X-Ray Spectroscopy of SN 1006 with Suzaku
Hiroya Yamaguchi, Katsuji Koyama, Satoru Katsuda, Hiroshi Nakajima, John P. Hughes, Aya Bamba, Junko S. Hiraga, Koji Mori, Masanobu Ozaki, Takeshi Go Tsuru
aa r X i v : . [ a s t r o - ph ] A p r PASJ:
Publ. Astron. Soc. Japan , 1– ?? , c (cid:13) X-Ray Spectroscopy of SN 1006 with Suzaku
Hiroya
Yamaguchi , Katsuji
Koyama , Satoru
Katsuda , Hiroshi
Nakajima , John P.
Hughes , Aya
Bamba , Junko S.
Hiraga , Koji
Mori , Masanobu
Ozaki , and Takeshi Go Tsuru Department of Physics, Kyoto University, Kitashirakawa-oiwake-cho, Sakyo-ku, Kyoto [email protected] Department of Earth and Space Science, Osaka University, 1-1 Machikaneyama, Toyonaka, Osaka 560-0043 Department of Physics and Astronomy, Rutgers University, 136 Frelinghuysen Road, Piscataway, NJ 08854-8019, U.S.A. Institute of Space and Astronautical Science, JAXA, 3-1-1 Yoshinodai, Sagamihara, Kanagawa 229-8510 RIKEN (The Institute of Physical and Chemical Research), 2-1 Hirosawa, Wako, Saitama 351-0198 Department of Applied Physics, University of Miyazaki, 1-1 Gakuen Kibana-dai Nishi Miyazaki, 889-2192 (Received 2007 May 6; accepted 2007 June 12)
Abstract
We report on observations of SN 1006 with Suzaku. We firmly detected K-shell emission from Fe, forthe first time, and found that the Fe ionization state is quite low. The broad-band spectrum extractedfrom southeast of the remnant was well-fitted with a model consisting of three optically thin thermalnon-equilibrium ionization plasmas and a power-law component. Two of the thermal models are highlyoverabundant in heavy elements, and hence are likely due to ejecta. These components have differentionization parameters, n e t ∼ . × cm − s and n e t ∼ . × cm − s; it is the later one that producesFe-K emission. This suggests that Fe has been heated by reverse shock more recently than the otherelements, consistent with a picture where the ejecta are stratified by composition with Fe in the interior.On the other hand, the third thermal component is assumed to be solar abundance, and we associateit with emission from the interstellar medium (ISM). The electron temperature ( kT e ∼ . Key words:
Supernova Remnants: individual (SN 1006) — X-Rays: spectra
1. Introduction
Based on the historical record, SN 1006 is widely re-garded to be one of the Galactic type Ia supernova rem-nants (SNRs) similar to Tycho’s SNR (Schaefer 1996).The current supernova (SN) models predict that Fe pro-duction in type Ia SNe is far larger than that of core-collapse SNe (e.g., Nomoto et al. 1984; Iwamoto etal. 1999). Therefore, measuring the Fe abundance is oneof the essential clues to classify the SN type of any specificremnant. The K-shell X-ray emission lines from ionizedFe offer the most direct information for abundance deter-mination. In fact, strong Fe-K α lines had been observedfrom Tycho’s SNR (e.g., Hwang et al. 1998; Decourchelleet al. 2001).In SN 1006, Vink et al. (2000) suggested the possiblepresence of an Fe-K emission line at 6.3 ± ∼
500 eV @6 keV). Other recent X-ray missions,Chandra and XMM-Newton, have not succeeded in de-tecting an Fe-K line. Cold Fe in the interior of SN 1006is known to exist based on ultraviolet absorption studies.Blue and red-shifted Fe II absorption lines were detected inthe spectra of background stars (Wu et al. 1993; Hamilton et al. 1997; Winkler et al. 2005), and interpreted as beingdue to unshocked Fe in the interior of the remnant. Theseresults show that portions of the Fe-rich ejecta are stillexpanding freely, and have not yet been overrun by thereverse shock. However the amount of Fe inferred in thesestudies ( < ⊙ ) is much less than the amount (0.6–0.8 M ⊙ ) predicted to be produced in the thermonucleardisintegration of a ∼ ⊙ white dwarf.In the early phases of supernova remnant evolution,even the shock-heated plasma is far from thermal equi-librium in terms of either the ionization or particle (elec-tron and ion) temperatures. The ionization age, a key di-agnostic of the non-equilibrium ionization (NEI) state, isdefined as n e t , the product of the electron density and thetime since the gas was heated. Typically, n e t is requiredto be ≥ cm − s for full ionization equilibrium (Masai1984). The thermal X-ray spectrum of SN 1006 suggestsan ionization timescale of n e t ∼ × cm − s (Vink etal. 2000), which is far from the full ionization equilibrium,and lower than nearly all other Galactic SNRs. The elec-tron temperature at the northwest rim of the remnant hasbeen estimated to be kT e ∼ . ∼ . ∼ − (Ghavamian et al. 2002), non-equilibration of ion and electron temperatures is present H. Yamaguchi et al. [Vol. , Fig. 1.
XIS-BI mosaic image of SN 1006 in the 0.55–0.6 keVband. Exposure and vignetting effects are corrected. Thecoordinates (R.A. and Dec) are referred to epoch J2000.0.The black square shows the FOV of the XIS on the SE regionof SN 1006 where this paper is concentrated. The blue squaresshow the FOVs on the other three quadrants (NE, NW, andSW). in SN 1006 as well.The extreme non-equilibrium state and high shock ve-locity are likely due to the low density of the ambient gas,because SN 1006 ( b = +14 .
6) is further from the Galacticplane than other Galactic SNRs, such as Cas A ( b = − . b = +1 . b = 6 .
8) is nearly as far above the plane as SN1006 (whentheir respective distances are included), Kepler appears tobe evolving into a dense circumstellar medium. Therefore,the evolutionary state of SN 1006 may be the lowest ofthese youngest Galactic SNRs, even though the real ageis ∼ ∼ × cm,following Winkler et al. (2003). The errors quoted in thispaper represent the 90% confidence level, unless otherwisestated.
2. Observations and Data Reduction
Four pointings were made of SN 1006 with Suzaku dur-ing the performance verification (PV) phase. Suzakuhas one Hard X-ray Detector (HXD; Takahashi etal. 2007, Kokubun et al. 2007), and four X-ray Imaging
Fig. 2.
Background-subtracted XIS spectra extracted fromthe whole SE quadrant (SN 1006 SE). The black and redpoints represent the FI and BI spectra, respectively.
Spectrometers (XIS; Koyama et al. 2007) each placed inthe focal plane of an X-Ray Telescope (XRT; Serlemitsoset al. 2007). This paper reports on the imaging andspectral results obtained with the XIS. The XIS con-sist of three Front-Illuminated (FI) CCDs and one Back-Illuminated (BI) CCD. The advantages of the former arehigh detection efficiency and low background level in theenergy band above ∼ × × ≤ ∼
50 ks, for each ofthe pointings. The response matrix files (RMF) and ancil-lary response files (ARF) were made using xisrmfgen andxissimarfgen (Ishisaki et al. 2007) version 2006-10-17.
3. Overall Structure
The field of view (FOV) of the XIS is ∼ ′ × ′ , andhence a four-point mosaic can completely cover SN 1006(about 30 ′ diameter). Since we intended to study the ther-mal plasma, we searched for the region of the remnantwith the most prominent thermal emission. We hencemade a mosaic image in the O VII K α line band (0.55–0.6 keV), because this line is a major component of thethermal plasma in SN 1006. The result is shown in fig-ure 1. We can see bright emission in this band in thenortheast (NE) and southwest (SW) regions of the rem-nant. However, these are dominated by non-thermal emis-sion (Koyama et al. 1995), from which it is difficult too. ] Suzaku Observation of SN 1006 3 Fig. 3.
XIS intensity map at the Fe-K α line (a: 6.33–6.53 keV band), from which the continuum flux at 6.1–6.3 keV band [shownin (b)] is subtracted. In both images, exposure and vignetting effects are corrected. The data from the three FIs are combined. Twocorners of the calibration sources are removed. The black squares indicate each FOVs of the XIS. The red ellipse shows the regionwhere we extracted the spectra for a detailed analysis. extract thermal spectra. Other regions are dominated bythermal emission. In particular, as shown in figure 1, thesoutheast quadrant (here SN 1006 SE) is brighter thanthe northwest (NW). Therefore, our study concentratedon this region. The spectral results of non-thermal rims(NE and SW) are reported by Bamba et al. (2008), andthose of the other regions (NW quadrant and the centerof the remnant) will be reported in another paper in thefuture. The FOV of the XIS on SN 1006 SE is outlinedby the black solid square in figure 1, which was observedon 2006 January 30 (Obs. ID=500016010). We extracted the spectrum of the entire SN 1006 SE re-gion, excluding the two corners in the FOV that containcalibration source emission. For the background, we usedthe North Ecliptic Pole (NEP) data (Obs. date = 2006February 10, Obs. ID = 500026010, Exp. time = ∼
82 ks).Although we have been monitoring and correcting the in-crease of the Charge-Transfer Inefficiency (CTI) for theXIS (see Koyama et al. 2007), recovery of the energy res-olution cannot be made. However, we can ignore the dif-ference of the spectral resolution between the SN 1006 SEand NEP observations, because the latter observation wasmade only ten days after the former. To minimize the un-certainty due to the background subtraction, in particularthat of NXB, we applied the same data-screening criteriato both the SN 1006 SE and NEP observations, and tookbackground spectra from the same detector coordinatesas the source regions after excluding point-like sourcesdetected in the NEP data. The background-subtractedspectra are shown in figure 2. Since the data from the three FIs are nearly identical, we merged those individualspectrum to improve the photon statistics.As shown in figure 2, we found clear K-shell (K α ) linesfrom Ar, Ca, and Fe, for the first time. With a power-lawplus Gaussian-line fit, we determined the line center en-ergy of the Fe-K α to be ∼ We show in figure 3a an image in a relatively narrowband (6.33–6.53 keV) that contains the Fe-K α line. Thisimage was generated by subtracting the continuum flux atenergies of 6.1–6.3 keV. (The image in this band is shownin figure 3b.)We can see that the Fe-K α flux is enhanced at thesouthern part of the remnant (outlined in red with a el-lipse), except for the NE and SW quadrants where thenon-thermal emission is dominant. The mean surfacebrightness at 6.33–6.55 keV within the elliptical regionis 8 . ± . × − photons cm − s − arcmin − , whilethat outside it (only in the SE and NW quadrants) is4 . ± . × − photons cm − s − arcmin − . In orderto study the thin-thermal spectrum with the best S/N ra-tio for Fe-K line, we extracted the X-ray spectrum fromwithin the elliptical region, excluding the corner of the cal-ibration sources. The background subtraction was madein the same way as that of the full-field spectrum. The re-sults are given in figure 8. Hereafter, all detailed analysesare made using this spectrum. H. Yamaguchi et al. [Vol. , Table 1.
The center energies and widths of the emissionlines.
Line Center energy ∗ (eV) Width † (eV)Mg-K α < α α α < α < α < ∗ Errors (statistical only) are given in the parentheses(see text). † One standard deviation (1 σ ). In order to study the line features, we fitted the spectraextracted from the elliptical region with a phenomenolog-ical model; a power-law for the continuum and Gaussiansfor the emission lines. The best-fit central energies andwidths for the emission lines are shown in table 1. Sincethe absolute energy calibration error is ± α lines. We also note that the widths of the Gaussiansidentified with Si-K α and S-K α are significantly broaderthan the instrumental energy resolution at those energies.Figure 6a would help to compare the widths of the datawith those of narrow lines. In subsection 4.3, we discussthis matter in detail.
4. Spectral Structure in the Narrow Bands
In order to study the plasma characteristics in relationto the various elements, we at first divided the spectruminto three representative energy bands: the 0.4–1.1 keVband for the O and Ne (light elements) lines, the 1.2–2.8 keV band for the Mg, Si, and S (medium elements)lines, and the 5–10 keV band for the Fe (heavy element)line. With the fitting to these individual band spectra,we constrained the plasmas including light, medium, andheavy elements separately.
We fitted the 5–10 keV spectrum with a solar abun-dance (Anders and Grevese 1989) VNEI model. Since theNEIvers 2.0 plasma code does not include K-shell emissionlines for ions below the He-like state, we reverted to theNEIvers 1.1 code. If we fix the Fe abundance to be solar,then the model cannot reproduce the Fe-K α profile or flux,although the χ /d.o.f. of 25/28 is acceptable. Allowingthe Fe abundance to be free results in a greatly improvedbest-fit χ /d.o.f. of 10/27. The best-fit kT e , n e t , and Feabundance are 3.8 (1.7–26) keV, 6.1 (2.2–10) × cm − s,and 6.5 ( > Fig. 4.
XIS spectra in the 5–10 keV band with the VNEImodel. The black and red data points represent the FI andBI spectra, respectively. In the BI spectrum, the energy bandabove 8 keV is ignored. pure thermal emission. The best-fit spectrum is shown infigure 4.
Figure 5 shows the 0.4–1.1 keV band spectra. This en-ergy band includes K α lines of the light elements, withdominant emission from the oxygen K-shell line series.The spectra were fitted with a VNEI model, allowing theabundances of C, N, O, and Ne to be free parameters.Also, we let the abundances of Ca and Fe to be free, be-cause the L-shell lines of these elements fall into this en-ergy band (0.4–1.1 keV). Interstellar absorption was fixedto a hydrogen column density of 6 . × cm − (Dubneret al. 2002). In the initial fits, we found a significantinconsistency between the FI and BI spectra near the en-ergy of the O-edge (0.54 keV). This may be due to incom-plete calibration information for the contamination layeron the optical blocking filter (OBF) of the XIS (Koyamaet al. 2007). Since the calibration of the OBF contamina-tion for the BI is more accurate than that for the FI, wedecided to retain the BI data across this energy band andignore the 0.5–0.63 keV band in the FI spectrum. The re-sults are shown in figure 5a, with the best-fit temperatureand ionization parameter of kT e = 0.58 (0.56–0.59) keVand n e t = 6.7 (6.6–6.8) × cm − s.In figure 5a, we find an apparent disagreement betweenthe data and model in the ∼ s − p ( ∼
730 eV) and 3 d − p ( ∼
830 eV)from Fe
XVII . Although the fluxes of these two lines in NEImodels are nearly equal over a wide range of plasma condi-tions, extrapolation to the extremely low ionization statesthat we see here in SN 1006 is quite uncertain. In par-ticular, for the young Type Ia SNR E0509 − .
5, Warrenand Hughes (2004) found that the Fe L-shell emission wasdominated by an emission from a line feature near 0.73keV, which was not sufficiently strong in the spectral mod-els and had to be included as a separate Gaussian compo-nent. Based on the strength of other Fe-L emission lineso. ] Suzaku Observation of SN 1006 5and the weakness of O emission in E0509 − . VII and O
VIII , respectively. In most astrophysical plas-mas, the best-fit temperature of ∼ VII compared to those of O
VIII . On theother hand, in lower temperature plasmas (e.g., kT e ∼ . − . VII K α line is dominant, theline fluxes decrease rapidly along the K-shell transitionseries (K α,β,γ,δ,ǫ,ζ , etc.). Therefore, K-shell lines in thehigher transitions can be safely ignored, and hence conven-tional NEI codes do not include O VII
K-shell transitionlines higher than K δ . This is the reason that no oxygenline in the ∼ kT e is moderate, so the fluxes ofhigher level K-shell transitions from O VII may be rela-tively strong and cannot be ignored. To account for themwe added Gaussians at 714 eV, 723 eV, and 730 eV to rep-resent the O
VII K δ , K ǫ , and K ζ lines, respectively. At aplasma temperature of 0.6 keV, the flux ratio of O VIII Ly ǫ to Ly δ predicted by the NEIvers 1.1 code is ∼ VII lines follow thesame pattern, namely K ǫ /K δ = K ζ /K ǫ = 0.5. The best-fit results with these additional lines, see figure 5c, are areasonably good fit. The additional artificial lines are alsoshown as the dotted lines in figure 5b. The temperatureand ionization parameter are almost the same as those ob-tained without the additional higher K-shell transitions ofO VII .The best-fit intensity ratio of O
VII K β to O VIII Ly α is ∼ VII K β /O VIII Ly α ∼ VII
K-shell line series are key spectralcomponents for all over the SN 1006 plasma.
This band includes K-shell lines of medium-weight ele-ments (Mg, Si, and S). As shown in table 1, we see cansignificant broadening of the Si and S K α lines. Therefore,the 1.2–2.8 keV band spectra cannot be reproduced witha single-component plasma model. In fact, we found sig-nificant disagreement between the data and model aroundthe Si-K α and S-K α lines (figure 6a) in the one-componentVNEI fits. The lack of data in the 1.83–1.85 keV band isdue to current calibration errors of the XIS; there is asmall gap in energy ( ∼
10 eV) at the Si K-edge, which isnot implemented in the current response function. Theapparent line broadening of the Si K α line, however, ismuch larger ( ∼
40 eV) than this gap. Therefore, the dis-
Fig. 5. (a) XIS spectra in the 0.4–1.1 keV band fitted with aVNEI model. The black and red data points represent the FIand BI spectra, respectively. (b) Spectrum in photon spacecorresponding to the model used in (a), but only showingO lines. It is binned every 2 eV. The black and red linesrepresent O
VII series and O
VII series, respectively. The dot-ted lines are additional transitions of K δ (714 eV), K ǫ (723eV), and K ζ (730 eV) of O VII added as separate Gaussians(see text). (c) Same spectra as (a), but for fits includingGaussians representing K δ (714 eV), K ǫ (723 eV), and K ζ (730 eV) of O VII (see text). The residuals in the energy band0.7–0.85 keV seen in (a) are largely removed.
H. Yamaguchi et al. [Vol. ,
Fig. 6. (a) XIS spectra in the 1.2–2.8 keV band fitted with the one-component VNEI model. The black and red represent the FIand BI, respectively. (b) Same to (a), but fitted with the two-component VNEI. agreement is not due to a calibration error, but is real.If the intrinsic broadening of the Si-K α line ( ∼
40 eV;see table 1) is due to thermal Doppler broadening, the ion(silicon) temperature must be ∼
13 MeV, which requiresa shock velocity of ∼ . × km s − . Ghavamian etal. (2002) determined the shock velocity (from measure-ments of the H α line width) to be ∼ − in theNW portion of SN 1006. This is where the blast wave isinteracting with significant amounts of interstellar matter,compared to other parts of the rim. In fact, the averagesize of the remnant ( ∼ . ′ ∼ km s − . The true currentblast wave speed in SN 1006 is almost surely bracketed bythese two values. Furthermore, the reverse shock, which islikely to be the heating source for the Si that we see, typ-ically moves into the ejecta at only a fraction of the blastwave speed. Therefore, we consider it unlikely that theline broadening we see is due to thermal Doppler broad-ening.Another possible explanation proposes that the emis-sion consists of several thermal plasma components. Wefitted the spectra with 2-VNEI models, in which we al-lowed for different kT e values, but the same n e t betweencomponents. However, no combination of parameterswould fit the data, even if we let the abundances of Mg,Si, and S be free parameters. Not only 2-VNEI, butalso 2-VPSHOCK models failed to fit the data when thetwo components were forced to have the same ionizationtimescale. All of the models that we tried, but rejected, Table 2.
Best-fit parameters and χ values of the spectralfitting in the 1.2–2.8 keV band with the various models. Model kT e kT e n e t χ /d.o.f.(keV) (keV) (cm − s)1-VNEI 1.5 – 4.5 × × × × n e t values. Thebest-fit reduced χ /d.o.f. was greatly improved to 401/346(see table 2, for comparison). Models with different elec-tron temperature and ionization timescale (2- kT e and 2- n e t ) gave no significant improvement of the reduced χ .We thus conclude that a 2-component VNEI model withdifferent n e t values is necessary to fit the medium ele-ment plasma band. The best-fit parameters and spectraare given in table 3 and figure 6b, respectively.
5. Model Fit of the Full Band Spectra
In section 4, we discuss how we separately derived spec-tral parameters for the three-energy bands that representtypical plasma conditions for the main light elements (Oand Ne), medium elements (Mg, Si, and S), and heavyelement (Fe). A summary is given in table 4. We usedthese parameters as the initial values to search for plasmaparameters in the full energy band of 0.3–10 keV; the re-sulting overall best-fit parameters must be consistent withthe data over the entire XIS energy band.Based on table 4, we assume that the model that de-scribes the full spectral range must consist of, at least,three plasma components: (1) high- kT e ( ∼ n e t ( ∼ cm − s), (2) similarly high- kT e withlow- n e t ( ∼ cm − s), and (3) low- kT e ( ∼ . n e t ( ∼ × cm − s). First, we considerwhether or not the full band spectra can be well repro-duced with only these components. We also consider abroader range of kT e and n e t values for the Fe-K emis-sion.As shown in table 3, components (1) and (2) are sig-nificantly enhanced in Mg, Si, and S (relative to solar).In the present spectral fits, we made the assumption thatthese components are composed purely of metals withoutany admixture of hydrogen or helium. This assumptiono. ] Suzaku Observation of SN 1006 7 Table 3.
Best-fit parameters of the spectral fitting in the 1.2–2.8 keV band with 2-component VNEI model with different n e t values. Parameter Component 1 Component 2 N H (cm − ) 6.8 × (fixed) kT e (keV) 1.1 (1.0–1.2)Mg 4.2 (3.5–5.2) 6.1 (3.5–13)Si 19 (15–24) 15 (13–24)S 24 (19–34) 23 (17–31) n e t (cm − s) 1.3 (1.1–1.7) × × n H n e V (cm − ) 2.8 (2.3–3.4) × × χ /d.o.f. 401/346 = 1.16 Table 4.
Plasma components determined from the narrow band spectra.
Band (keV) Major elements kT e (keV) n e t (cm − s) Component ∗ × (3)1.2–2.8 Mg, Si, S 1.1 (1.0–1.2) 1.3 (1.1–1.7) × (1)1.1 (1.0–1.2) 7.9 (6.2–9.8) × (2)5.0–10 Fe 3.8 (1.7–26) 6.1 (2.2–10) × ∗ The plasma component identifications described in text.arose because hydrogen and helium emit only continuumemission in the X-ray band. Given the complexity of ourspectral model, it is just not possible with the XIS data toobtain reliable estimates for the level of continuum emis-sion from the three plasma components separately. As afurther assumption, we fixed the abundance of component(3) to be the solar value. (Anticipating the results givenbelow, we found that most of the continuum emissionabove 1 keV actually comes from a hard power-law likespectral component.) Operationally, we fixed the oxygenabundance in both the pure-metal components (i.e., nos. 1and 2) to a large value (1 × ), and fitted for the abun-dances of other elements relative to oxygen. The C and Niabundances were assumed to be the same as Ne and Fe,respectively. The emission measure ( EM ) in the plasmacode is defined (in XSPEC) as n H n e V , even if the plasmais dominated by heavy elements. However, the oxygendensity can be calculated as n O = (8 . × − n H ) × ,where the numerical value is the solar abundance of oxy-gen from Anders and Grevesse (1989). Details of thismethod can be found in Vink et al. (1996).Since the NEIvers 1.1 spectral model does not includeK-shell emission lines from Ar and low ionization states ofCa, we added Gaussians at the energies of their K α tran-sitions: 3.02 keV (Ar) and 3.69 keV (Ca) (see table 1), Forcomponent (1) we additionally freely varied the Ca abun-dance, which effectively allowed for the Ca L-shell lines tocontribute at low energies ( < Fig. 7.
XIS spectra in the 0.33–10 keV band fitted with thethree-component plasma model (see text). The black and redpoints represent the FI and BI data, respectively. The modelfails to describe the continuum emission above ∼ OBF (see Koyama et al. 2007). Energies of 0.5–0.63 keV(only for FI) and 1.83–1.85 keV were also ignored for thesame reasons as noted in subsections 4.2 and 4.3. Sincethe absolute gain of the XIS has an uncertainty of ± χ /d.o.f.obtained was 1081/833. The fitted spectra are shownin figure 7. All line features are well fitted, even Fe-K,suggesting that including additional thermal componentswould not be justified. However, there is a systematic dataexcess in the continuum at energies > ∼ χ /d.o.f. = 996/831. The H. Yamaguchi et al. [Vol. , Table 5.
Best-fit parameters of the broadband spectral fitting.
Component Parameter ValueAbsorption N H (cm − ) 6.8 × (fixed)VNEI 1 (Ejecta 1) kT e (keV) 1.2 (1.1–1.4)Abundance (10 solar) C 0.19 ∗ (0.08–0.30)N 0 (fixed)O 1.0 (fixed)Ne 0.19 ∗ (0.08–0.30)Mg 4.1 (3.7–4.7)Si 17 (15–20)S 23 (20–27)Ca 11 (9.6–13)Fe 0.68 † (0.62–0.76)Ni 0.68 † (0.62–0.76) n e t (cm − s) 1.4 (1.2–1.6) × n H n e V (cm − ) 2.7 (2.4–3.0) × VNEI 2 (Ejecta 2) kT e (keV) 1.9 (1.5–2.6)Abundance (10 solar) C 0.33 ‡ (0.23–0.45)N 0 (fixed)O 1.0 (fixed)Ne 0.33 ‡ (0.23–0.45)Mg 3.0 (2.2–3.9)Si 10 (8.6–12)S 12 (9.1–16)Ca 18 § (5.1–44)Fe 18 § (5.1–44)Ni 18 § (5.1–44) n e t (cm − s) 7.7 (6.7–9.2) × n H n e V (cm − ) 1.8 (1.3–2.5) × NEI (ISM) kT e (keV) 0.51 (0.31–0.55) n e t (cm − s) 5.8 (5.7–6.1) × n H n e V (cm − ) 1.1 (1.0–1.2) × Power-law Γ 2.9 (2.8–3.0)Norm k (photons cm − s − ) 6.3 (5.5–7.2) × − Gaussian lines to complement the incomplete VNEI code (see text)Line Center Energy(keV)(fixed) Norm (photons cm − s − )O VII -K δ × − O VII -K ǫ × − O VII -K ζ × − ∗∗ Ar-K 3.01 5.7 (4.4–7.0) × − Ca-K 3.69 2.4 (1.3–3.5) × − Gain Offset (eV) –4.8 χ /d.o.f. 996/831 = 1.20 ∗ , † , ‡ and § represent linked parameters. k The differential flux at 1 keV.
Fixed to 50% of the normalization of O
VII -K δ . ∗∗ Fixed to 50% of the normalization of O
VII -K ǫ .o. ] Suzaku Observation of SN 1006 9 Fig. 8.
XIS spectra in the 0.33-10 keV band which were extracted from the red elliptical region shown in figure 3. The black andred data points represent the FI and BI spectra, respectively. The solid lines, colored green, blue, light blue, orange, and magentashow the components of the ISM, Ejecta 1, Ejecta 2, Power-law, and additional Gaussians of the best-fit model for the BI spectrum.Note that the Gaussian like structures (magenta) at ∼ ∼ best-fit parameters are given in table 5. As already noted,the gain may have an uncertainty of ±
6. Discussion
In section 5, we analyzed the full energy band spec-trum extracted from the elliptical region shown in figure 3,and found that it could be described adequately with aparameterized model including three thermal plasmas innon-equilibrium ionization (VNEI 1, VNEI 2, and NEI, intable 5) and one power-law component (Power-law, in ta-ble 5). Two of the plasmas, VNEI 1 and VNEI 2, were as-sumed to have non-solar abundances, and the other, NEI,was assumed to have solar composition. In the followingwe discuss the implications of our spectral results, butthe reader is cautioned that these results are subject tochange if the assumptions we have made are changed.
The NEI model, assumed to have solar abundances, pro-duces most of the observed low-energy X-rays, particularlythe K α lines from O VII , O
VIII , and Ne IX (see figure 8).This component is rather uniformly extended over the en-tire remnant, as can be seen in the O VII line band map(figure 1). We associate this component with the swept-upISM. The other plasmas, VNEI 1 and VNEI 2, with their non-solar elemental abundance ratios, are plausibly of ejectaorigin. Since VNEI 1 has a larger ionization parameterthan the other plasmas, we suggest that this plasma washeated by reverse shock in the early stage of remnant evo-lution (here Ejecta 1). On the other hand, VNEI 2 hasan extremely low ionization parameter, and hence shouldhave been heated much more recently (here Ejecta 2). Wehave firmly detected iron K α emission for the first time.The low ionization state of Fe, plus its overabundance inthe Ejecta 2 component, is generally consistent with theType Ia origin of SN 1006. We compare our best-fit relative abundances with thepredicted nucleosynthesis yield of the widely-used W7 SNIa model (Nomoto et al. 1984). For Ejecta 1, althoughthe abundances of C, Ne, Mg, Si, and S relative to O arebroadly consistent with the W7 model, Ca and Fe fall farbelow their predicted values, as shown in figure 9a. In thecase of Ejecta 2, on the other hand, the heavy elements,albeit with large errors, are broadly consistent with theabundance pattern from the W7 model, as shown in fig-ure 9b. These results, along with the difference in the ion-ization timescale between the components just mentioned,are consistent with a layed composition of the ejecta withthe higher-Z elements more concentrated toward the cen-ter of SN 1006.0 H. Yamaguchi et al. [Vol. ,
Fig. 9.
Metal abundances as a function of atomic number derived from the spectral fitting. The data points of (a) and (b) representthose of Ejecta 1 and Ejecta 2, respectively. The solid lines show the abundance relative to oxygen calculated in the W7 model fora Type Ia supernova by Nomoto et al. (1984).
For this simple interpretation, one expects the Fe lineemission from the low-ionization timescale component tobe spatially located interior to the Mg and Si emissionfrom the high-ionization component, i.e., near the centerof SN 1006. As shown in figure 3, though, the Fe fluxappears to peak close to the southeastern rim. In sum-mary, our two-component spectral model for the ejectais a highly simplified view of what is surely a complexmulti- n e t and multi- kT e structure that varies throughoutthe interior of the SNR. A longer observation, particularlyof the inner regions of SN 1006, will help to improve thestatistical accuracy of the Fe-K line flux, and allow us tostudy this apparent discrepancy in more detail. The plasma parameters given in table 5 were derivedfrom the southeast solid ellipse of figure 3, which is π × . ′ × . ′ in area. This correspondsto 3.3 × cm , from which we estimated the emis-sion volume to be V = (3.3 × ) / = 6.0 × cm .Therefore, the EM of the heated ISM derived from thespectral fits corresponds to an electron density of n e =0.15 f − . cm − , where f is the filling factor. Since theage of SN 1006 is ∼ n e t ∼ . × cm − s, which isalmost consistent with the best-fit value of the ionizationparameter of the NEI (ISM) component. Such a low ion-ization age suggests that the temperatures of the ions andelectrons may also be far from equilibrium. Accordingto equation (3) of Laming (2001), under the assumptionthat the initial ratio between kT e and kT H just at theshock front is the ratio of the electron and proton masses,the proton temperature is estimated to be, kT H = 7 . (cid:18) n e t . × cm − s (cid:19) − (cid:18) kT e .
51 keV (cid:19) / [keV] . This result is consistent with that of the H α observation of T e /T H ≤ .
07 (Ghavamian et al. 2002). The extreme non-equilibrium state of the plasma is due to the low density of the ambient medium. According to equations (2) and(5) of Ferri`ere (2001), the density of H I and H II at the Z -height of SN 1006 ( Z = 550 pc) is ∼ − . Thisis consistent with our estimate of the ambient density of n H / ∼ .
03 cm − . The low density of the ambient medium will allow thevelocity of the shock front to remain high for a relativelylong time. This may be one reason why this remnantshows such efficient particle acceleration up to very highenergies ( ∼ eV), observed as power-law (synchrotron)emission from the NE and SW rims (Koyama et al. 1995).The photon index from these rims is Γ ∼ ∼ ∼ × − ergs cm − s − arcmin − in the2–10 keV band. This value is ∼
7. Summary
We have analyzed high-quality spectra obtained withSuzaku of a region in the southern part of SN 1006, se-lected because it is bright in Fe-K α . The results and in-terpretations are summarized as follows:1. The spectrum can be described by a model with atleast three NEI thermal plasmas and one power-lawo. ] Suzaku Observation of SN 1006 11component.2. The fits yield different ionization parameters of n e t ∼ × cm − s (low), ∼ × cm − s(medium), n e t ∼ cm − s (high).3. The low- n e t plasma are highly overabundant inheavy elements, in which we found K α lines fromFe, for the first time. The abundance pattern isconsistent with that of type Ia SN ejecta.4. The abundance of the medium- n e t plasma is as-sumed to be solar, and we associate this componentwith the shocked ISM. Although oxygen is not over-abundant, K α lines of O VII and O