Abstract
Despite their importance, a detailed understanding of Type Ia supernovae (SNe Ia) remains elusive. X-ray measurements of the element distributions in supernova remnants (SNRs) offer important clues for understanding the explosion and nucleosynthesis mechanisms for SNe Ia. However, it is challenging to observe the entire ejecta mass in X-rays for young SNRs, because the central ejecta may not have been heated by the reverse shock yet. Here we present over 200 kilosecond Chandra observations of the Type Ia SNR G344.7–0.1, whose age is old enough for the reverse shock to have reached the SNR center, providing an opportunity to investigate the distribution of the entire ejecta mass. We reveal a clear stratification of heavy elements with a centrally peaked distribution of the Fe ejecta surrounded by intermediate-mass elements (IMEs: Si, S, Ar Ca) with an arc-like structure. The centroid energy of the Fe K emission is marginally lower in the central Fe-rich region than in the outer IME-rich regions, suggesting that the Fe ejecta were shock-heated more recently. These results are consistent with the prediction for standard SN Ia models, where the heavier elements are synthesized in the interior of an exploding white dwarf. We find, however, that the peak location of the Fe K emission is slightly offset to the west with respect to the geometric center of the SNR. This apparent asymmetry is likely due to the inhomogeneous density distribution of the ambient medium, consistent with our radio observations of the ambient molecular and neutral gas.
Export citation and abstract BibTeX RIS
1. Introduction
Type Ia supernovae (SNe Ia) are thought to result from thermonuclear explosions of white dwarfs in a binary system. Since the peak luminosity in the optical band is almost uniform among the objects, SNe Ia can be utilized as distance indicators in cosmology (e.g., Perlmutter et al. 1998; Riess et al. 1998). SNe Ia also play an important role as major suppliers of iron, contributing to the chemical enrichment of the Milky Way (e.g., Reddy et al. 2006) and galaxy clusters (e.g., Sato et al. 2007). Despite this importance, however, many fundamental aspects of these explosions remain elusive.
X-ray observations of supernova remnants (SNRs) offer a unique way to address the relevant open questions, as they allow us to measure the composition and distribution of heavy elements that were synthesized during the SN explosion. In fact, recent observational studies of young, luminous Type Ia SNRs, such as Kepler, Tycho, and SN 1006, have provided important insights into the evolution of the explosions of their progenitors (e.g., Decourchelle 2017; Katsuda 2017; Vink 2017, and references therein). In these young historical SNRs, however, the central ejecta might not yet have been heated by the reverse shock and thus are invisible in X-rays. Thus, while some extensive studies have been performed with these young Ia SNRs, our knowledge of how the SN Ia explosion begins or what happens in the deepest layers of the exploding white dwarf is still limited.
The Type Ia SNR G344.7–0.1 is an ideal target for a comprehensive X-ray ejecta study of SN Ia, because its age is significantly older (3–6 kyr: Combi et al. 2010; Giacani et al. 2011) than the historical Ia SNRs, for which the reverse shock has likely reached the SNR center. This SNR was discovered in the radio band with the Molonglo Observatory Synthesis Telescope (MOST) and the Parkes Observatory (Clark et al. 1975). X-rays from this region were first detected by the ASCA Galactic Plane Survey (Sugizaki et al. 2001), and identified as emission originating from an optically thin thermal plasma associated with the SNR (Yamauchi et al. 2005). The following studies with Chandra and XMM-Newton enabled more detailed imaging and spectroscopic analysis, where the abovementioned SNR ionization age was estimated (Combi et al. 2010; Giacani et al. 2011). G344.7–0.1 was initially classified as a core-collapse (CC) SNR, based on its highly asymmetric X-ray morphology (Lopez et al. 2011) and possible association with star-forming regions (Giacani et al. 2011). However, elemental abundances of the SN ejecta were not well constrained in these studies, mainly due to the short exposure, ∼25 ks, for both Chandra and XMM-Newton.
The situation was largely changed by Suzaku observations, where strong, ejecta-dominated Fe K emission was detected (Yamaguchi et al. 2012). Its centroid energy ∼6.46 keV corresponds to a relatively low ionization state (Fe17+), which is proposed to be a common characteristic for Type Ia SNRs (Yamaguchi et al. 2014a). Notably, the atomic processes involved in the Fe K fluorescence emission are theoretically well understood (e.g., Yamaguchi et al. 2014b). Thus, the detection of the Fe K emission from this SNR is a remarkable advantage compared to other Type Ia SNRs (with similar ages of several kyr, e.g., G299.2–2.9: Post et al. 2014; DEM L71: Hughes et al. 2003), where the X-ray spectra are dominated by Fe L-shell emission with highly uncertain atomic physics. Unfortunately, the angular resolution of Suzaku was not good enough to constrain the detailed distribution of the Fe ejecta in G344.7–0.1. This motivated us to carry out deep Chandra observations to reveal the distribution of the entire SN ejecta mass, including Fe, the major nucleosynthesis product of SNe Ia.
It should also be noted that the previous Chandra observation detected a point-like source in the soft X-ray band, named CXOU J170357.8–414302, at the geometrical center of G344.7–0.1 (Combi et al. 2010). Since the absorption column density of this source (NH ≈ 1 × 1022 cm−2) was significantly lower than that of the SNR (5 × 1022 cm−2), it was concluded in their work that the source was a foreground object unassociated with the SNR. However, one may still suspect the possibility of a central compact object (CCO) originating from a CC progenitor, because of the poor photon statistics from the previous observation. Our additional 200 ks Chandra observations also allow us to confirm that this object is unrelated to G344.7–0.1 with better statistics and to search for other point-like sources near the SNR center.
In Section 2, we describe details of our Chandra observations and data reduction. In Section 3, we present our imaging and spectral analysis. We discuss the results in Section 4 and present our conclusions in Section 5. The errors quoted in this work are at 1
2. Observations and Data Reduction
We used the ACIS-I array (Garmire et al. 2003) for our deep observations of G344.7–0.1 to cover the whole X-ray emitting region (∼8' in diameter) within a single field of view (FoV). The ACIS-I has a lower background level than the S3 chip, especially at energies above 5 keV, offering a great advantage for our primary objective: detection and localization of the Fe K emission.
As summarized in Table 1, the observations were split into two series, one in May and the other in 2018 July, with the exposure totaling ∼205 ks. Our quick-look analysis after the first observations series revealed that the Fe K emission partially fell into the gaps between the ACIS-I CCDs, where the exposure is reduced as shown in Figure 1. Therefore, we slightly shifted the aim point for the later observation series such that the Fe-rich regions received the maximum exposure. The difference in the aim points and roll angles between the two series is indicated in Figure 1, with the integrated exposure map overplotted with the radio intensity contours of the SNR.
Table 1. Observation Log
Obs. ID | Date | Exposure Time (ks) | Aim Pointa | Roll Angle |
---|---|---|---|---|
20308 | 2018 May 16 | 29.2 | 25599, −4172 | 5021 |
20309 | 2018 May 12 | 55.3 | 25599, −4172 | 5121 |
21093 | 2018 May 18 | 33.6 | 25599, −4172 | 5021 |
21094 | 2018 May 19 | 14.9 | 25599, −4172 | 5021 |
21095 | 2018 May 20 | 26.7 | 25599, −4172 | 5021 |
21096 | 2018 Jul 3 | 27.2 | 25606, −4171 | 3032 |
21117 | 2018 Jul 5 | 18.8 | 25606, −4171 | 3032 |
Note.
aIn the coordinate.Download table as: ASCIITypeset image
We reprocess the data in accordance with the standard reduction procedures using CIAO version 4.11 (Fruscione et al. 2006) and the calibration database (CALDB) version 4.8.2. For spectral analysis presented in Section 3.2, and Appendix A, we extract ACIS spectra and generated RMFs and ARFs for each individual ObsID, and those taken in the same observation series (i.e., May or July) are summed up to improve the photon statistics. The two merged spectra taken from the same sky regions but different series are jointly fitted with the identical spectral parameters.
3. Analysis and Results
3.1. Imaging Analysis
3.1.1. Extended Emission
Figure 2(a) shows a three-color, exposure-corrected image of G344.7–0.1, where red, green, and blue correspond to emission from the Si K (1.76–1.94 keV), hard continuum (3.0–6.0 keV), and Fe K (6.35–6.55 keV) band, respectively. The image indicates that the softer X-ray emission (red) spreads across the entire SNR, whereas the hard X-ray continuum (green) is relatively faint in the northeast region, implying a temperature gradient from the southwest (high) to the northeast (low). This gradient is more clearly highlighted in Figure 2(b), a map of the mean energy of the detected photons in the broad energy band of 1.5–5.0 keV generated using the mean_energy_map script in the CIAO package.12 Note, however, that this energy band includes several emission lines, and thus a more quantitative spectral analysis (Section 3.2) is required to determine the accurate temperature distribution.
Download figure:
Standard image High-resolution imageFigures 2(c) and (d) are the narrowband images of the Si K and Fe K emission, identical to the red and blue images in Figure 2(a), respectively. It is remarkable that the Fe K line-enhanced region is surrounded by the arc-like structure of the Si K emission. Also notable is that the Fe K peak location is about 2' offset to the west with respect to the geometric center of the SNR determined by the radio band image. In addition to the arc of the Si K emission, a filamentary structure is found at the west rim in the same energy band, indicated with the dashed ellipse in Figure 2(c). All of these features are discovered in this work for the first time, owing to the deep observations with the Chandra ACIS-I.
We generate in Figure 3 radial profiles of the surface brightness in the narrow energy bands that correspond to the K-shell emission of Si (1.76–1.94 keV), S (2.25–2.60 keV), Ar (2.95–3.25 keV), Ca (3.60–4.10 keV), and Fe (6.35–6.55 keV). We confirm that the Fe K emission profile indeed peaks at the interior to other species. We also find that the profiles of the intermediate-mass elements (IMEs) are similar to each other.
Download figure:
Standard image High-resolution image3.1.2. Point-like Sources
Figure 2(e) shows a 0.5–1.2 keV image of the same region as the other panels of Figure 2. Because of the large foreground extinction (NH ∼ 5 × 1022 cm−2: Yamauchi et al. 2005; Combi et al. 2010; Giacani et al. 2011; Yamaguchi et al. 2012), soft X-rays from G344.7–0.1 are almost fully absorbed. Therefore, photons in this energy band are likely dominated by foreground objects. Our simple eyeball inspection detects the two bright point-like sources, labeled No. 1 and 2 in Figure 2(e). Source No. 1 corresponds to CXOU J170357.8–414302, previously reported by Combi et al. (2010). This source is located near the geometric center of the SNR, and does not positionally coincide with the Fe K peak location. The other (No. 2) is newly detected with our deep Chandra observation. We also search for fainter point sources using the wavdetect algorithm in the two energy ranges of 0.5–1.2 keV and 1.2–2.5 keV with the Encircled Counts Fraction (ECF) parameter set to 0.9. Table 2 gives the detected sources with their coordinates, spatial extent relative to the size of the point-spread function (PSFRATIO), and the counts in the 0.5–2.0 keV (soft) and 2.0–9.0 keV (hard) bands. We regard those with PSFRATIO <1 as true point sources. About 50 sources are detected based on these criteria.
Table 2. List of the Detected Point-like Sources
No. | PSFRATIOa | Softb | Hardc | No. | PSFRATIOa | Softb | Hardc | ||
---|---|---|---|---|---|---|---|---|---|
1 | 17:03:57.8, −41:43:02.2 | 0.56 | 146 | 44 | 26 | 17:03:36.8, −41:40:46.5 | 0.35 | 29 | 65 |
2 | 17:04:04.3, −41:41:01.0 | 0.48 | 114 | 38 | 27 | 17:03:54.4, −41:40:19.9 | 0.70 | 31 | 59 |
3 | 17:03:47.5, −41:46:47.1 | 0.77 | 65 | 67 | 28 | 17:04:09.3, −41:40:11.3 | 0.69 | 22 | 29 |
4 | 17:04:04.1, −41:45:36.2 | 0.24 | 24 | 2 | 29 | 17:04:19.3, −41:38:56.3 | 0.84 | 29 | 85 |
5 | 17:04:01.2, −41:43:35.6 | 0.25 | 8 | 2 | 30 | 17:03:55.1, −41:46:33.7 | 0.90 | 56 | 110 |
6 | 17:03:53.4, −41:42:32.2 | 0.34 | 35 | 27 | 31 | 17:03:53.9, −41:41:33.3 | 0.62 | 36 | 80 |
7 | 17:03:45.7, −41:41:56.4 | 0.43 | 46 | 33 | 32 | 17:03:49.8, −41:41:01.8 | 0.81 | 78 | 128 |
8 | 17:04:09.1, −41:39:56.9 | 0.64 | 28 | 31 | 33 | 17:03:38.5, −41:45:15.2 | 0.64 | 41 | 124 |
9 | 17:04:07.0, −41:39:49.6 | 0.72 | 32 | 17 | 34 | 17:03:37.2, −41:44:36.8 | 0.80 | 137 | 306 |
10 | 17:04:11.8, −41:39:31.0 | 0.33 | 17 | 1 | 35 | 17:03:52.7, −41:44:27.8 | 0.51 | 20 | 32 |
11 | 17:03:41.6, −41:46:28.9 | 0.47 | 21 | 39 | 36 | 17:03:35.9, −41:44:22.0 | 0.51 | 49 | 118 |
12 | 17:04:03.0, −41:44:06.2 | 0.28 | 8 | 3 | 37 | 17:03:46.8, −41:44:13.8 | 0.68 | 39 | 109 |
13 | 17:04:11.7, −41:39:53.4 | 0.28 | 11 | 4 | 38 | 17:03:44.7, −41:43:57.6 | 0.37 | 15 | 34 |
14 | 17:04:20.1, −41:39:47.7 | 0.35 | 28 | 13 | 39 | 17:03:34.4, −41:43:20.2 | 0.67 | 80 | 170 |
15 | 17:04:18.5, −41:39:21.0 | 0.38 | 46 | 22 | 40 | 17:03:55.2, −41:43:12.4 | 0.30 | 7 | 7 |
16 | 17:04:16.3, −41:38:37.9 | 0.36 | 24 | 7 | 41 | 17:03:56.1, −41:43:00.6 | 0.38 | 10 | 20 |
17 | 17:04:14.4, −41:38:34.2 | 0.62 | 80 | 70 | 42 | 17:03:43.9, −41:42:43.3 | 0.48 | 25 | 57 |
18 | 17:03:38.8, −41:46:36.1 | 0.45 | 40 | 38 | 43 | 17:03:37.0, −41:42:20.0 | 0.76 | 116 | 191 |
19 | 17:03:36.8, −41:39:40.1 | 0.55 | 35 | 44 | 44 | 17:04:00.3, −41:42:07.1 | 0.51 | 8 | 14 |
20 | 17:03:51.7, −41:44:45.8 | 0.95 | 118 | 187 | 45 | 17:03:46.1, −41:40:52.3 | 0.24 | 14 | 20 |
21 | 17:04:04.7, −41:43:47.9 | 0.39 | 15 | 9 | 46 | 17:03:43.4, −41:40:55.5 | 0.60 | 70 | 130 |
22 | 17:04:02.3, −41:46:23.0 | 0.60 | 39 | 39 | 47 | 17:03:50.4, −41:40:16.8 | 0.90 | 105 | 160 |
23 | 17:04:19.7, −41:42:52.4 | 0.43 | 19 | 10 | 48 | 17:04:11.1, −41:40:01.8 | 0.91 | 51 | 69 |
24 | 17:03:33.9, −41:45:49.8 | 0.60 | 68 | 108 | 49 | 17:03:53.4, −41:38:37.4 | 0.91 | 41 | 130 |
25 | 17:04:14.0, −41:41:28.7 | 0.79 | 22 | 31 | ⋯ | ⋯ | ⋯ | ⋯ | ⋯ |
Notes.
aA parameter of wavdetect that indicates the spatial extension of the sources with respect to the PSF size. The source can be regarded as a true point source when this value is unity or less. bPhoton counts in 0.5–2.0 keV. cPhoton counts in 2.0–9.0 keV.Download table as: ASCIITypeset image
3.2. Spectral Analysis
3.2.1. Extended Emission
Here we analyze ACIS-I spectra extracted from the regions given in Figure 2, where the point sources detected in the previous section (Table 2) are excluded. In our analysis we model the background emission instead of subtracting it. Figure 4 shows the spectra of several representative regions, where black and red are the data from the May and July observations, respectively. The emission lines of the IMEs and Fe are clearly resolved in each region. Background data are extracted from an annulus region surrounding the SNR with inner and outer radii of and , respectively (Figure 1). We model and fit the background spectra simultaneously with the source spectra, instead of subtracting them from the source data, to properly take into account the difference in the effective area between the source and background regions. The details of the background spectral modeling are presented in Appendix A. In the following spectral analysis, we fit unbinned data using the C-statistics (Cash 1979) to estimate the model parameters and their error ranges without bias, although the binned spectra are shown in the figures for clarity.
Download figure:
Standard image High-resolution imageWe first fit the spectra with a model of an optically thin thermal plasma in the non-equilibrium ionization (NEI) state, i.e., the vnei model in the XSPEC package, based on the latest atomic database (AtomDB13 version 3.0.9). We use the tbnew_gas model14 for the foreground absorption, assuming the photoelectric absorption cross sections taken from Verner et al. (1996). The free parameters are the absorption column density NH, electron temperature kTe, ionization timescale net, volume emission measure (VEM), and elemental abundances of Mg, Si, S, Ar, Ca, and Fe relative to the solar abundance table of Wilms et al. (2000). This model gives good fits to the spectra below 5 keV, but fails to reproduce the Fe K emission. We find that the model generally predicts an Fe K centroid energy higher than the observed values, which indicates that the Fe K emission originates from another plasma component with a lower ionization state.
We thus add a Gaussian component to reproduce the K-shell emission from the low-ionized Fe. The addition of this component significantly improves the fit, with the C-stat value reduced from 1003.00/805 to 920.43/803 in the 5.0–8.0 keV band, including the strong Fe K line. The best-fit model spectra and parameters are given in Figure 4 and Table 3, respectively. The abundances of the IMEs (but for Mg) are generally higher than the solar values, implying a significant ejecta contribution to the vnei component. We also find that both the electron temperature (of the vnei component) and the Fe K centroid energy vary among the regions. The lowest temperature is observed in the N and NE, consistent with the mean photon energy map (Figure 2(b)). The Fe K centroid energy is marginally lower in the Fe peak region than in most of the outer regions, indicating that the Fe ejecta in the former has a low ionization degree. The absorption column density is found to be almost uniform throughout the SNR, ∼6.5 × 1022 cm−2. This value is slightly higher than the previous measurements of NH = 4.0–5.5 × 1022 cm−2 (Yamauchi et al. 2005; Combi et al. 2010; Giacani et al. 2011; Yamaguchi et al. 2012). This discrepancy is likely due to the difference in the reference solar abundance; the previous work referred to Anders & Grevesse (1989), while we refer to Wilms et al. (2000). In fact, if we fit the ACIS-I spectra using the solar abundance table of Anders & Grevesse (1989), we obtain NH ∼ 4.5 × 1022 cm−2, consistent with the previous work.
Table 3. Best-fit Spectral Parameters for the Extended Emission
Components | Parameters | Fe | Si N | Si E | Si SW | N |
---|---|---|---|---|---|---|
Absorptiona | ||||||
NEI plasmab | ||||||
Mg | ||||||
Si | ||||||
S | ||||||
Ar | ||||||
Ca | ||||||
VEMc | ||||||
Fe K | E (keV) | |||||
100 (fix) | <93 | 100 (fix) | ||||
Fluxd | ||||||
Photon Counts (104 conut) | 1.75 | 1.83 | 2.05 | 4.22 | 1.24 | |
Background Contribution (%) | 18.4 | 19.1 | 16.4 | 21.4 | 48.8 | |
C-stat | 2658.68 | 2690.62 | 2636.78 | 2739.07 | 2655.14 | |
dof | 2311 | 2312 | 2311 | 2311 | 2312 | |
NE | NW | E | SE | Filament | ||
Absorptiona | ||||||
NEI plasmab | ||||||
Mg | <1.70 | <1.23 | ||||
Si | ||||||
S | ||||||
Ar | ||||||
Ca | e | |||||
VEMc | ||||||
Fe K | E (keV) | ⋯ | e | e | ⋯ | |
⋯ | ⋯ | |||||
Fluxd | ⋯ | ⋯ | ||||
Photon Counts (104 conut) | 3.73 | 0.69 | 1.48 | 1.52 | 0.40 | |
Background Contribution (%) | 35.5 | 45.6 | 46.5 | 40.6 | 35.8 | |
C-stat | 2704.56 | 2694.10 | 2601.55 | 2643.46 | 2671.64 | |
dof | 2311 | 2314 | 2313 | 2312 | 2314 |
Notes.
aThe absorption cross sections are taken from Verner et al. (1996). bSolar abundances of Wilms et al. (2000) are assumed. cIn units of . The volume emission measure (VEM) is given as , where V and D are the volume of the emission region (cm3) and the distance to the emitting source (cm), respectively. dIn units of . eThese are determined using spectra extracted from a larger region containing both E and SE, since the values cannot be constrained for each individual region.Download table as: ASCIITypeset image
Next, we replace the Gaussian component with another vnei component to account for the emission from pure Fe ejecta. The SNR is now divided into three regions, the Fe peak, Si-arc (containing Si E, N, and SW), and rim (containing N, NE, NW, SE, S, and Filament), and their spectra are jointly fitted with a common NH value. Table 4 gives the parameter values obtained for the pure Fe ejecta component. We find that the temperature of this component (kTe > 2.8keV) is significantly hotter than that of the IME component (kTe = 0.8–1.1 keV). We also find that net of the Fe ejecta <2 × 1010 s cm−3 is lower than that of the IME component ∼2 × 1011 s cm−3, and that the lowest net of the Fe ejecta component is achieved at the Fe peak region.
Table 4. Best-fit Spectral Parameters for the Pure Fe Component
Parameters | Fe Peak | Si-arc | Rim |
---|---|---|---|
<0.14 | |||
VEMa |
Note.
aThe VEM is given as .Download table as: ASCIITypeset image
3.2.2. Point-like Sources
We perform spectral analysis of two bright sources (No. 1 and No. 2) detected in the soft X-ray band, both embedded in the diffuse emission from the SNR (Figure 2(e)). The spectra of source No. 1 (CXOU J170357.8–414302) and No. 2 are shown in Figure 5. We fit the spectra of both sources with the following three models: an absorbed power law (tbnew_gas×powerlaw), an absorbed blackbody (tbnew_gas×bbody), and an absorbed optically thin thermal plasma with solar abundances (tbnew_gas×apec). We use background spectra extracted from the annulus surrounding each source region. For source No. 1, we obtain 160 counts for the background-corrected spectrum with 16% background contribution, and 136 counts with 8.6% for No. 2. The best-fit parameters we obtain are given in Table 5. The estimated column densities are significantly lower than that of the SNR ( ≈ 6.5 × 1022 cm−2) for any model assumption, consistent with the foreground stellar object explanation as suggested by Combi et al. (2010) for No. 1. The spectra of the other detected sources (Table 2) with sufficient photon counts are analyzed as well, obtaining statistically acceptable results (C-stat/dof ≲ 1.2) for all the sources. The results are discussed in more detail in Section 4.3.
Download figure:
Standard image High-resolution imageTable 5. Best-fit Spectral Parameters for the Bright Point Sources
No. | a | a | b | a | b | |
---|---|---|---|---|---|---|
1 | <0.08 | <0.06 | ||||
2 | <0.16 |
Notes.
aIn units of . bIn units of keV.Download table as: ASCIITypeset image
4. Discussion
4.1. Stratified Elemental Composition
Thanks to deep observations with Chandra, we have revealed the centrally peaked Fe K emission surrounded by the arc-like structure of the IME emission. This radially stratified chemical composition with Fe at the interior is consistent with the standard picture of SN Ia nucleosynthesis dominated by carbon detonation (e.g., Iwamoto et al. 1999; Seitenzahl et al. 2013). Several other SN Ia explosion models, such as those involving pure carbon deflagration (e.g., Fink et al. 2014) or He-shell detonation (e.g., Sim et al. 2012), predict a substantial amount of 56Ni (which decays into Fe) synthesized outside of the IME. We do not find such evidence in the X-ray data of G344.7–0.1, which may rule out these explosion scenarios as the origin of this SNR. We note that our spectral data below 1 keV, where Fe L lines fall, are missing due to the strong absorption, leaving the possibility that a significant amount of Fe is mixing into the IME-rich layer. Unless this is the case, our results prefer the carbon detonation model for the origin of G344.7–0.1.
We have also found that the peak location of the Fe emission (and thus the centroid of the IME arc) is offset to the west by (corresponding to ∼2.8 pc at the distance of 6 kpc) with respect to the geometric center of the SNR. This implies that either the SN explosion was asymmetric, or the Fe K peak location is the actual explosion center but an apparent asymmetry has formed during the SNR evolution. The spatial distribution of the plasma conditions may help disentangle this degeneracy. In particular, the ionization state of the Fe ejecta that is inferred from the Fe K centroid energy offers a useful diagnostic of the time duration since the ejecta were heated by the reverse shock (e.g., Badenes et al. 2006; Yamaguchi et al. 2014b). Figure 6(a) shows the Fe K centroid energy observed in each region as a function of the angular distance from the location of the Fe emission, which reveals a positive correlation between the two quantities. The mean values of the centroid energy in the Fe peak and Si arc regions correspond to the charge states of Fe10+ and Fe17+, respectively. Similarly, Table 4 indicates that a lower ionization timescale is achieved in the Fe peak region than in the outer regions. This suggests that the Fe ejecta at the inner regions were heated by the reverse shock more recently. We also plot in Figure 6(b) the electron temperatures of the NEI component of each region. In contrast to the Fe K centroid, the temperature is anticorrelated with the distance from the Fe peak location. We thus conclude that the location of the Fe peak region is the true explosion center of this SNR. The origin of the asymmetric morphology is discussed in more detail in Section 4.2.
Download figure:
Standard image High-resolution imageOne caveat is that the variation in the Fe K centroids may also be contributed by the line-of-sight velocity of the Fe ejecta, as was claimed for the Kepler SNR (Kasuga et al. 2018). We find that the observed variation in the Fe K centroid (∼50 eV) requires a line-of-sight velocity vsight of 2400 km s−1 for the Fe ejecta at the Fe peak region. If the Fe ejecta of the Fe peak location have been freely expanding for 3000 yr with vsight ∼2400 km s−1, the ejecta reach 7.4 pc from the SN explosion center, larger than the radius of this SNR (5 pc; assuming a distance of 6 kpc). Given that the ejecta must have been decelerated by the reverse shock, we should take this estimate as a lower limit. Since an unnaturally large asymmetry in the explosion is required to explain this line shift solely with the Doppler effect, the ionization effect is likely dominant in the observed variation in the Fe K centroid energies. The degeneracy between the ionization effect and Doppler shift in the Fe K emission can be disentangled by measurement of the Fe K
Now we estimate the three-dimensional distribution of the vnei component that is dominated by the IME ejecta, assuming that the Fe peak region is the true explosion center. Since the surface-brightness profiles of Si, S, Ar, and Ca (Figure 3) are virtually identical with one another, we use a radial profile of the 1.5–5.0 keV emission as representative of the IME distribution. Figure 7(a) shows the extracted radial profile compared with a simple two-zone spherical model assuming a uniform density and uniform abundance distribution in each zone. We also take into account the temperature gradient, and thus the emissivity variation, between the two zones, where ∼1.1 keV at <1' and ∼0.9 keV at 1'–3'. Figure 7(b) shows the continuum-subtracted surface brightness of the Fe K and 1.5–5.0 keV emission, suggesting an enhanced density of Fe ejecta at <1'. We find that the observed profile is well reproduced by a model consisting of the model schematized in Figure 7(c): a low-density (nin) inner sphere with a radius of 1' and a thick high-density (nshell = 2.6nin) shell with inner and outer radii of 1' and 3', respectively. The IME arc structure found in Figure 2(c) corresponds to the inner part of the thick shell, where the highest surface brightness is achieved. Note that the angular distance between the explosion center and the west rim is ∼3', thus the excess in the brightness beyond this radius (Figure 7(a)) is solely due to the emission extended to the northeast.
Download figure:
Standard image High-resolution image4.2. Asymmetry
As discussed in the previous section, the identified explosion center is offset from the geometric center of the SNR. Therefore, the X-ray morphology of G344.7–0.1 is relatively asymmetric. Lopez et al. (2011) classified this SNR as a possible CC SNR because of such highly asymmetric X-ray morphology, based on the analysis of the power-ratio method (PRM). However, the data set used in Lopez et al. (2011) has more limited photon statistics than our 200 ks Chandra data. Thus, we apply the PRM on our new data set of G344.7–0.1 to measure the asymmetry of this SNR. We replicate the analysis from Lopez et al. (2011), where the PRs are calculated on the 0.5–2.1 keV band exposure-corrected image of SNR, and obtain , .15
The values are now comparable with those for the other well-established Type Ia SNRs, such as Kepler. We note that the surface-brightness-weighted geometric center, i.e., the aperture center for the PRM, is (
The nonuniform density structure of the ambient medium is indeed found in our radio data. Figure 8 shows the total interstellar proton column density map combining both the molecular (CO: NANTEN2) and atomic (H i: McClure-Griffiths et al. 2005) components. The details of the radio observations and data analysis are described in Appendix B. The velocity range of the gas density map is from −118.0 km s−1 to −109.0 km s−1, which corresponds to the kinematic distance of 6.2–6.4 kpc adopting a model of the Galactic rotation curve by Brand & Blitz (1993). The velocity range is determined by CO/H i cavities in the position–velocity diagrams (see Appendix B.3), suggesting an expanding gas motion originated by accretion winds from the progenitor system of the SNR (e.g., Sano et al. 2017, 2018). Therefore, the molecular and atomic clouds at the velocity range are likely associated with the SNR. The velocity range and distance to the SNR are consistent with the previous H i study by Giacani et al. (2011).
Download figure:
Standard image High-resolution imageWe find denser clouds possibly interacting with the SNR in the west (Figure 8), consistent with the previous results from the Mopra Southern Galactic Plane CO Survey (Burton et al. 2013; Braiding et al. 2015; Lau et al. 2019). The SNR expansion is likely to decelerated by this interaction toward the west, resulting in the asymmetric morphology we observe. This interpretation is supported by the temperature gradient inferred from the mean energy map (Figure 2(b)); the lower temperature is achieved in the northeast region probably due to more efficient adiabatic expansion. A similar temperature gradient due to the asymmetric ambient density distribution is observed in other middle-aged SNRs, e.g., W49B (Lopez et al. 2013; Yamaguchi et al. 2018). Alternatively, the SN ejecta at the west could have been reheated by reflection shocks originating from the interaction with the dense clouds, as was observed in, for instance, Kes 27 (Chen et al. 2008).
We have also discovered a local asymmetric structure at the west rim, "Filament" in Figure 2(c). Interestingly, the measured IME abundances are enhanced with respect to the solar values. Moreover, the relative abundances of S, Ar, Ca to Si are comparable to the solar composition and that of Mg is substantially lower, which is consistent with the typical nucleosynthesis yields of the incomplete Si burning (e.g., Iwamoto et al. 1999; Seitenzahl et al. 2013; Townsley et al. 2016). This implies an ejecta origin of this structure, despite its filamentary morphology and location being more consistent with an interpretation of the forward shock front that propagates into the ISM. If the structure is indeed dominated by the SN ejecta, it is reminiscent of the ejecta knots in Tycho SNR, suggesting that such local asymmetry is somewhat common in a certain type of Type Ia SNRs.
4.3. Point-like Sources
We have analyzed the spectra of the point-like sources detected with total photon counts of 30 or more (see Table 2). The best-fit parameters for the absorbed blackbody model are given in Table 6, where the angular distance from the explosion center to the sources is given as well. If the sources are physically associated with the SNR, the measured absorption column density must be consistent with that of the SNR (5–7 × 1022 cm−2). We find that only eight sources, No. 31, 36, 39, 41, 42, 43, 48, and 49, satisfy this criterion, and the other sources are most likely either foreground or background objects.
Table 6. Spectral Properties of All Point Sources
No.a | Luminosityb () | Distance c | ||
---|---|---|---|---|
1 | <0.06 | 1.66 | ||
2 | <0.16 | 3.34 | ||
3 | <0.09 | 3.99 | ||
6 | <0.10 | 4.69 | ||
7 | <0.38 | 4.45 | ||
8 | <0.10 | 5.35 | ||
9 | <0.93 | 2.93 | ||
11 | <0.24d | 5.11 | ||
14 | <0.38 | 6.55 | ||
15 | <0.14 | 6.33 | ||
16 | <0.68 | 4.29 | ||
17 | <0.71 | 2.66 | ||
18 | <2.45 | 3.85 | ||
19 | <0.62 | 3.09 | ||
20 | <0.25 | 1.04 | ||
22 | <0.54 | 4.38 | ||
24 | <1.23 | 4.17 | ||
25 | <0.41 | 4.82 | ||
26 | 3.03 | |||
27 | 2.63 | |||
28 | <0.62 | 4.58 | ||
29 | <1.90 | >2.26 | <253 | 6.82 |
30 | <0.39 | 3.96 | ||
31 | <2.77 | 1.51 | ||
6.69e | ||||
32 | <0.18 | 1.74 | ||
33 | 3.17 | |||
34 | 2.89 | |||
35 | >710 | <197 | 1.83 | |
36 | <1.12 | 2.93 | ||
6.63e | ||||
37 | 1.52 | |||
38 | 1.45 | |||
39 | 2.80 | |||
41 | <28.3 | <4.19 | 1.33 | |
42 | 0.97 | |||
43 | <1.47 | 2.29 | ||
4.76e | ||||
45 | 1.97 | |||
46 | <0.06 | <416 | 2.12 | |
47 | 2.50 | |||
48 | <5.13 | 4.94 | ||
6.50e | ||||
49 | <11 | <0.37 | 4.22 |
Notes.
aFaint sources are excluded (see Section 3.1). bWe assume the distance to the SNR to be 6 kpc. cThe angular distance from the Fe peak location. dWe allow to vary in 0.2–0.5 keV for this source. eBest-fit values applied constraints in error ranges on each corresponding SNR region in Table 3.Download table as: ASCIITypeset image
Next, we compare the temperature and luminosity of these sources with typical values for CCOs: kTbb = 0.2–0.5 keV and 1033-34 erg s−1 (e.g., Pavlov et al. 2004; Gotthelf et al. 2013). We find that four sources, No. 31, 36, 43 and 48 show typical kTbb and luminosity as CCOs. These sources, however, have relatively large errors of 37%–90% in NH. Thus, we allow their NH values to vary in the error ranges of NH for the corresponding regions in Table 3 for each source. We finally find that all four sources have low luminosities with expected NH values of the different SNR regions. Therefore we conclude that there is no CCO associated with G344.7–0.1.
5. Conclusions
The elemental distribution in SNRs is key to understanding the explosion mechanism of SNe. We have presented the results of the deep Chandra observations of the Type Ia SNR G344.7–0.1 that revealed the distribution of the SN ejecta. The X-ray images and radial profiles of the surface brightness have revealed a centrally peaked distribution of the Fe ejecta, surrounded by an arc-like structure of the IMEs. The centroid energy of the Fe K emission is lower in the central Fe-rich region than in the outer IME-rich regions, suggesting that the Fe ejecta were heated by the reverse shock more recently in the IME regions. These results are consistent with predictions of standard SN Ia models, where the heavier elements are synthesized in the interior of an exploding white dwarf.
This work is financially supported by the Chandra GO Program grant GO8-19052A and Grants-in-Aid for Scientific Research (KAKENHI) of the Japanese Society for the Promotion of Science (JSPS) grants No., 19H00704 (HY), and 24224005 and 19H05075 (HS). S.K. is supported by the Leading Initiative for Excellent Young Researchers, Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan. H.S. is supported by "Building of Consortia for the Development of Human Resources in Science and Technology" of MEXT (grant No. 01-M1-0305). NANTEN2 is an international collaboration of 11 universities: Nagoya University, Osaka Prefecture University, University of Bonn, University of Cologne, Seoul National University, University of Chile, University of New South Wales, Macquarie University, University of Sydney, University of Adelaide, and University of ETH Zurich.
Software: CIAO (Fruscione et al. 2006), XSPEC (Arnaud 1996).
Appendix A: Background Estimation
An X-ray background consists of three major components, non X-ray background (NXB), cosmic X-ray background (CXB), and the Galactic thermal emission. The third component is non-negligible in our study, since G344.7–0.1 is located in the Galactic plane.
The NXB spectra contain several fluorescence lines and continuum emission. We reproduce the NXB component with a model of bknpowerlaw + 5 Gaussians. Gaussians are for instrumental emission lines of Al K
Download figure:
Standard image High-resolution imageTable A1. Best-fit Parameters for the Background Model
Components | Parameters | |
---|---|---|
NXB | ||
c | ||
Absorptiona | ||
CXB | ||
c | ||
Galactic thermal emissionsb | ||
VEMd | ||
VEM d | ||
C-statistic/dof | 1410.75/1152 |
Notes.
aThe absorption cross sections are taken from Verner et al. (1996). bSolar abundances of Wilms et al. (2000) are assumed. cIn units of at 1 keV. dIn units of . The volume emission measure (VEM) is given as , where V and D are the volume of the emission region () and the distance to the emitting source (), respectively.Download table as: ASCIITypeset image
The intensities of the Galactic center X-ray emission (GCXE) and Galactic ridge X-ray emission (GRXE) are modeled by Uchiyama et al. (2013) based on many Suzaku observations. We compared the intensity of the Galactic emission component in our model with the predictions of the GCXE and GRXE intensities in this paper. We calculated the intensity in the 2.3–5.0 keV band based on Equation (1) in Uchiyama et al. (2013) and derived a value of 12.6 × 10−7 photon s−1 cm−2 arcmin−2, which compares with the value in our model of photon s−1 cm−2 arcmin−2 (within 90% errors). Therefore, our model intensity for the Galactic emission is consistent with the model predictions in Uchiyama et al. (2013).
Appendix B: Radio Observations
B.1. CO and H i
Observations of 12CO (J = 1–0) line emission at 115.271202 GHz were executed in 2012 May 18th using the NANTEN2 millimeter/submillimeter radio telescope. The telescope is installed at 4865 m altitude of the Atacama Desert in Chile, operated by Nagoya University. We observed an area of 1° × 1° region using the on-the-fly mapping mode with Nyquist sampling. The 4 K cooled Nb superconductor-insulator-superconductor mixer receiver was used for the front end. The system temperature including the atmosphere was ∼200 K in the double-side band. The backend was a digital Fourier-transform spectrometer with 1684 channels or 1 GHz bandwidth, corresponding to a velocity coverage of 2600 km s−1 and a velocity resolution of 0.16 km s−1. After convolving the cube data with a two-dimensional Gaussian kernel of 90'', the final beam size was ∼180'' in FWHM. The absolute intensity was calibrated by observing the standard source IRAS 16293–2422 [(
The H i line data at 1.4 GHz is from the Southern Galactic Plane Survey (SGPS; McClure-Griffiths et al. 2005). The archival data set was obtained using a combination of the Australia Telescope Compact Array (ATCA) and Parkes radio telescope. The angular resolution is 130'' and the velocity resolution is 0.82 km s−1. The typical noise fluctuation is ∼1.9 K at the velocity resolution of 0.82 km s−1.
B.2. Estimation of Total ISM Protons
To obtain the total proton column density map, we estimate proton column densities in both the molecular and atomic clouds (e.g., Fukui et al. 2012, 2017; Kuriki et al. 2018; Sano et al. 2019). The proton column density of molecular cloud Np(H2) can be estimated from the following equations:
where N(H2) is the column density of molecular hydrogen in units of cm−2, XCO is the CO-to-H2 conversion factor in units of (K km s−1)−1 cm−2, and W(CO) is the integrated intensity of CO in units of K km s−1. In the present paper, we used XCO = 2.0 × 1020 (K km s−1)−1 cm−2 (Bertsch et al. 1993).
On the other hand, an estimation of the proton column density of atomic cloud Np(H i) is tricky because we should consider the optical depth of H i. According to Fukui et al. (2015), 91% of atomic hydrogen in local clouds has an optical depth of 0.5 or higher with respect to the H i line emission. Therefore we cannot use the well-known equation assuming the optical depth of H i ≪1 (e.g., Dickey & Lockman 1990):
where W(H i) is the integrated intensity of H i in units of K km s−1.
To estimate the optical-depth-corrected proton column density of atomic cloud Np(H i)', we used a relation between W(H i) and Np(H i)' presented by Fukui et al. (2017). The authors derived Np(H i)' as a function of W(H i) using the dust opacity map at 353 GHz obtained from IRAS and Planck data sets (Planck Collaboration et al. 2014) taking into account nonlinear dust properties (e.g., Roy et al. 2013; Okamoto et al. 2017; Hayashi et al. 2019a, 2019b). In the present paper, we derived a conversion factor from W(H i) to Np(H i)' fitted by the result of Fukui et al. (2017) as a function of W(H i). We obtained averaged value of Np(H i)' toward G344.7–0.1 is ∼1.2 × 1021 cm−2, which is ∼2.3 times higher than that of Np(H i) assuming the optically thin case. Finally, we derived the total proton column density map combining ' and whose angular resolution is smoothed to match the FWHM of NANTEN2 beam size of ∼180'' (see Figure 8).
B.3. An Expanding Gas Motion
Giacani et al. (2011) argued that G344.7–0.1 is likely associated with an open H i shell at the central velocity of ∼115 km s−1, corresponding to the kinematic distance of 6.3 ± 0.1 kpc. Here, we investigate the cloud association using both the CO and H i data sets.
Figures B1(a) and (c) show the integrated intensity maps of H i and CO at a velocity of ∼115 km s−1, respectively. We found a shell-like structure not only in the H i map, but also in the CO map. The CO clouds nicely surround the SNR, especially for the northern half of the shell. Figures B1(b) and (d) show the position–velocity diagrams of H i and CO, respectively. We found cavity-like structures of H i and CO (dashed lines) in the velocity range from −118.0 km s−1 to −109.1 km s−1, which have a similar diameter to G344.7–0.1 in terms of the decl. range. These cavity-like structures represent an expanding gas motion similar to a "wind-blown shell," originated by strong winds from the progenitor system: e.g., stellar winds from a high-mass progenitor or accretion winds (also known as "disk winds") from a single degenerate progenitor system of Type Ia SNR. For the case of Type Ia SNR G344.7–0.1, the latter origin is favored. The expansion velocity of molecular and atomic clouds
Download figure:
Standard image High-resolution imageFootnotes
- 12
The mean photon energy (MPE) is defined as = (where Ei and N are the energy of each photon and the total number of photons detected in the pixel x, y). An MPE map is known to represent the temperature distribution, when the photons are extracted from the broad energy band (e.g., Troja et al. 2008; David et al. 2009).
- 13
- 14
- 15
High P2/P0 indicates elliptical or elongated morphology, and high P3/P0 indicates asymmetric or nonuniform surface-brightness distribution (Lopez et al. 2011).