Suzaku Observations across the Cygnus Loop from the Northeastern to the Southwestern Rim
Masashi Kimura, Hiroshi Tsunemi, Satoru Katsuda, Hiroyuki Uchida
aa r X i v : . [ a s t r o - ph ] O c t Suzaku Observations across the Cygnus Loop from theNortheastern to the Southwestern Rim
Masashi
Kimura , Hiroshi Tsunemi , Satoru Katsuda , , Hiroyuki Uchida Department of Earth and Space Science, Graduate School of Science, Osaka University,1-1 Machikaneyama, Toyonaka, Osaka 560-0043, Japan NASA Goddard Space Flight Center, Greenbelt, MD 20771, [email protected] (Received 2008 August 8; accepted 2008 October 21)
Abstract
We have observed the Cygnus Loop from the northeast (NE) rim to the southwest(SW) rim using Suzaku in 10 pointings that is just north of previous XMM-Newtonobservations. The observation data obtained were divided into 45 rectangular regionswhere the width were configured so that each region holds 8000 − kT e -component non-equilibrium ionization (NEI) model or with two- kT e -component NEI model. Thetwo- kT e -component model yields significantly better fit in almost all the non-rimregions. Judging from abundances and flux, the high- kT e -component (0.4 − kT e -component ( ∼ ◦ . We also found that the ejecta distributions wereasymmteric to the geometric center: the ejecta of O, Ne and Mg were distributedmore in the NE, while the ejecta of Si and Fe were distributed more in the SW of theCygnus Loop. We calculated the masses for various metals and estimated the originof the Cygnus Loop as the 12 − ⊙ core-collapse explosion. Key words:
ISM: abundances – ISM: individual (Cygnus Loop) – ISM: supernovaremnants – X-rays: ISM 1 . Introduction
The Cygnus Loop is a nearby (540pc: Blair et al. 2005) middle-aged supernova remnant(SNR). Its large apparent size (2 ◦ . × ◦ .
5: Levenson et al. 1997; Aschenbach & Leahy 1999)and high surface brightness enable us to study soft X-ray emission from the Cygnus Loop.The bright surface emission is mainly from interstellar medium (ISM) swept-up by forwardshock wave. Miyata et al. (1994) observed the northeast (NE) rim of the Loop with AdvancedSatellite for Cosmology and Astrophysics (ASCA) and revealed metal deficiency there. Sinceprevious measurements indicate that the ISM around the Cygnus Loop has metal deficientabundances (Parker et al. 1967), they concluded that the plasma in the NE rim is dominatedby the ISM. In the contrast of the rim region, ASCA detected Si, S, Fe rich plasma at thecenter portion of the Cygnus Loop (Miyata et al. 1998), which is thought to be the ejecta. Therelative abundances of the ejecta support the idea that the Cygnus Loop was the result from acore-collapse supernova, and the progenitor mass is estimated to be 25M ⊙ .Recent XMM-Newton observations (Tsunemi et al. 2007) across the Cygnus Loop fromthe NE rim to the southwest (SW) rim revealed ejecta distributed inside of ∼ R s of theLoop, where R s is the shock radius. The relative abundances inferred for the total ejecta arealmost consistent with those expected for the core-collapse SN, whose progenitor mass is 13M ⊙ (Katsuda & Tsunemi 2008). Suzaku observed just south of XMM-Newton observation path(Katsuda et al. 2008) and revealed asymmetric metal distributions of the Cygnus Loop. Theyfound that Mg is distributed more in the NE, while Si, S, and Fe are distributed more in theSW of the Loop. In order to extent our knowledge on the ejecta distribution as well as theoverlying ISM, we observed the Cygnus Loop from the NE rim to the SW rim using Suzakuobservatory. Figure 1 shows our field of view (FOV) which cover just north regions of theXMM-Newton observation path.
2. Observation and Data Screening
The observations comprised ten points: NE3 , P1 − P7, P9 and P10. We employed re-vision 2.0 of the cleaned event data since there was no sufficient flares in the light curve. Forbackground subtraction, the spectra acquired from Lockman Hole were used. We selected theLockman Hole data whose observation dates were close to those of our Cygnus Loop observa-tions. Since there was no photons above 3.0 keV after the background subtraction, the energyranges of 0.3–3.0 keV and 0.4–3.0 keV were used for XIS1 (back-illuminated CCD; BI CCD) andXIS0, 3 (front-illuminated CCD; FI CCD), respectively (Koyama et al. 2007). We summarizedObs IDs, nominal points, observation dates, and effective exposure times after the screening intable 1. Figure 2 right shows a merged XIS1 three-color image. Red, green, and blue colorscorrespond to narrow energy bands of 0.52–0.70 keV (O
VII K α ),0.70–0.85 keV (Fe L) and0.85–0.94 keV (Ne IX K α ), respectively. The scale R shows the distance from an observation2enter in arcmin. We can see strong green color around the SW portion (i.e., P6, P7, P9),whereas red and blue colors are enhanced in the NE (i.e., P1 − P4) regions and P10.
3. Spatially resolved spectral analysis
We divided the entire FOV into 2 parts, namely the NE part and the SW part. The NEpart contains pointigs of NE3, P1, P2, P3, P4 and the SW part contains pointings of P5, P6,P7, P9, P10. The two parts were divided into total of 45 rectangular regions where their widthswere configured so that each area holds 8000 − xisrmfgen (ver. May 2007) and xissimarfgen (ver. March2008). In order to reduce the effect of radiation damage of XIS, the data were taken by usingthe spaced row charge injection (SCI) method (Prigozhin et al. 2008) which recovers the energyresolution of radiation damaged CCD. The SCI method is not fully supported by CALDB at thewriting phase of this paper, and our data clearly showed better energy resolution than that ofthe RMF generated by normal use of xisrmfgen . Therefore we searched the appropriate RMFby changing the date obs parameter of xisrmfgen which represents the observation date. Bychanging the date obs to earlier date, xisrmfgen can generate the RMF for earlier observationwhich has better energy resolution. We changed date obs by every 3 months from September2005 to September 2007, which generate 9 different RMFs in total. Then the generated RMFswere used to fit the 10 sample spectra taken from red regions in figure 2. We applied modeldescribed in next section to test the RMFs in order to find the most appropriate RMF. Sincethe RMF where we set date obs = December 2005 showed the least reduced-chi squared valuein every spectrum, those RMFs were used to perform further spectral fit. kT e VNEI model
The extracted spectra were first fitted by an absorbed non-equailibrium ionization (NEI)model with a single component [the wabs; Morrison & McCammon 1983 and the VNEI model(NEI ver. 2.0); e.g., Borkowski et al. 2001 in XSPEC v12.4.01]. Free parameters are thehydrogen column density, N H ; electron temperature, kT e ; the ionization time, τ , where τ isthe electron density times the elapsed time after the shock heating; the emission measure EM(EM= R n e n H dl , where n e and n H are the number densities of electrons and hydrogens and dl is the plasma depth); the abundances of C, N, O, Ne, Mg, Si, Fe and Ni. We set abundanceof S equal to that of Si and that of Ni equal to that of Fe. Abundances of the other elementswere fixed to the solar values (Anders & Grevesse 1989). The black line in figure 3 shows thereduced-chi square value as a function of R. The one-component model gave us large reduced-chi square values in some regions; therefore, we added an extra component for our spectralfit. 3 .2. Two- kT e VNEI model
We tried to apply this model by freeing all parameters, but since this did not give asreasonable values, we decided to fix certain values. The kT e , τ , and EM are free parameters inboth components. N H is also a free parameter but shares same value between two components.Since abundances of the swept-up matter region is well known, we set the abundances of low- kT e -component to those values determined in the NE region by Uchida et al. (2006). Theabundances of O (=C=N), Ne, Mg, Si (=S), and Fe (=Ni) for the high- kT e -component areset as free parameters. The fit improved significantly ( >
99% based on F-test probability) innon-rim region ( − ′ < R < ′ ) by adding another component (see figure 3). On the otherhand, the NE rim ( R < ′ ) and the SW rim ( R > ′ ) did not show notable improvement on thefit. Furthermore, parameters of the high- kT e -component had large uncertainties in rim regions.Therefore, we employed two- kT e -component model in the non-rim regions ( − ′ < R < ′ ),while one- kT e -component model in the rim regions. Figure 4 shows spectra extracted from redregions shown in figure 2. The top 2 panels are fitted with one-component model (NE3 andP10) while the bottom 8 panels are fitted with two-component model. The best-fit parametersare summarized in table 2. The residuals between the data and the model in bottom 4 panels(P5, 6, 7, 9) of figure 4 show some structures around 1.2keV. We should keep in mind thatthe two- kT e VNEI model still did not give us acceptable fits from the statistical point of view,which suggests that our model is too simple.Figure 5 shows kT e , τ and metal abundance as a function of R . We found that thehigh- kT e -component shows distinctly higher temperature and metal abundance compared tolow- kT e -component. The temperature of the low- kT e -component stays almost constant ( ∼ kT e -component has a peak value of 0.8 keV around R = − ′ and decrease towards the SW. We also found asymmetry in metal abundance ofhigh- kT e -component. The abundance of O[=C=N], Ne and Mg are higher in the NE part( R < ′ ) while Si is in the SW ( R > ′ ). The flux of each component is shown in top leftpanel of figure 5. The flux of the low- kT e -component is dominant in the NE part and italso shows shell brightening in both NE and SW rims, therefore, we confirm that low- kT e -component surrounds the Loop and the high- kT e -component fill its interior . All of these factssupport our assumption where low- kT e -component originates from the swept-up matter andthe high- kT e -component originates from the ejecta.
4. Discussion and Conclusion
We observed the Cygnus Loop from the NE to the SW with Suzaku in ten pointings.Our FOV covers more to the north of previous observation carried out by XMM-Newton andSuzaku. Dividing the entire FOV into 45 rectangular regions, we extracted spectra from all theregions, and performed spectral analysis to them. For spectral fit, we employed models used4y the previous observation (Katsuda et al. 2008; Tsunemi et al. 2007) and obtained similarresults. The one-component VNEI model showed fairly good fit in the rim of the Loop but notin the non-rim part of the Loop. The fit of non-rim part improved significantly by applyingtwo-component VNEI model. Judging from the abundanes and the flux distribution of eachcomponent, the high- kT e component must be the ejecta, while the low- kT e component comesfrom the swept-up matter. The temperature for the swept-up matter component is significantly lower than that forthe ejecta component. This temperature is similar to that obtained for the rim of the Loop,where we expect no contamination of the ejecta (e.g., Miyata et al. 2007). Therefore, webelieve that we surely separated the X-ray emission of the ejecta inside the Loop from that ofthe surrounding matter. The EM distribution of the swept-up matter is inhomogeneous in ourFOV, as shown in top right panel of figure 6. The shell of the swept-up matter seems to be thinaround 10 ′ < R < ′ relative to that in the NE part ( R < ′ ). The flux of swept-up matter alsoshows this tendency; its flux in the SW part is about a third of the NE part. Such trend is alsoreported by XMM-Newton (Tsunemi et al. 2007) and previous Suzaku observations (Katsuda etal. 2008). Tsunemi et al. (2007) divided XMM-Newton observation into north path and southpath and found that this thin shell region to be 5 ′ in south path and 20 ′ in north path. Theyestimated this thin shell region to have diameter of 1 ◦ and centering (20 h m s , ◦ ′ ′′ ).Our observation path goes though right in middle of this region and our result was consistentwith their prediction. Since the EM of this region is also about − of the NE part, and theEM of ejecta is stronger compared to the NE part, there might be a blowout in the directionof our line of sight, just like a blowout in the south of the Loop (Uchida et al. 2008) and weare only seeing the one side of the shell. We detected no ejecta emission from the NE rim( R ≤ − ′ ) and the SW rim ( R ≥ ′ ) of the Loop, which is about 15% of our FOV. The ejectaoccupy the major part of the inner side of the Loop while the swept-up matter surrounds it.This structure is identical to those in previous results (Miyata et al. 1998; Katsuda & Tsunemi2008). We have calculated the EM of various heavy elements in the ejecta such as O[=C=N],Ne, Mg, Si, and Fe [=Ni]. Figure 6 shows the EM distribution for these elements as a functionof R . The red marks show the results from our FOV while the green and the blue marks showthe results from XMM-Newton north path and south path. The comparison of these resultsshow some similarity near the both rim of the Loop ( R < − ′ , ′ < R ), but shows discrepancyin the center portion of the Loop. This is especially notable in Si and Fe. Since our FOV isfurther north from the geometric center of the Loop compared to XMM-Newton observation,these discrepancies suggests onion layer structures at a SN time of explosion. For Si and Fe,5ome structures can be seen in both Suzaku and XMM-Newton observations, such as the bumparound R = − ′ and the drop at 0 ′ < R < ′ .From the result from our FOV, the NE part and the SW part have different characteristicof metal distributions. We calculated the mass ratio in the NE part and those in the SW partfrom Figure 6. They are O ∼ .
7, Ne ∼ .
2, Mg ∼ .
3, Si ∼ .
36, Fe ∼ .
62 assuming unity fillingfactor. The O-Ne-Mg group is heavily distributed in the NE part of the Loop by factor of2 −
5. In contrast to the O-Ne-Mg group, the Si-Fe group is distributed more in the SW partof the Loop by factor of 2 −
3. These asymmetries can also be seen in all of the abundancedistribution (figure 5) except for Fe. However EM of the elements is the clearer representation ofthe amounts of elements, therefore we believed that Fe is also distributed asymmetrically moreto the SW. This asymmetry is quantitatively consistent from previous observation (Katsudaet al. 2008). A natural explanation for the asymmetry is an asymmetry at the time of theSN explosion of the Cygnus Loop. Recent theoretical models describe asymmetric supernovaexplosion resulting from hydrodynamic instability (Burrows et al. 2007).We calculated the mass of ejecta to be 21M ⊙ from EM. We estimated the plasma depthto be 26pc, unity filling factor and n e = 1 . n H , although the fossil ejecta might be deficientin hydrogen. If this is the case, the total mass of the ejecta reduces to ∼ ⊙ However thisvalue strongly depends on the plasma structure of the ejecta. Therefore we used SN explosionmodel to calculate the mass. In order to compare our data with the SN explosion models, wecalculated the ratios of elements relative to O. Figure 7 shows the number ratios of Ne, Mg,Si and Fe relative to O of the ejecta component. We plotted the core-collapse model (Woosley& Weaver 1995) for the various progenitor mass and Type Ia supernova model (Iwamoto et al.1999) for comparison. Since the distribution characteristic of elements showed large differencebetween the NE and the SW, we also plotted the number ratios calculated from each part. Theorange line represents the number ratios calculated from all of observations including XMM-Newton and previous Suzaku observations (Tsunemi et al. 2007, Katsuda et al. 2008). Thenumber ratios from the SW part shows fairly good agreement with Type Ia models in Mg andSi but that the number ratios of Ne and Fe show large inconsistency from the model. On thecontrary, the number ratios of Ne, Mg, Si, from NE are in agreement with 15M ⊙ model, butFe is about 4 times higher in our data. Because of the asymmetric distributions of ejecta, thenumber ratios of heavy elements give very different values by choosing which part of the Loopwe use. The number ratios calculated from our FOV and all FOV (2 Suzaku and 1 XMM-Newton) are both in good agreement with that of 12M ⊙ model. However, just like any otherdata, the number ratio of Fe shows disagreement. It is possible that the Fe is overabundantbecause all FOV are across the center of the Loop, where we expect to be Fe rich region.6 cknowledgements This work is partly supported by a Grant-in-Aid for Scientific Research by the Ministry ofEducation, Culture, Sports, Science and Technology (16002004). This study is also carried outas part of the 21st Century COE Program, ‘
Towards a new basic science: depth and synthesis ’.H.U. and S.K. are supported by JSPS Research Fellowship for Young Scientists. We also liketo thank Hiroko Kosugi for careful reading of this manuscript.7 ig. 1.
ROSAT HRI image of the entire Cygnus Loop. The Suzaku FOV (NE3, P1, P2, P3, P4, P5, P6,P7, P9, P10) are shown as white rectangles. Dotted circles and rectangles represent previous XMM-Newtonand Suzaku observations. able 1. Information of observations of the Cygnus Loop and Lockman Hole
Obs. ID Coordinate (RA, DEC) Obs. Date Effective ExposureCygnus Loop500022010 (NE3) 313.746, 32.188 2005.11.29 12.2 ks501012010 (P1) 313.510, 31.975 2007.11.13 9.8 ks501013010 (P2) 313.265, 31.779 2007.11.14 16.4 ks501014010 (P3) 313.032, 31.574 2007.11.14 7.5 ks501015010 (P4) 312.799, 31.369 2007.11.14 18.3 ks501016010 (P5) 312.547, 31.180 2007.11.15 19.3 ks501017010 (P6) 312.297, 30.991 2007.11.11 28.7 ks501018010 (P7) 312.078, 30.776 2007.11.12 21.0 ks501019010 (P9) 311.809, 30.603 2007.11.12 16.2 ks501020010 (P10) 311.566, 30.407 2007.11.13 14.6 ksLockman Hole100046010 (for NE3) 163.4063, 57.6108 2005.11.14 49.3 ks102018010 162.9257, 57.2581 2007.05.03 68.9 ks
Fig. 2.
Left: Surface brightness map of our FOV. Right: Three-color image of ten XIS FOV (Red: O
VII K α , Green: Fe L, Blue: Ne IX K α ). The data were binned by 8 pixels and smoothed by a Gaussian kernelof σ = 25 ′′ The scale R shows the distance from observation center in arcmin. The effects of exposure,vignetting, and contamination are corrected for both figure. able 2. Best-fit parameter for example spectra. NE and SW (P10) are fitted with one-component VNEI model while the others are fitted with two-component VNEImodel
Elements NE3 P1 P2 P3 P4 P5 P6 P7 P9 SW(P10)Free abundance component n H [ × cm − ] 3 . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . . ± . kT e [keV] . . . . . . . +0 . − . +0 . − . ± ± .
02 0.55 ± ± ± ± ± .
01 0.32 ± . . . . . . . . . . . . . . . ± . ± . . . . . . . . . . . . . . . . +0 . − . =O =O =O =O =O =O =O =O 0 . +0 . − . O . . . . . . . . . . . . . . +0 . − . ± .
07 0.73 ± .
06 1 . ± .
07 0.52 ± ± . ± . < . . ± .
02 0.16 +0 . − . Ne . . . . . . . . . . . . . +0 . − . . ± .
08 0.76 ± . ± .
07 0.18 ± ± .
03 0 . ± . < . . ± .
02 0.21 +0 . − . Mg . . . . . . . . . . . . +0 . − . . ± .
07 0.47 ± .
06 0 . ± .
07 0.13 ± ± .
02 0.16 ± .
03 0.12 ± . ± .
03 0.14 ± . . . . . . . . . . . . . . . ± .
09 0 . ± . ± . . ± . ± ± . ± . ± . . ± . ± . . . . . . . . . +0 . − . . ± .
05 0.66 ± . ± .
04 0.47 ± ± .
02 0.88 ± .
02 0.53 ± .
01 0 . ± .
02 0.19 ± τ [ × cm − s] 0.99 ± . ± .
06 1.00 ± .
04 1 . ± . +0 . − . +0 . − . +0 . − . +0 . − . . +0 . − . ± × cm − ] 4 . +0 . − . . ± .
01 0 . ± .
009 0 . +0 . − . . +0 . − . . ± .
003 0 . +0 . − . . +0 . − . . ± .
01 0.37 +0 . − . Fixed abundance component kT e [keV] . . . . . . . − − − ± ± ± .
01 0.27 ± .
01 0.34 ± ± ± ± . − − − Abundances . . . . − − − (fixed to those determined for the NE rim of the Cygnus Loop) † − − − τ [ × cm − s] . − − − . ± .
04 0.97 ± .
04 1 . ± .