5–6 age kyr. range of 5–6 kyr. tioabundance patternsratio (lesspatterns IGEs), which (less IGEs), are more which deviated are more from deviated the ob- servedfrom the ratios. observed ratios. Besides the inconsistent abundance ratios, none of the above- 2013a),(Lopez etwhich al. 2013a is considered), which tois requireconsidered a > to25 requireM progenitor. a >25 M A Besides the inconsistent abundance ratios, none of the above- mentioned CC models explain the observed amount of IGEs. >progenitor.25M progenitor A >25 M insteadprogenitor would instead create a would wind create bubble a in wind the mentioned CC models explain the observed amount of IGEs. The Cr, Mn, and Fe ejecta masses in the hotter-phase gas are molecularbubble in the cloud molecular of 21 cloud pc. of 21 pc. The Cr, Mn, and Fe ejecta masses in the hotter-phase gas are = . . 3 3 = . . 3 3 = ∼ ∼ MMCrCr = 449.9 1 1.1 1010− −M M, M,MnMMn6 =6 61.69 110.9− M10−, MMFe , +0.10 ± × ± × 0.32= .M +±,0. respectively.10 × The nucleosynthesis± models× used in MFe 0.090 32 0.09 M , respectively. The nucleosynthesis models 3.1.2. Missing Missing compact compact object? object? usedthis− studyin this predict− study predict higher higher iron-group iron-group production production for the for stars the 44 starswith with higher higher masses. masses. A 25 A. 252 M.2 Mstarstar produces produces6 6.4.4 1010−− MM A massive single single star star with with MM .. 2020–25–25 M ends its life with 4 × ∗ of Cr, .3.2 104 − M of Mn, and. 0.048 M of Fe in× a spher- a NS (Heger et al. 2003). The∗ NSs, which are born extremely of Cr, 3 2 10− M of Mn, and 0 048 M of Fe in a spherical ex- a NS (Heger et al. 2003). The NSs, which are born extremely ical explosion,× × each of which is about an order of magnitude hot (>101010 K), cool predominantly via neutrino emission from plosion, each of which is about an order of magnitude lower than hot (> 10 K), cool predominantly via neutrino emission from lower than the value obtained from X-ray observations. A 25 M 44 55 the value obtained from X-ray observations. A 25 M under- the interior interior for for 1010 ––1010 yryears after after birth. birth. The The cooling cooling curve curve or 52 ∼∼ goingundergoing energetic energetic (1052 erg) (10 sphericalerg) spherical explosion explosion produces produces slightly orluminosity-age luminosity-age relation relation depends depends sensitively sensitively on on the the equation equation of 4 4 moreslightly Cr, more Mn, and Cr, Fe Mn, (0. and001 FeM (,0 3.001.5 M10−, 3M.5 , and10− 0.007M ,M and, ofstate state of of dense dense matter, matter, NS NS mass, mass, and and envelope composition. × respectively)0.007 M , respectively) compared comparedto normal to explosion× normal explosion energies, energies, but this Page et al.( (2004,2004, 2009 2009)) proposed proposed the the minimal minimal cooling cooling scenario scenario remainsbut this insuremainsfficient insufficient to create to the create observed the observed yields. Similarly, yields. Sim- all including Cooper Cooper pair pair breaking breaking and and formation formation process process to to explain explain ofilarly, the bipolar all of the CC bipolarexplosion CC models explosion predict models much predict smaller much IGE the observed observed luminosities luminosities of of considerably considerably young young NSs. NSs. According Accord- yieldssmaller than IGE the yields observed than values. the observed The bipolar values. energetic The bipolar explosion ener- ingto the to minimal the minimal cooling cooling scenario, scenario, young young NSs radiate NSs radiate X-ray X-ray emis- ofgetic a 25 explosionM star (model of a 25 25A)M star produces (model Cr, 25A) Mn and produces Fe masses Cr, Mn of emissionsion in the in initial the initial few kyr few and kyrs their and temperatures their temperatures and luminosities and lumi- 4 4 and. Fe masses4 of. 7.5 104 − M , 2.3. 10− M , and 0.082 M , are a function of the NS age. 7 5 10− M , 2 3 10− M , and 0 082 M , respectively. nosities are a function of the NS age. respectively.× × × × Figure8 shows the cooling curve (luminosity evolution) of Hence, the CC models fail to explain the observed abundance Figure 8 shows the cooling curve (luminosity evolution) of Hence, the CC models fail to explain the observed abundance NSs based on the minimal cooling paradigm,77 which assumes ratios and under-predict the mass of IGEs. Due to the large abun- NSs based on the minimal cooling paradigm , which assumes ratios and under-predict the mass of IGEs. Due to the large abun- that no enhanced neutrino emission is allowed in NSs (see Page dances of Cr and Mn, this conclusion is unchanged, even taking that no enhanced neutrino emission is allowed in NSs (see Page dances of Cr and Mn, this conclusion is unchanged, even taking et al. 2009). All models are for 1.4 M stars built using the into account the large systematic uncertainties in Mh (> 18 M ; et al. 2009). All models are for 1.4 M stars built using the NS into account the large systematic uncertainties in Mh (>18 M ; NS equation of state of Akmal et al. (1998). The model pre- see Section 2.4). equation of state of Akmal et al. (1998). The model predicts a see Sect. 2.4). = . 32 32 .34 341 1 luminositydicts a luminosity range of rangeL = of 7L.0 107 0 –2.107 –102 7 erg10 s− forerg thes− for the NS at an age∞ of W49B∞ × (5–6× kyr;× see× discussion in NS at an age of W49B (5–6 kyr; see discussion in Section 3.4). 3.2.3.2. Aspherical Aspherical Type-Ia Type-Ia explosion explosion Sect.Nine 3.4). of the 24 detected point-like sources are in the lumi- nosityNine range of the predicted 24 detected for an point-like NS at an sources age of are5–6 in kyr the according luminos- 3.2.1.3.2.1. Metal abundances and yields ity range predicted for an NS at an age of 5–6 kyr according to the to the minimal cooling paradigm (regions labeled in red in Fig- Type-Ia SNe are the dominant factories of IGEs (see a recent ureminimal 6; distance cooling of paradigm 10 kpc is (regions assumed; labeled between in red the in dashed Fig.6; dis- and Type-Ia SNe are the dominant factories of IGEs (see a recent re- tance of 10 kpc is assumed; between the dashed and dot-dashed viewreview of of Seitenzahl Seitenzahl & & Townsley Townsley 2017). 2017). The The large large masses masses of of Cr, Cr, dot-dashed curves in Figure 7). We note that some of the sources Mn, and Fe and the high IGE/intermediate-mass element (IME) maycurves not in be Fig. real7). point We sources,note that but some compact of the clumps sources of may gas associ-not be Mn, and Fe and the high IGE/intermediate-mass element (IME) real point sources, but compact clumps of gas associated with ratioratio in in W49B W49B clearly clearly suggest suggest a a Type-Ia Type-Ia origin. origin. As As shown shown in in the the ated with W49B, while sources outside the SNR boundary would bottom panels of Fig.9, we compare the abundances ratios of the needW49B, a large while transverse sources outside velocity the (& SNR103 boundarykm s 1) towould establish need a bottom panels of Figure 9, we compare the abundances ratios of 3 1 − SNR to the predicted results of different Type-Ia SNe models, connectionlarge transverse with velocity the SNR. (& As10 allkm the s− point) to establishsources are a connection off-center, the SNR to the predicted results of different Type-Ia SNe mod- with the SNR. As all the point sources are off-center, the NS1 including DDT models followed with spherical or extremely the NS may have been born with a high velocity (> 400 km s− els, including DDT models followed with spherical or extremely ooff-centerff-center slow slow flagrations, flagrations, according according to to the the three-dimensional three-dimensional 7 http://www.astroscu.unam. model from Seitenzahl et al.(2013b) and two-dimensional mod- 7 Cooling code from Dany P. Page: model from Seitenzahl et al. (2013b) and two-dimensional mod- mx/neutrones/NSCool/cooling code from Dany P. Page: els from Maeda et al.(2010b). The classical deflagration model http://www.astroscu.unam.mx/neutrones/NSCool/ els from Maeda et al. (2010b). The classical deflagration model A150, page 8 of 14 Article number, page 8 of 15 P. Zhou & J. Vink: Asymmetric Type-Ia supernova origin of W49B

Typical CC (Sukhbold et al.P.P. 2016) Zhou Zhou & & J. J. Vink: Vink:Energetic Asymmetric Asymmetric CC Type-Ia Type-Ia(Nomoto supernova supernova et al. 2006) origin origin of of W49B W49BBipolar CC (Maeda & Nomoto 2003) obs obs obs 3 3 25.2 M • 40 M • 30E 40A: 40 M • 10.9 E θ = 15° Typical OCC (Sukhbold et al. 2016) EnergeticO CC51 (Nomoto et al. 2006) 3Bipolar CC O(Maeda51 θ& = Nomoto 45° 2003) 20.1 M O• 30 MO • 20E51 40B: 40 MO • 1.2 E51 17.0 M • 25 M • 10E 25A: 25 M • 6.7 E θ = 15° obs O obs O 51 obs O 51 θ = 45° 15.2 M O• 20 MO • 10E51 25B: 25 MO • 0.6 E51 3 • 3 • • θ = 15° 25.2 13.0 M MO • 40 MO 30E51 40A: 40 MO 10.9 E51

O • O • O

• O 3 20.1 M • 30 MO • 20E 40B: 40 MO • 1.2 E θ = 45° 2 12.0 MO O• 2 51 51 17.0 M • 25 M • 10E 2 25A: 25 M • 6.7 E θ = 15° O O 51 O 51 θ = 45° 15.2 M O• 20 MO • 10E51 25B: 25 MO • 0.6 E51 13.0 M •

O • O • O • O 12.0 M • 2 O 2 2 [X/Ar]/[X/Ar] [X/Ar]/[X/Ar] [X/Ar]/[X/Ar] 1 1 1 [X/Ar]/[X/Ar] [X/Ar]/[X/Ar] [X/Ar]/[X/Ar] 1 1 1 0 0 0 Si S Ar Ca Cr Mn Fe Si S Ar Ca Cr Mn Fe Si S Ar Ca Cr Mn Fe 0 0 0 Type-Ia (Maeda et al. 2010b) Si S Ar CaType-IaCr Mn(SeitenzahlFe et al. 2013b) Si S Ar Ca Cr Mn Fe Si S Ar Ca Cr Mn Fe 12 obs obs N1 3 C-DDT Type-IaN10 (Seitenzahl et al. 2013b) O-DDTType-Ia (Maeda et al. 2010b) 10 N100 12 obsN1600 obs N1 3 C-DDT • O • O 8 N10 O-DDT 10 N100 2 N1600

6 • O • O 8 2 [X/Ar]/[X/Ar] [X/Ar]/[X/Ar] 4 1 6 [X/Ar]/[X/Ar] [X/Ar]/[X/Ar] 2 4 1 0 0 2 Si S Ar Ca Cr Mn Fe Si S Ar Ca Cr Mn Fe 0 0 Fig. 9. The abundances of Si, Ar, Ca,Si Cr,S Mn,Ar andCa Fe relativeCr Mn ArFe (with error bars) comparedSi S withAr Ca the predictionsCr Mn Fe of supernova nucleosynthesis models. The upper panels show the CC spherical explosion models (left), and energetic spherical (middle) and bipolar explosion models (right; the explosion energy are 10–30 1051 erg, E stands for 1051 erg, θ is the opening half-angle of the jet) for different progenitor masses. Bottom-left Fig.Fig. 9. TheAbundances abundances of Si, of Ar, Si, Ca,Ar, Cr,Ca, Mn, Cr,51 Mn,and Fe and relative Fe relative Ar (with Ar (witherror bars) error compared bars) compared with the with predictions the predictions of supernova of supernova nucleosynthesis nucleosynthesis models. panel shows three-dimensional× DDT Type-Ia models, where the deflagration is ignited in 1, 10, 100, and 1600 spherical sparks, respectively, near models.Upper panels: The upperCC sphericalpanels show explosion the CC models spherical (left explosion), and energetic models spherical (left), and (middle energetic) and spherical bipolar (middle) explosion and models bipolar (right explosion; the explosion models (right; energy the are the WD center.51 Bottom-right panel51 compares the DDT models which followed the two-dimensional spherical deflagration (C-DDT) and extremely explosion10–30 10 energyerg, areE51 10–30stands for101051 erg,ergE, θ isstands the opening for 1051 half-angleerg, θ is the of the opening jet) for half-angle different progenitor of the jet) masses. for differentBottom-left progenitor panel: masses.three-dimensional Bottom-left off-center× deflagration (O-DDT) ignitions,51 respectively. Solar progenitor metallicity is used for all models. panelDDT showsType-Ia three-dimensional models, where× the DDT deflagration Type-Ia models, is ignited where in 1, the 10, deflagration 100, and 1600 is ignited spherical in 1, sparks, 10, 100, respectively, and 1600 spherical near the sparks, WD center. respectively,Bottom-right near thepanel: WDcompares center. Bottom-right the DDT models panel compares which followed the DDT the models two-dimensional which followed spherical the two-dimensional deflagration (C-DDT) spherical and deflagration extremely (C-DDT) off-center and deflagration extremely o(O-DDT)ff-center deflagrationignitions, respectively. (O-DDT) ignitions, Solar progenitor respectively. metallicity Solar is progenitor used for all metallicity models. is used for all models. W7 Nomoto et al. (1984) was excluded since it over-predicts the W7Fe abundance.(Nomoto et al. 1984) was excluded since it over-predicts the 12 Si W7Fe abundance. Nomoto et al. (1984) was excluded since it over-predicts the S Fe abundance.For the models in the bottom-left panel of Fig.9, the defla- Ar 1210 Ca grationFor is the assumed models to in be the ignited bottom-left from different panel of numbers Figure 9, of the sparks defla- SiFe 9 3 S ingration a WD is with assumed a central to be density ignited of from2.9 diff10erentgcm numbers− . The of model sparks Ar N1in ais WD for the with single a central spot ignition density andof 2. N1009× 10 denotes9 g cm the3. The 100-spot model 108 Ca For the models in the bottom-left panel× of Figure− 9, the defla- Fe grationignition.N1 is for is Theassumed the singleoccurrence to spotbe ignited ignition of multi-spot from and di N100fferent ignitions, denotes numbers which the of 100-spot sparks covers 9 3 inaignition. range a WD of with The offset a occurrence central radii,is density a of probable multi-spot of 2.9 consequence10 ignitions,g cm− of. which The the model turbu- covers 86 N1lenta range is convection for theof o singleffset of radii, the spot WD ignition is a prior probable and to the N100× consequence thermonuclear denotes the of 100-spotthe runaway turbu- (Garcia-Senzlent convection & Woosley of the WD 1995 prior; Woosley to the thermonuclear et al. 2004; Iapichino runaway ignition. The occurrence of multi-spot ignitions, which covers Average abundance 64 aet(Garcia-Senz range al. 2006 of). off Theset & radii, Woosleyfewer is sparks a 1995; probable burn Woosley less consequence materials et al. 2004; of to the power Iapichino turbu- the lentexpansionet al. convection 2006). of the The of WD thefewer WD and sparks thus prior theburn to the deflagration less thermonuclear materials is weaker. to powerrunaway The the Average abundance 42 (Garcia-Senzmoderatelyexpansion strongof & the Woosley expansionWD and 1995; thus causes Woosley the a deflagration high et central al. 2004; is density weaker. Iapichino at theThe onsetmoderately of detonation, strong expansion and most ofcauses the remaining a high central fuel density is burned at theto et al. 2006). The fewer sparks burn less materials to power the 0 50 100 150 200 250 300 2 expansionIGEsonset by of detonation. detonation, of the WD Conversely, and and most thus of the a the larger deflagration remaining number fuelis of weaker. ignition is burned The ker- to Position angle (degree) moderatelynelsIGEs produces by detonation. strong relatively expansion Conversely, more causes IMEs a larger a during high number central incomplete of density ignition burning at the ker- 0 50 100 150 200 250 300 (Seitenzahlnels produces et al. relatively 2013b). more IMEs during incomplete burning Fig. 10. TheAverage average abundances abundances as aas function a function of of the the P.A. P.A. (= (=0◦0to to the the onset of detonation, and most of the remaining fuel is burned to Fig. 10. Position angle (degree) IGEs(SeitenzahlThe by abundancedetonation. et al. 2013b). ratios Conversely, in W49B a larger can be number well described of ignition by ker- the north);north; counter-clockwise orientation). The abundance abundance of of each each element element multi-spark ignition models N100 and N1600, except for the ele- isis averaged averaged inside inside the the dashed dashed circle circle denoted denoted in in Fig. Figure4. 4. nels produces relatively more IMEs during incomplete burning Fig. 10. The average abundances as a function of the P.A. (= 0 to the ment Si. Hence, the model with moderate to large numbers of (Seitenzahl et al. 2013b). north; counter-clockwiseh m s orientation). The abundance of each element αJ2000 = 19 11 07.6, δJ2000 = 09◦0601000. 2, which is denoted by ignitionThe sparks abundance reproduces ratios in the W49B abundances can be welland yields described in W49B by the is averaged inside the dashed circle denoted in Figure 4. well, whereas the fewer ignition sparks result in Fe/IME ratios a3.2.2. green Aspherical cross in the explosion bottom-right panel of Fig.2. In particular, multi-spark ignition models N100 and N1600, except for the el- the IGEs, synthesized in the densest part of the exploding WD, thatement areSi. too Hence, high. the model with moderate to large numbers of The ejecta distribution in W49B is not spherical, as shown in The abundance ratios in W49B can be well described by the 3.2.2.showAspherical strong lateral explosion distribution. The abundance of Fe is evi- multi-sparkignition sparks ignition reproduces models N100 the abundances and N1600, and except yields for in the W49B el- the abundance–position angle (P.A.) diagram (Figure 10). Here 3.2.2. Aspherical explosion dently enhanced in the eastern part of W49B, and in the range ementwell, Si.whereas Hence, the the fewer model ignition with moderate sparks result to large in Fe numbers/IME ratios of Thethe ejecta explosion distribution center is in assumed W49B> is at not the spherical, approximate as shown geometric in PA 10–130◦ (meanh [Fe]m s 6). The Si–Ca abundances are ele- ignitionThethat ejecta are sparks too distribution high. reproduces inW49B the abundances is not spherical, and yields as in shown W49B in thevatedcenter abundance–position∼ inαJ2000 a similar= 19 P.A.11 anglerange,07.6, (P.A.)δ butJ2000 also= diagram09 in◦06 a nearly0 (Figure1000.2, opposite which 10). Here is PA de- well,the abundance–position whereas the fewer angleignition (PA) sparks diagram result (Fig. in Fe 10/IME). Here ratios the therange explosion ( 210–300 center◦). The is assumed lateral distribution at the approximate of Fe likely geometric reflects ∼ h m s Article number, page 9 of 15 thatexplosion are too center high. is assumed at the approximate geometric center centerintrinsicαJ2000 asymmetries= 19 11 07 of.6, theδJ2000 SN explosion.= 09◦06010 The00.2, IMEswhich have is de- a

Article number,A150, page page 9 9 of of 15 14 A&A 615, A150 (2018) more axial symmetric morphology, which could be intrinsic to initial metallicity of the progenitor (Badenes et al. 2008; Park the explosion as well, but as discussed in Sect. 3.3 could also et al. 2013), since in these cases mostly the outer layers of the be caused by the structure of the circumstellar medium in which supernovae were assumed to be shocked, and the 55Mn yield W49B is evolving. in the outer layer does depend on the initial metallicity. Carbon A feasible explanation for an asymmetric Type-Ia SN explo- simmering as the cause of the high Mn yield seems unlikely, sion is off-center ignition of a WD (Röpke et al. 2007; Maeda since it only affects the neutron excess of the core. However, a et al. 2010b), which has been used to interpret the spectral evo- high metallicity was suggested in the young SNR N103B based lution diversity observed in Type-Ia SNe (Maeda et al. 2010b). on metal ratios, which seems highly unlikely given that this Maeda et al.(2010a) modeled nucleosynthesis results of an SNR is located in the low-metallicity environment of the Large extremely off-center deflagration of a Chandrasekhar-mass WD Magellanic Cloud (Badenes 2016). In this case, but perhaps also followed with a DDT (model “O-DDT”), and compare them in hindsight for the cases of Tycho’s and Kepler’s SNRs, the with those from a spherically symmetric explosion (model high yields may be affected by simmering, and radial mixing “C-DDT”). As shown in the bottom-right panel of Fig.9, the of material from the core and outer regions (e.g., Gamezo et al. off-center ignition model (29 sparks assumed) better describes 2005) may have affected the 55Mn abundance of outer ejecta the abundance ratios in W49B. Furthermore, the off-center igni- (but see Badenes et al. 2005). Finally, a large Mn/Cr ratio was tion model predicts an offset distribution of stable Fe-peak also found in the Type-Ia SNR candidate 3C 397, which could elements, which appears to explain the lateral Fe distribution. be contributed by the neutron-rich NSE region (Yamaguchi et al. The abundances and yields of two-dimensional simulations of an 2015). On the other hand, a dense, low-carbon WD with a solar- off-center SN explosion are identical to those of the 100 sparks and subsolar-metallicity progenitor can also produce the high ignition model as shown in the bottom-left panel of Fig.9 (see Mn/Cr ratio in 3C 397 (Dave et al. 2017). 3 Seitenzahl et al. 2013b, for more comparison of the two models), We report here a Mn mass of MMn = 6.6 1.9 10− M for and also similar to the nucleosynthesis results of a recent gravita- W49B, based on fitting the X-ray spectra with± the×vrnei model. tional detonation model with a single off-center bubble ignition The mass ratio of Mn and Cr is 1.3 (0.8–2.2). If we would (Seitenzahl et al. 2016). attribute this to the effect of progenitor∼ metallicity, a super-solar +0.14 +10 Although the current study does not pin down a specific metalicity of Z = 0.12 0.07 = 8 4 Z is implied (using the solar model among many Type-Ia models, we have found that W49B ratio of Asplund et al. 2009− ). − may provide a unique and important example of an asymmetric W49B is in a later stage of its SNR evolution than young Type-Ia explosion, which may offer more clues about how WDs SNRs like Tycho’s and Kepler’s SNR, so the Mn/Cr ratios can explode asymmetrically. reported here likely reflect the combination of both core and outer ejecta, that is, from both “normal” freeze-out from NSE 3.2.3. The high Mn abundance and its implications and incomplete silicon burning. In this case, the ratio also depends on the selected initial parameters such as central den- Most of the 55Mn is produced by the decay of 55Co via 55Fe, sity and numbers of flaming bubbles (Seitenzahl et al. 2013b) while the 55Co is mainly synthesized in the incomplete Si- and the explosion mechanisms (e.g., Dave et al. 2017). burning and “normal” freeze-out from nuclear statistical equi- As shown in the bottom-left panel of Fig.9, the three- librium (NSE; Seitenzahl et al. 2013a). The production of the dimensional models with solar-metallicity progenitor and differ- 55 55 neutron-rich element Co (and hence Mn) requires the pres- ent ignition sparks predict a range of MMn/MCr (0.6–2.2), which ence of neutron-rich elements (Badenes et al. 2008; Bravo 2013). overlaps with the mass ratio of 1.3 (0.8–2.2) reported here for The abundance of these elements are influenced by several W49B. Therefore, from the point of view of existing models, effects. First of all, it may be caused by the presence of the ini- and the SNR phase W49B in super-solar metallicity models may tial abundance of the neutron-rich element 14N, in which case not be required to explain our estimated high Mn/Cr mass ratio. a high 55Mn yield may reflect the high initial metallicity of the Moreover, the high Mn/Cr ratio, and the high Mn/Fe ratio of progenitor (Timmes et al. 2003). Using a one-dimensional Type- [Mn/Fe]/[Mn/Fe] of 1.4–3.3 in W49B strongly suggest a WD Ia explosion model (spherical, single ignition), Badenes et al. explosion almost the mass of Chandrasekhar (sub-solar Mn/Fe 0.65 (2008) proposed a relation MMn/MCr = 5.3 Z for estimating ratio for sub-Chandrasekhar cases; Seitenzahl et al. 2013a), the progenitor Z with the Mn and Cr supernova× yields. Secondly, which is best explained by the single-degenerate scenario. in accreting C/O WDs close to the Chandrasekhar limit, the rise in temperatures in the core will result in carbon fusion several 3.2.4. Properties for typing the SNR centuries before the onset of the explosion. During this so-called carbon-simmering phase, weak interactions enhance the neu- Table1 summarizes the main properties for typing an SNR and tron fraction in the core (Piro & Bildsten 2008), resulting in a the results of W49B, in which their importance in reference is higher 55Mn yield of the ensuing Type-Ia supernova. Finally, a reflected in the sequence. high yield of 55Mn from “normal” freeze-out from NSE requires PWN/NS: There is no PWN detected inside W49B, or any that the nucleosynthesis occurs under conditions of high density clear evidence of an associated NS, although the presence of 8 3 (ρ & 2 10 cm− ; Thielemann et al. 1986). Such high den- an NS cannot yet be excluded, as shown by the presence of sity can× be provided by WDs with masses larger than 1.2 M , numerous point sources in and around W49B (Fig.6). An asso- which is inconsistent with the sub-Chandrasekhar WD scenarios ciation of one of them with an NS would strongly suggest a CC (Seitenzahl et al. 2013a). But the high yield of Mn is consistent origin. with the single-degenerate Type-Ia scenario, for which the mass Ejecta: The masses of the IGEs in the SNR are heavily of the exploding WD should be close to the Chandrasekhar limit. overabundant compared to CC nucleosynthesis models. The Even for Chandrasekhar-limit explosions, the Mn/Fe abundance observed abundance ratios instead suggest a Type-Ia origin. ratio is rather sensitive to the explosion conditions. Asymmetry: CC SNRs appear to be statistically more asym- For a few young SNRs, such as Tycho’s and Kepler’s SNRs, metrical than Type-Ia SNRs (Lopez et al. 2011). However, there the high Mn/Cr ratio has been taken as evidence for a high is emerging evidence that a fraction of the Type-Ia SN explosion

A150, page 10 of 14 P. Zhou & J. Vink: Asymmetric Type-Ia supernova origin of W49B

Table 1. Properties for typing SNRs and W49B’s results. Our density maps show that the cold component reveals a high-density distribution that has a similar east-west orienta- Properties W49B Bipolar CC Normal CC Type-Ia tion, which is less prominently present in the hot component. PWN No + + The strong association with the cold component, which is likely ± NS Unclear shocked ambient medium, suggests that the bar-like structure is ± ± ± Ejecta Iron-group rich + most likely the result of a structure in the ambient medium. − − Asymmetry Yes + + However, the X-ray morphology is clearly a combination of ± Wind bubble .6 pc + the metal-rich ejecta distribution and the density distribution. − ± A possible explanation is that W49B evolves inside a more or Notes. The and + symbols under each explosion mechanism indi- − less barrel-shaped cavity, with lower densities in the North and cate negative and positive evidence, respectively, given the properties of South, which agrees well with the shape of the infrared (Keohane W49B listed in the second column. The symbols indicate that some et al. 2007) and radio emission (Moffett et al. 1993). The higher known SNRs (not for all) with similar properties± match the scenario, or that the explosion mechanism cannot be constrained due to some densities in the equatorial direction are reflected in the higher unclear property. densities of the cold component along the bar. The high densi- ties along the bar probably also triggered the early formation of is intrinsically aspherical (Maeda et al. 2010b; Uchida et al. 2013, the reverse shock, and therefore also resulted in the ejecta being for SN 1006). Moreover, the morphology of evolved SNR is sub- shocked at higher densities in the central region of W49B. ject to the shaping by the inhomogeneous environment. Given The intrinsic asymmetry of the ejecta is reflected not in the the nonuniform environment, Type-Ia SNe can also evolve to overall bar-like morphology of the X-ray line emission, but in mixed-morphology SNRs. For W49B, the centrally filled mor- the abundance maps (Fig.2), which indeed reveal not so much a phology is mainly due to density enhancement in the SNR bar-like morphology, but a fan-like region that reveals higher Fe interior (see discussion below). We note that also the candidate abundances in the Eastern part of the bar-like region. Type-Ia SNR 3C 397 (Chen et al. 1999; Safi-Harb et al. 2000; A bipolar explosion jet seems unlikely, since the base of the Yang et al. 2013; Yamaguchi et al. 2014, 2015, the last reference jet should be displaced several thousand years after the explosion suggests a CC origin) is highly aspherical, with many properties (see the numerical simulation in González-Casanova et al. 2014). that are similar to W49B. Instead the bar-like region goes through the center of the W49B, Wind bubble: The presence and size of a wind bubble, which is why this SNR is labeled a mixed-morphology SNR. and/or the stellar environment itself provides clues about the progenitors of SNRs. W49B is suggested to be located inside a 3.4. Recombining gas and SNR age wind-blown bubble, whose relatively small size suggests that the progenitor had a mass of 13 M , or smaller (if the SNR shock As mentioned in Sect. 2.4, the recombining plasma occupies the ∼ majority of the SNR regions of W49B ( 3/4 by area), not only has already transgressed sufficiently into the shell) if it is a CC ∼ SNR. However, Type-Ia SNRs can also be associated with wind in the southwest, but also in the bar-like structure in the SNR bubbles if there is outflow driven by their single-degenerate pro- interior and in the northeast. The recombination age in the north- genitor systems, such as Tycho (Zhou et al. 2016) and the Type-Ia east is larger than in the southwest (see Fig.4). The gas that has candidate RCW 86 (Badenes et al. 2007; Williams et al. 2011; reached CIE is distributed in the southeastern shell and spreads Broersen et al. 2014). Moreover, observations of Type-Ia SNe in patches throughout the northeastern part of the remnant. have shown that a considerable population explodes “promptly” The distribution and origin of the recombining/over-ionized and associated relatively young stellar populations, although the plasma in W49B has been studied in several papers, and rapid delay time of their explosions (40–420 Myr) is still larger than cooling is regarded as the cause (e.g., Ozawa et al. 2009). Ther- that of CC SNe (.40 Myr; Maoz et al. 2012). The presence of mal conduction was initially proposed as the cooling mechanism a small cavity surrounded by a dense ambient medium there- (Kawasaki et al. 2005), but adiabatic expansion is more favored fore either suggests a CC SNR with a low-mass progenitor, or a in the later studies (Miceli et al. 2010; Lopez et al. 2013a). Type-Ia SNR, with a moderately sized wind-blown bubble. Zhou et al.(2011) performed a hydrodynamic simulation of the We favor the Type-Ia origin of W49B after a comparison of W49B evolution in a circumstellar dense cylinder as indicated all direct and indirect properties for SNR typing, although some by the infrared observations. They found that the recombining of its indirect properties can also be explained by a normal CC plasma originates in both cooling processes: (1) the mixing of explosion. Both observations and models indicate that there is hot plasma and cold gas evaporated from the dense circum- some diversity among Type-Ia SNe. In that regard, W49B is an stellar medium; and (2) rapid adiabatic expansion. The density intriguing object revealing some properties different from other enhancement along the bar-like structure (see Fig.2) supports Type-Ia SNRs. Further studies, especially on the details of the the existence of the dense circumstellar matter for process explosion process and its environment, are needed to achieve an (1) to occur. The adiabatic expansion can explain the large-scale integrated understanding of this remnant. recombining plasma extended to the southwestern boundary. The recombination age shown in Fig.4 is derived from the 3.3. “Bar-like” morphology due to a density enhancement recombination timescale and the electron density in the hot phase (tr = τr/ne, where the density is assumed to be a constant with The bar-like X-ray morphology, which has an east-west orien- time). It describes the time elapse starting from the moment tation and is sharper in Fe–K emission, has been attributed to that plasma reached ionization equilibrium, and then began cool- either an asymmetric SN explosion origin (Miceli et al. 2006; ing. Our measurements indicate recombination ages between Lopez et al. 2013b), the density distribution of the ambient 2000 and 6000 yr, based on an over-ionization model. The medium (Miceli et al. 2006) shaped by the progenitor winds true∼ age of∼ the plasma may well differ as the ionization state (Keohane et al. 2007; Miceli et al. 2008; Zhou et al. 2011) or depends on the temperature and density history of the plasma. to two opposite SN jets perpendicular to the bar-like structure But one should also note that over-ionization can only occur once (Bear & Soker 2017). ionization equilibrium has been reached, and the plasma turns

A150, page 11 of 14 A&A 615, A150 (2018) from under-ionized into over-ionized (Kawasaki et al. 2005). The – W49B is evolving in a dense environment. The mean 3 latter would imply that our recombination age measurements densities are 24 and 5 cm− for the cool and hot X-ray even underestimate the true plasma age. Although the connec- components, respectively. The total masses of the X-ray- +41 2.5 tion between recombination age and SNR age is complex, the emitting gas are 484 34d10 M in the cool phase and recombination ages that we find have implications for the age of 2.5 − 52 8d10 M in the hot phase. W49B. – We± obtained the mass-weighted average abundances In previous studies, the age of W49B was estimated to be +0.8 +1.1 +1.2 +1.2 [Si] = 3.4 0.7, [S] = 4.9 1.0, [Ar] = 5.1 1.0, [Ca] = 5.2 1.0, and between 1000 yr (Pye et al. 1984; Lopez et al. 2013b) and −+1.2 − − − ∼ [Fe] = 5.7 1.1. The Cr, Mn, and Fe abundances according 4000 yr (Hwang et al. 2000). The recombination ages mea- − ∼ to a fit to the global spectra of W49B are [Cr] = 6.6 0.8, sured by us suggest that the SNR is older, as the SNR cannot [Mn] = 12.5 2.7, and [Fe] = 3.2 0.1. ± be younger than the oldest plasma it contains, and the recombi- – The element± Fe shows a strong lateral± distribution. The abun- nation age probably underestimates the true age of the plasma. dance of Fe is evidently enhanced in PA range 10–130◦ Moreover, for the plasma to first reach ionization equilibrium, ∼ 12 3 (SNR east). The Si–Ca abundances are also elevated in a a timescale of &10 cm− s (Smith & Hughes 2010) is needed. 3 similar PA range, but also in the nearly opposite P.A. range Even for densities around n 10 cm− , typical for the Western H ≈ ( 210–300◦). The lateral distribution of Fe suggests intrin- region, the associated timescale is &2600 yr. The recombina- sic∼ asymmetries of the SN explosion. The nearly axially tion timescale of 2000 yr for that region should be added to ∼ symmetric distribution of IMEs may also reflect that the that. Therefore, the total plasma age must be of the order of explosion was not sphericially symmetric, but in this case 4000–5000 yr. In the eastern regions, the recombination ages are also the density distribution of the circumstellar medium as high as 6000 yr, or the plasma is in equilibrium (white pixels may play a role (see point 8 below). in Fig.4). Since the density is lower in that region ( n 4 cm 3), H ≈ − – We have found 24 point-like sources in the vicinity of W49B CIE implies plasma ages of 6600 yr, close to maximum recom- 32.1 36.2 2 1 ∼ with luminosities in the range 10 − d10 erg s− . Nine of bination ages we find. The recombination ages therefore suggest them have the luminosities of a cooling NS at the age of an age of at least 5000–6000 yr. W49B (at an assumed distance of 10 kpc). Therefore, even One can also estimate the age based on the Sedov model if W49B is a CC SNR, there is still the possibility that it combined with an estimate of the forward shock velocity based contains a cooling NS, rather than a black hole. on the plasma temperature. Here one has the choice between – None of the CC nucleosynthesis models (spherical explosion taking the hot plasma component or the cool plasma compo- or bipolar explosion) explain the abundance ratios in W49B. nent. However, the hot plasma component seems to be very The iron-group yields predicted by the CC models are insuf- much metal enriched, and shows a lot of temperature varia- ficient to explain the observed masses in the hotter phase: tion. This component is likely associated with plasma shocked 3 3 MCr = 4.9 1.1 10− M , MMn = 6.6 1.9 10− M , by the reverse shock. Associating the cooler component with ±+0.10 × ± × MFe = 0.32 0.09 M . The energetic CC explosion scenario the forward shock, we infer a forward shock velocity of vs = − 1/2 1/2 1 matches even more poorly than the normal CC scenario [16kTc/(3¯µmH)] = 476(kTc/0.27 keV) km s− , where the given the small molecular cavity surrounding the SNR and mean atomic weight µ¯ = 0.61 is taken for fully ionized plasma nine point-like sources in the vicinity of W49B (probably and mH is the hydrogen atomic mass. For a uniform ambient only a projection effect). medium, the velocity corresponds to a Sedov age of tsedov = 1/2 – A DDT Type-Ia model with multi-spot ignition of a 2Rs/(5vs) 5.3(kTc/0.27 keV)− kyr, and an explosion energy Chandrasakhar-mass WD well describes the observed abun- E = 25/(4∼ ξ)(1.4n m )R3v2 1.3 1051n erg (Ostriker & 0 0 H s s ∼ × 10 dance ratios. This model based on solar-metallicity can also McKee 1988), where we adopt an SNR radius Rs = 6.4 pc, explain the high Mn-to-Cr ratio (MMn/MCr = 0.8–2.2) found and ξ = 2.026, with n10 the ambient density of hydrogen atom 3 in W49B. A feasible explanation of the asymmetric Type-Ia in units of 10 cm− . The temperature of the cool component is SN explosion is off-center ignition of a WD. therefore also consistent with an age of 5000–6000 yr, and the – The centrally filled/bar-like morphology of W49B is mainly canonical supernova explosion energy of 1051 erg for the high ∼ due to density enhancement projected to the SNR center density environment in which W49B expands. 3 (nc > 30 cm− ), given the good spatial correlation between the gas density and the X-ray brightness. This suggests that W49B evolves in a barrel-shaped cavity, which also lead the 4. Conclusion ejecta to be shocked at higher densities and projected to the center. The overall morphology, triggered by the ambient We have performed a spatially resolved X-ray study of SNR medium structure, combined with our conclusion that W49B W49B using a state-of-the-art adaptive binning method in order is a Type-Ia SNR, suggests that Type-Ia supernovae can also to uncover its explosion mechanism and the origin of the cen- result in mixed-morphology SNRs. trally filled morphology. An asymmetric Type-Ia explosion is the – The recombination age of the plasma suggests an SNR age most probable explanation for the abundances, yields, and metal of 5–6 kyr, similar to the estimated Sedov age of 5.3 kyr. distribution in W49B. A density enhancement with an east-west ∼ orientation is the main reason for the bar-like X-ray morphology. Acknowledgements. P.Z. acknowledges the support from the NWO Veni Fel- The detailed results are summarized as follows. lowship, grant no. 639.041.647 and NSFC grants 11503008, 11590781, and – The X-ray emission in W49B is well characterized by 11233001. two-temperature gas containing a cool component with kTc 0.27 keV and a hot, ejecta-rich component with kT ∼0.6–2.2 keV. There is a large gradient of kT from the References h ∼ h northeast to the southwest. The detailed distribution of gas H. E. S. S. Collaboration (Abdalla, H., et al.) 2018, A&A, 612, A5 temperature and other physical parameters across the SNR Akmal, A., Pandharipande, V. R., & Ravenhall, D. G. 1998, Phys. Rev. C, 58, are shown (see Figs.2 and4). 1804

A150, page 12 of 14 P. Zhou & J. Vink: Asymmetric Type-Ia supernova origin of W49B

Anders, E., & Grevesse, N. 1989, Geochim. Cosmochim. Acta, 53, 197 Nomoto, K., Thielemann, F.-K., & Yokoi, K. 1984, ApJ, 286, 644 Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009, ARA&A, 47, 481 Nomoto, K., Tominaga, N., Umeda, H., Kobayashi, C., & Maeda, K. 2006, Nucl. Badenes, C. 2016, Supernova Remnants: An Odyssey in Space After Stellar Phys. A, 777, 424 Death, 105 Olbert, C. M., Clearfield, C. R., Williams, N. E., Keohane, J. W., & Frail, D. A. Badenes, C., Borkowski, K. J., & Bravo, E. 2005, ApJ, 624, 198 2001, ApJ, 554, L205 Badenes, C., Hughes, J. P., Bravo, E., & Langer, N. 2007, ApJ, 662, 472 Ostriker, J. P., & McKee, C. F. 1988, Rev. Mod. Phys., 60, 1 Badenes, C., Bravo, E., & Hughes, J. P. 2008, ApJ, 680, L33 Ozawa, M., Koyama, K., Yamaguchi, H., Masai, K., & Tamagawa, T. 2009, ApJ, Bear, E., & Soker, N. 2017, MNRAS, 468, 140 706, L71 Blitz, L. 1993, Protostars and Planets III, eds. E. H. Levy, & J. I. Lunine (Tucson: Page, D., Lattimer, J. M., Prakash, M., & Steiner, A. W. 2004, ApJS, 155, 623 Univ. of. Arizona), 125 Page, D., Lattimer, J. M., Prakash, M., & Steiner, A. W. 2009, ApJ, 707, 1131 Borkowski, K. J., Rho, J., Reynolds, S. P., & Dyer, K. K. 2001, ApJ, 550, 334 Park, S., Muno, M. P., Baganoff, F. K., et al. 2005, ApJ, 631, 964 Branch, D., Livio, M., Yungelson, L. R., Boffi, F. R., & Baron, E. 1995, PASP, Park, S., Slane, P. O., Hughes, J. P., et al. 2007, ApJ, 665, 1173 107, 1019 Park, S., Badenes, C., Mori, K., et al. 2013, ApJ, 767, L10 Bravo, E. 2013, A&A, 550, A24 Pavlov, G. G., Sanwal, D., & Teter, M. A. 2004, in Young Neutron Stars and Broersen, S., Chiotellis, A., Vink, J., & Bamba, A. 2014, MNRAS, 441, 3040 Their Environments, eds. F. Camilo, & B. M. Gaensler, IAU Symp., 218, 239 Brogan, C. L., & Troland, T. H. 2001, ApJ, 550, 799 Piro, A. L., & Bildsten, L. 2008, ApJ, 673, 1009 Cappellari, M., & Copin, Y. 2003, MNRAS, 342, 345 Post, S., Park, S., Badenes, C., et al. 2014, ApJ, 792, L20 Chen, Y., Sun, M., Wang, Z.-R., & Yin, Q. F. 1999, ApJ, 520, 737 Pye, J. P., Becker, R. H., Seward, F. D., & Thomas, N. 1984, MNRAS, 207, Chen, Y., Zhou, P., & Chu, Y.-H. 2013, ApJ, 769, L16 649 Chen, Y., Jiang, B., Zhou, P., et al. 2014, in Supernova Environmental Impacts, Radhakrishnan, V., Goss, W. M., Murray, J. D., & Brooks, J. W. 1972, ApJS, 24, eds. A. Ray, & R. A. McCray, IAU Symp., 296, 170 49 Chevalier, R. A. 1999, ApJ, 511, 798 Reynolds, S. P., Lyutikov, M., Blandford, R. D., & Seward, F. D. 1994, MNRAS, Chiotellis, A., Schure, K. M., & Vink, J. 2012, A&A, 537, A139 271, L1 Dave, P., Kashyap, R., Fisher, R., et al. 2017, ApJ, 841, 58 Rho, J., & Petre, R. 1998, ApJ, 503, L167 Diehl, S., & Statler, T. S. 2006, MNRAS, 368, 497 Röpke, F. K., Hillebrandt, W., Schmidt, W., et al. 2007, ApJ, 668, 1132 Dubner, G., & Giacani, E. 2015, A&ARv, 23, 3 Safi-Harb, S., Petre, R., Arnaud, K. A., et al. 2000, ApJ, 545, 922 Gamezo, V. N., Khokhlov, A. M., & Oran, E. S. 2005, ApJ, 623, 337 Sakano, M., Warwick, R. S., Decourchelle, A., & Predehl, P. 2004, MNRAS, Garcia-Senz, D., & Woosley, S. E. 1995, ApJ, 454, 895 350, 129 González-Casanova, D. F., De Colle, F., Ramirez-Ruiz, E., & Lopez, L. A. 2014, Seitenzahl, I. R., & Townsley, D. M. 2017, ArXiv e-prints [arXiv:1704.00415] ApJ, 781, L26 Seitenzahl, I. R., Cescutti, G., Röpke, F. K., Ruiter, A. J., & Pakmor, R. 2013a, Gvaramadze, V. V., Langer, N., Fossati, L., et al. 2017, Nat. Astron., 1, 0116 A&A, 559, L5 Heger, A., Fryer, C. L., Woosley, S. E., Langer, N., & Hartmann, D. H. 2003, Seitenzahl, I. R., Ciaraldi-Schoolmann, F., Röpke, F. K., et al. 2013b, MNRAS, ApJ, 591, 288 429, 1156 Helfand, D. J., Becker, R. H., White, R. L., Fallon, A., & Tuttle, S. 2006, AJ, 131, Seitenzahl, I. R., Kromer, M., Ohlmann, S. T., et al. 2016, A&A, 592, A57 2525 Smith, R. K., & Hughes, J. P. 2010, ApJ, 718, 583 Hillebrandt, W., & Niemeyer, J. C. 2000, ARA&A, 38, 191 Smith, A., Peacock, A., Jones, L. R., & Pye, J. P. 1985, ApJ, 296, 469 Hughes, J. P., Hayashi, I., & Helfand, D., et al. 1995, ApJ, 444, L81 Sukhbold, T., Ertl, T., Woosley, S. E., Brown, J. M., & Janka, H.-T. 2016, ApJ, Hwang, U., Petre, R., & Hughes, J. P. 2000, ApJ, 532, 970 821, 38 Iapichino, L., Brüggen, M., Hillebrandt, W., & Niemeyer, J. C. 2006, A&A, 450, Tananbaum, H. 1999, IAU Circ., 7246, 1 655 Thielemann, F.-K., Nomoto, K., & Yokoi, K. 1986, A&A, 158, 17 Jones, T. W., Rudnick, L., Jun, B.-I., et al. 1998, PASP, 110, 125 Timmes, F. X., Brown, E. F., & Truran, J. W. 2003, ApJ, 590, L83 Kawasaki, M., Ozaki, M., Nagase, F., Inoue, H., & Petre, R. 2005, ApJ, 631, 935 Uchida, H., Yamaguchi, H., & Koyama, K. 2013, ApJ, 771, 56 Keohane, J. W., Reach, W. T., Rho, J., & Jarrett, T. H. 2007, ApJ, 654, 938 Vink, J. 2012, A&ARv, 20, 49 Khokhlov, A. M. 1991, A&A, 245, L25 Vink, J. 2016, Supernova 1604, Kepler’s Supernova, and its Remnant (Cham: Kinugasa, K., & Tsunemi, H. 1999, PASJ, 51, 239 Springer), Handbook of Supernovae [arXiv:1612.06905] Lazendic, J. S., & Slane, P. O. 2006, ApJ, 647, 350 Wang, B., & Han, Z. 2012, New Astron. Rev., 56, 122 Livio, M. 2000, eds. J. C. Niemeyer, & J. W. Truran, Type Ia Supernovae, Theory Weaver, R., McCray, R., Castor, J., Shapiro, P., & Moore, R. 1977, ApJ, 218, 377 and Cosmology (Cambridge: Cambridge University Press), 33 Webbink, R. F. 1984, ApJ, 277, 355 Lopez, L. A., Ramirez-Ruiz, E., Huppenkothen, D., Badenes, C., & Pooley, D. A. Whelan, J., & Iben, Jr., I. 1973, ApJ, 186, 1007 2011, ApJ, 732, 114 Williams, B. J., Blair, W. P., Blondin, J. M., et al. 2011, ApJ, 741, 96 Lopez, L. A., Pearson, S., Ramirez-Ruiz, E., et al. 2013a, ApJ, 777, 145 Wilms, J., Allen, A., & McCray, R. 2000, ApJ, 542, 914 Lopez, L. A., Ramirez-Ruiz, E., Castro, D., & Pearson, S. 2013b, ApJ, 764, 50 Winkler, P. F., Gupta, G., & Long, K. S. 2003, ApJ, 585, 324 Maeda, K., & Nomoto, K. 2003, ApJ, 598, 1163 Wolszczan, A., Cordes, J. M., & Dewey, R. J. 1991, ApJ, 372, L99 Maeda, Y., Baganoff, F. K., Feigelson, E. D., et al. 2002, ApJ, 570, 671 Woosley, S. E., Wunsch, S., & Kuhlen, M. 2004, ApJ, 607, 921 Maeda, K., Benetti, S., Stritzinger, M., et al. 2010a, Nature, 466, 82 Yamaguchi, H., Ozawa, M., Koyama, K., et al. 2009, ApJ, 705, L6 Maeda, K., Röpke, F. K., Fink, M., et al. 2010b, ApJ, 712, 624 Yamaguchi, H., Badenes, C., Petre, R., et al. 2014, ApJ, 785, L27 Maoz, D., Mannucci, F., & Brandt, T. D. 2012, MNRAS, 426, 3282 Yamaguchi, H., Badenes, C., Foster, A. R., et al. 2015, ApJ, 801, L31 Maoz, D., Mannucci, F., & Nelemans, G. 2014, ARA&A, 52, 107 Yang, X. J., Tsunemi, H., Lu, F. J., et al. 2013, ApJ, 766, 44 Miceli, M., Decourchelle, A., Ballet, J., et al. 2006, A&A, 453, 567 Zhang, G.-Y., Chen, Y., Su, Y., et al. 2015, ApJ, 799, 103 Miceli, M., Decourchelle, A., Ballet, J., et al. 2008, Adv. Space Res., 41, 390 Zhou, X., Miceli, M., Bocchino, F., Orlando, S., & Chen, Y. 2011, MNRAS, 415, Miceli, M., Bocchino, F., Decourchelle, A., Ballet, J., & Reale, F. 2010, A&A, 244 514, L2 Zhou, P., Chen, Y., Zhang, Z.-Y., et al. 2016, ApJ, 826, 34 Moffett, D. A., Goss, W. M., & Reynolds, S. P. 1993, AJ, 106, 1566 Zhu, H., Tian, W. W., & Zuo, P. 2014, ApJ, 793, 95

A150, page 13 of 14 A&A 615, A150 (2018)

Appendix A: Point-like sources in the vicinity of W49B Table A.1. Information of the point-like sources in the vicinity of W49B.

No. Coordinates Counts (hard) Counts (soft) Significance Photon flux f Hardness ratio kTBB log10(LBB)

1 2 ( fH fS) 1 RA Dec Total Net Total Net (σ) (photon s− cm− ) − (keV) ( erg s− ) ( fH+ fS) 1 19:11:05.7 09:02:33.6 75 62.5 191 163.5 29.3 1.8E 06 5.8E 08 0.24 0.02 0.25 34.2 − ± − − ± 2 19:11:14.6 09:03:28.5 17 0.9 126 73.8 11.0 4.5E 07 3.9E 08 0.96 0.08 0.12 36.2 − ± − − ± 3 19:11:21.2 09:05:02.7 26 12.9 226 168.8 17.8 1.2E 06 4.6E 08 0.79 0.03 0.17 35.2 − ± − − ± 4 19:11:06.1 09:05:33.4 166 75.4 90 6.9 5.4 7.4E 07 5.1E 08 0.89 0.05 1.18 32.4 − ± − ± 5 19:10:57.1 09:05:39.5 35 23.7 22 0.4 4.8 2.2E 07 3.5E 08 1.00 0.11 >5 – − − ± − ± 6 19:10:50.5 09:07:24.7 29 22.2 24 13.3 9.1 6.8E 07 8.3E 08 0.44 0.08 0.42 32.8 − ± − ± 7 19:11:04.0 09:07:37.6 155 20.6 338 116.4 9.1 1.0E 06 6.7E 08 0.54 0.06 0.21 34.4 − ± − − ± 8 19:10:57.4 09:04:02.1 1 1.0 35 24.5 5.8 2.2E 07 3.6E 08 1.00 0.14 <0.01 – − − ± − − ± 9 19:11:04.2 09:04:04.0 21 1.0 57 22.9 6.2 1.4E 07 3.6E 08 0.87 0.27 0.15 34.7 − ± − − ± 10 19:11:10.6 09:06:37.3 491 69.3 244 51.2 5.2 9.9E 07 7.0E 08 0.38 0.04 0.40 33.1 − ± − ± 11 19:11:21.8 09:06:43.0 9 5.2 12 4.5 3.7 7.8E 08 2.8E 08 0.30 0.24 0.37 32.1 − ± − ± 12 19:11:07.5 09:06:54.9 113 21.5 87 4.4 3.3 2.1E 07 4.9E 08 1.00 0.17 >5 – − − ± − ± 13 19:11:08.1 09:07:27.0 90 36.1 63 10.4 3.9 4.3E 07 4.9E 08 0.71 0.07 0.60 32.3 − ± − ± 14 19:11:09.9 09:09:04.4 31 10.1 77 53.3 9.3 4.1E 07 4.1E 08 0.52 0.09 0.21 34.0 − ± − − ± 15 19:11:05.3 09:02:53.2 1 0.1 7 5.3 2.3 5.6E 08 3.3E 08 0.95 0.58 0.13 34.9 − ± − − ± 16 19:11:14.1 09:03:08.9 31 21.0 22 1.2 5.5 2.1E 07 3.5E 08 0.93 0.12 1.83 32.1 − ± − ± 17 19:11:18.9 09:04:03.8 10 2.0 54 23.6 5.1 1.5E 07 3.4E 08 1.00 0.23 <0.01 – − − ± − − ± 18 19:11:12.5 09:05:32.4 298 27.5 400 78.6 5.0 7.1E 07 6.5E 08 0.27 0.07 0.25 33.8 − ± − − ± 19 19:10:59.2 09:05:48.6 180 51.1 136 24.7 5.2 6.2E 07 5.4E 08 0.54 0.05 0.47 32.7 − ± − ± 20 19:11:13.0 09:06:09.3 307 15.3 342 60.7 4.7 5.0E 07 6.5E 08 0.41 0.12 0.23 33.9 − ± − − ± 21 19:11:10.1 09:06:56.0 197 12.2 196 53.0 4.5 4.9E 07 6.6E 08 0.45 0.13 0.22 34.0 − ± − − ± 22 19:11:05.9 09:07:30.9 245 90.0 221 52.7 8.5 1.4E 06 7.0E 08 0.48 0.03 0.44 33.1 − ± − ± 23 19:11:07.9 09:10:19.7 1 0.7 19 13.0 3.6 9.7E 08 2.8E 08 1.00 0.25 <0.01 – − − ± − − ± 24 19:11:15.7 09:10:26.3 4 1.9 11 7.9 3.5 8.6E 08 3.3E 08 0.43 0.32 0.22 33.2 − ± − − ±

We detected 24 point-like sources in the 0.7–5 keV energy band scales of 1, √2, and 2 pixels. Adding a larger spatial scale of in the vicinity of W49B using the Mexican-Hat wavelet source 2 √2 would result in a detection of 76 sources, while most of detection tool wavdetect in CIAO (see Fig.6). We combined the these sources are likely clumpy plasma inside the SNR. We three Chandra observations to detect faint sources, where the examined the detected sources in the hard (2.5–5 keV) and soft exposure-weighted PSF map was use to run wavdetect on (0.7–2.5 keV) bands in order to study their spectral proper- the merged dataset8. The energy band is selected to highlight ties. Table A.1 summarizes the coordinates, total (background- the emission of the point-like sources over the SNR background. included) and net (background-subtracted) counts in the hard The compact object is assumed to have an effective temperature and soft bands, photon fluxes f , and hardness ratios HR of the of 0.07–0.5 keV for heavy element atmosphere or 0.11–0.5 keV sources, where f is the net count rate divided by the effec- for Hydrogen atmosphere, with a foreground absorption col- tive area of the camera. The hardness ratio defined as HR = umn density 5.5–8.8 1022 cm 2. Two criteria are considered ( f f )/( f + f ) has a value between 1 and 1, where the × − H − S H S − for the NS temperature range: (1) The observed central compact fH and fS are the photon fluxes in the hard and soft bands, objects show blackbody temperatures in the range 0.2–0.5 keV respectively. (Pavlov et al. 2004) and (2) the minimal paradigm (Page et al. The blackbody temperatures kTBB and luminosities LBB of 2004) predicts a temperature of 0.11 keV for the light elements the point-like sources are estimated using hardness ratio HR envelop and 0.07 keV for the heavy envelop of an NS with an age and photon flux f (see Fig.7). The foreground absorption and of 6000 yr. distance are assumed to be N = 8 1022 cm 2 and 10 kpc, H × − To minimize the contamination from the SNR structures, we respectively, the same as those of W49B. The best-fit kTBB and extracted the most compact sources using the detection spatial LBB are also listed in Table A.1.

8 http://cxc.harvard.edu/ciao/threads/wavdetect_merged

A150, page 14 of 14