nbhkdz.com冰点文库

XMM-Newton observations of the diffuse X-ray emission in the starburst galaxy NGC 253

时间:2011-08-18


Astronomy & Astrophysics manuscript no. ngc253di?use February 2, 2008

c ESO 2008

arXiv:0711.3182v1 [astro-ph] 20 Nov 2007

XMM-Newton observations of the diffuse X-ray emission in the starburst galaxy NGC 253?
M. Bauer1 , W. Pietsch1 , G. Trinchieri2 , D. Breitschwerdt3 , M. Ehle4 , M.J. Freyberg1 and A. Read5
1 2 3 4 5

Max-Planck-Institut f¨ ur extraterrestrische Physik, Giessenbachstra?e, 85741 Garching, Germany INAF Osservatorio Astronomico di Brera, via Brera 28, 20121 Milano, Italy Institut f¨ ur Astronomie der Universit¨ at Wien, T¨ urkenschanzstr. 17, A-1180 Wien, Austria XMM-Newton Science Operations Centre, ESA, P.O. Box 78, 28691 Villanueva de la Ca? nada, Madrid, Spain Department of Physics & Astronomy, University of Leicester, Leicester LE1 7RH, UK

Received XX XX 2007; accepted XX XX 2007.
ABSTRACT

Aims. We present a study of the di?use X-ray emission in the halo and the disc of the starburst galaxy NGC 253. Methods. After removing point sources, we analysed XMM-Newton images, hardness ratio maps and spectra from several regions in the halo and the disc. We introduce a method to produce vignetting corrected images from the EPIC PN data, and we developed a procedure that allows a correct background treatment for low surface brightness spectra, using a local background, together with closed ?lter observations. Results. Most of the emission from the halo is at energies below 1 keV. In the disc, also emission at higher energies is present. The extent of the di?use emission along the major axis of the disc is 13.6 kpc. The halo resembles a horn structure and reaches out to ?9 kpc perpendicular to the disc. Disc regions that cover star forming regions, like spiral arms, show harder spectra than regions with lower star forming activity. Models for spectral ?ts of the disc regions need at least three components: two thermal plasmas with solar abundances plus a power law and galactic foreground absorption. Temperatures are between 0.1 and 0.3 keV and between 0.3 and 0.8 keV for the soft and the hard component, respectively. The power law component may indicate an unresolved contribution from X-ray binaries in the disc. The halo emission is not uniform, neither spatially nor spectrally. The southeastern halo is softer than the northwestern halo. To model the spectra in the halo, we needed two thermal plasmas with solar abundances plus galactic foreground absorption. Temperatures are around 0.1 and 0.3 keV. A comparison between X-ray and UV emission shows that both originate from the same regions. The UV emission is more extended in the southeastern halo, where it seems to form a shell around the X-ray emission.

Key words. X-rays: galaxies – X-rays: ISM – Galaxies: individual: NGC 253 – Galaxies: halos – Galaxies: ISM – Galaxies: starburst

1. Introduction
The di?use X-ray emission of starburst galaxies can be quite prominent. Especially in galaxies that we see edge-on, we can ?nd very complex emission from galactic halos. One famous example is the starburst galaxy NGC 253 in the Sculptor Group. It is close enough (2.58 Mpc, 1′ =750 pc, Puche et al. 1991) to resolve structures in the disc and halo, and to separate the detected point sources from the di?use emission. Also, it is seen almost edge-on (78.5?, Pence 1980), so an unobscured analysis of the halo emission is possible. NGC 253 has been observed in X-rays many times. There are observations with Einstein (e.g. Fabbiano & Trinchieri 1984), ROSAT (e.g. Pietsch 1992; Read et al. 1997; Dahlem et al.
Send o?print requests to: M. Bauer, e-mail: mbauer@mpe.mpg.de ? Based on observations obtained with XMM-Newton, an ESA science mission with instruments and contributions directly funded by ESA Member States and NASA

1998; Vogler & Pietsch 1999; Pietsch et al. 2000), ASCA (e.g. Ptak et al. 1997), BeppoSAX (e.g. Cappi et al. 1999), XMM-Newton (e.g. Pietsch et al. 2001; Bauer et al. 2007), and Chandra (e.g. Weaver et al. 2002; Strickland et al. 2002, 2004a,b). While with some instruments one was not able to separate emission from point sources and di?use emission, other instruments, especially ROSAT, XMM-Newton, and Chandra, do have a narrow enough point spread function to do so. We here report on the ?rst extensive analysis of the di?use emission in NGC 253 with XMM-Newton.

2. Observations and data reduction
NGC 253 was observed with XMM-Newton (Jansen et al. 2001) during three orbits in June 2000 and June 2003, using all of the European Photon Imaging Camera (EPIC) instruments (Str¨ uder et al. 2001; Turner et al. 2001), the two co-aligned RGS spectrometers, RGS1 and RGS2 (den Herder et al. 2001),

2

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

Table 1. XMM-Newton EPIC PN NGC 253 observation log.

Nr. (1) 1 2 3 4

Revolution number (2) 89 89 186 646

Obs. id. (3) 0125960101 0125960201 0110900101 0152020101

Obs. dates (4) 2000-06-03 2000-06-04 2000-12-14 2003-06-19

Pointing direction RA/DEC (J2000) (5) 00:47:36.74 00:47:36.57 00:47:30.20 00:47:36.89 (6) -25:17:49.2 -25:17:48.7 -25:15:53.2 -25:17:57.3

P. A. (deg) (7) 56.9 57.0 233.8 53.8

Filter (8) Medium Thin Thin Thin

Texp (ks) (9) 60.8 17.5 24.4 113.0

Texp, clean (ks) (10) 24.3 3.1 4.4 47.9

and the Optical Monitor (Mason et al. 2001), for a total of about 216 ks. The revolution number, observation identi?er, observing date, pointings and orientation of the satellite (P.A.), and the total exposure times for the EPIC PN camera (Texp ) are shown in Table 1. Throughout the following analysis, we used the MOS data only to detect and remove point sources. We were especially interested in low-surface brightness di?use emission at energies below 1 keV, where the MOS detectors have a lower sensitivity than the EPIC PN. By not utilising the MOS data for the analysis of the di?use emission, we avoided a higher background noise level. We analysed the data using the Science Analysis System (SAS), version 7.0.0. In a ?rst step, we cleaned the EPIC PN and MOS observations by excluding times, where the count rate over the whole detector above 10 keV exceeded the count rate during quiescent times. This cleaning was done to avoid times with high particle background, which result in a higher background level. The observations showed no obvious additional times with solar wind charge exchange (e.g. Snowden et al. 2004), which would result in times with high background at energies below 1 keV, so no further exclusion of exposure time was necessary. The exposure times after screening for high background (Texp, clean ) are shown in Table 1. Summing up over the ?nal good time intervals, we ended up with 80 ks of exposure time in total. This means only about 37% of the original exposure time could be used for the analysis presented in this paper. This number is quite small, compared to typical exposure time fractions of usable times after screening of 60–70%1. Next, we screened for bad pixels that were not detected by the pipeline. In order to be able to merge images later on, we calculated sky coordinates (X,Y) for the events in all observations with respect to the centre reference position α2000 =00h 47m 33s.3, δ2000 =-25?17′ 18′′ . For the following analysis we splited the data set into ?ve energy bands: 0.2–0.5 keV, 0.5–1.0 keV, 1.0– 2.0 keV, 2.0–4.5 keV and 4.5–12 keV as bands 1 to 5.

2.1. Point source removal
In this paper we did not study the population of the point sources, but we focused on the di?use emission in the halo and the disc of the galaxy. To do so, we had to remove con1 see the XMM-Newton EPIC Background Working Group webpage http://www.star.le.ac.uk/?amr30/BG/BGTable.html

tributions from point sources. In order to run the source detection algorithm of the SAS-software package, we created images for the PN, selecting only single events (PATTERN=0) in energy band 1, and single and double events (PATTERN≤4) for the other bands. For MOS we used single to quadruple events (PATTERN≤12) in all bands. To avoid di?erences in the background over the PN detector, we omitted the energy range between 7.2 keV and 9.2 keV, where the detector background shows strong spatially variable ?uorescence lines (Freyberg et al. 2004). We created images, background images and exposure maps, and masked them to an acceptable detector area. The binning for all images is 2′′ . We searched for point sources in the ?eld of view (FOV), simultaneously in the 5 energy bands and three detectors. First, we searched in each observation separately, to correct for inaccuracies in the pointing positions. The resulting source lists were correlated to catalogues from USNO (Monet et al. 2003), SIMBAD2 , and Chandra (Strickland et al. 2002). O?sets were determined and applied to each observation. With the position corrected event ?les, we again created images on which we executed the ?nal point source detection. We searched in the merged images from observations 1, 2, and 4, and separately in the images from observation 3. The reason for merging only observations 1, 2 and, 4 is that observation 3 has a pointing o?set (?6′ ) into the northwestern halo and therefore we would have di?erent point spread functions on the same sky coordinates. Additionally, we created a point source catalogue for the Chandra observations. Point sources in ChandraObsID 3931 were identi?ed using the Wavelet-Based detection Algorithm (wavdetect in the ciao software, version 3.4, Freeman et al. 2002), in the 0.5–5.0 keV energy band using scales of 1′′ , 2′′ , 4′′ , 8′′ , and 16′′ . For ObsID 969 and ObsID 790 we adopted the published source list from Strickland et al. (2002). The combined XMM-Newton and Chandra source list will be published and further discussed in a forthcoming paper. The combined source list was used to remove the point sources from the data sets. The SAS-task region was used to produce elliptical regions that approximate the PSF with an analytical model at a given detector position and ?ux value (0.5 times the background ?ux at this position). Sources that were not detected in the XMM-Newton data sets, but are known from Chandra observations, were excluded with a circular region
2

http://simbad.u-strasbg.fr/simbad/

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

3

Fig. 1. Adaptively smoothed EPIC PN images with contours in the lower 4 energy bands: (top-left) 0.2–0.5 keV, (top-right) 0.5– 1.0 keV, (bottom-left) 1.0–2.0 keV, and (bottom-right) 2.0–4.5 keV. Contours are at (0.35, 0.50, 0.80, 1.6, 2.5, 6.0, 20, 100) × 10?5 ct s?1 pix?1 . Additionally we show the inclination corrected optical D25 ellipse in black. with a diameter of 8′′ . One might argue that these sources contribute only little to the overall emission. However, we took up a conservative position and also excluded these sources to keep any unwanted interference at a minimum. An extended source, most likely a galaxy cluster candidate in the background was additionally masked with a circular region of 1.5′ diameter.

4

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

Fig. 2. (left): Adaptively smoothed EPIC PN image with contours in the highest di?erent energy band (4.5–12 keV). Contour levels are the same as in Fig. 1. (right): The vignetting corrected exposure map of the merged four observations. The outer contour indicates 0 ks and the exposure increases linearly towards the centre by one seventh of the maximum (80 ks) per gray-scale level (0–11.4 ks, 11.4–22.9 ks, 22.9–34.3 ks, 34.3–45.7 ks, 45.7–57.1 ks, 57.1–68.6 ks, 68.6–80.0 ks). Except for a few pixels, all the detector gaps are covered by at least 4.4 ks in the central region.

2.2. Images
We used all 4 observations to produce images. The observations have di?erent pointing directions and position angles, so we obtained images where almost all the CCD gaps are ?lled. The single images, from the energy bands 1 to 5, were corrected for the detector background (electronic noise, high energy particles) by subtracting the surface brightness of the detector corners, that are outside of the ?eld of view. The images were exposure and vignetting corrected, and adaptively smoothed with a Gaussian kernel, with sizes between 10′′ and 47′′ (Fig. 1 and 2). For a detailed description of this method see App. B. A false-colour image was produced by combining the three lowest energy bands 1, 2, and 3, as channels red, green, and blue, respectively (Fig. 3). The image, after the point source removal is shown in Fig. 4.

Energy spectra of several regions (cf. Fig. 4) were extracted from the event ?les after removal of the point sources. To calculate the area of these regions, we used the task backscale. We produced background spectra using a region at the southwestern border of the FOV, together with observations where the ?lter wheel was closed. A detailed description of this procedure, which also handles the binning of the spectra, can be found in Appendix C. The ?nal, background subtracted source spectrum for each region has a signi?cance of at least 3 σ in each data bin. Since the emission is mostly con?ned to energies between 0.2 and 2.0 keV, we only calculated the hardness ratios HR1 and HR2, where HR1=(B2 -B1 )/(B2 +B1 ), and HR2=(B3 B2 )/(B3 +B2 ). B1 , B2 , and B3 are the count rates in the energy bands 1 to 3, i.e. 0.2–0.5 keV, 0.5–1.0 keV, and 1.0–2.0 keV, respectively. They were obtained by summing up the background subtracted counts in the spectra in the energy bands and observations. In order to ?t the spectra with physical models, we created the proper response and anxiliary response ?les for extended sources for each spectrum. In XSPEC 11.3.2, we linked the model parameters between the two observations and included a global renormalisation factor to account for di?erences between the observations to ?t the spectra from observations 1 and 4 simultaneously.

2.3. Hardness ratio maps and spectra
As a big advantage, compared to the observations by ROSAT and Chandra, the higher count rates in XMM-Newton allowed us to extract spectra with reasonable statistics from smaller regions in the disc and the halo. For the hardness ratios and spectra we again restricted ourselves to the EPIC PN data. We did not use observations 2 and 3 for hardness ratios and spectra, because after good time interval screening only little exposure was left (cf. Table 1).

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

5

Fig. 3. Adaptively smoothed EPIC PN image of NGC 253. The colours correspond to the energy bands (0.2–0.5 keV, red), (0.5–1.0 keV, green) , and (1.0–2.0 keV, blue). Overplotted in white is the inclination corrected optical D25 ellipse of NGC 253. Scale and orientation are indicated.

Fig. 4. Adaptively smoothed EPIC PN image of the di?use emission of NGC 253. Point sources have been removed. Overplotted in green are the regions that were used for extracting hardness ratios and spectra. The inclination corrected optical D25 ellipse is shown in white. gion 14) is shown in Fig. 6. The visible lines are from O VII (?0.57 keV), O VIII (0.65 keV), Fe XVII (0.73–0.83 keV), Ne IX (?0.91 keV), Ne X (1.0 keV), Mg XI (?1.3 keV), and Si XIII (?1.9 keV). To ?t the spectra, we tried several di?erent models, which all contain an absorption model (tbabs, Wilms et al. 2000) for the Galactic foreground NH of 1.3 × 1020 cm?2 (Dickey & Lockman 1990). Also the abundances were ?xed to solar values (see also Sec. 4.4) from Wilms et al. (2000). A simple one-temperature thin thermal plasma model (apec, Smith et al. 2001) did not result in a good ?t (i.e. χ2 ν ≤ 1.4) in any case. Similarly, a power law model did not give good ?ts. At least three components were necessary for most of the regions: two thin thermal plasmas plus a power law component. The power law was needed to account for the emission above ?1 keV and probably results from point sources below the point source detection limit, or incomplete source removal due to too small extraction radii. The obtained temperatures are quite uniform throughout the disc and vary from 0.1 to 0.3 keV and from 0.3 to 0.8 keV for the soft and the hard component, respectively. The intrinsic luminosity (corrected for Galactic absorption) of the di?use emission within the inclination corrected optical D25 ellipse is 2.4×1039 erg s?1 (0.2–10.0 keV), or 8.5×1038 erg s?1 (2.0– 10 keV). Both values were corrected for the area of cut-out point sources. The spectra decrease in hardness from the northwest to the southeast parallel to the minor axis, which can easily be seen in the hardness ratio maps (Fig. 5). This is not an e?ect caused by di?erent temperatures, but by the increasing strength of the

3. Results
To characterise the di?use emission in the disc and the halo, we analysed images in di?erent energy bands, and hardness ratios and spectra from several regions. In the disc the regions were chosen in a way that they follow the spiral arm structure. In the halo, we chose plane-parallel regions above the galactic disc. The projected heights of the halo regions are 2 kpc, with exception of the region furthest to the northwest (region 1), which has a projected height of 3 kpc. The regions are overplotted on top the false-colour X-ray image in Fig. 4. The hardness ratios in the di?erent regions are given in Table A.1 and shown graphically in Fig. 5.

3.1. Disc diffuse emission
The disc shows di?use emission in energies up to ?10 keV, where the harder emission is located close to the centre of NGC 253. The soft emission (<1 keV) shows the largest extent along the major axis. From the nucleus, it reaches ?7.0 kpc to the northwest and ?6.4 kpc to the southeast. A prominent feature in the disc is the lack of very soft emission northwest of the major axis. This is already known from ROSAT observations (e.g. Pietsch et al. 2000) and can be explained by the geometry of the system: The galaxy’s disc is oriented so that we see the underside of the disc. The emission from the northwestern halo behind the disc is therefore absorbed by the intervening disc material. The spectral properties in di?erent regions of the disc are summarised in Table A.1. An example of a disc spectrum (re-

6

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

soft spectral component towards the southeast (compared to the hard component), as the optical depth through the halo on the near side of the disc increases. The regions along the northwestern out?ow (regions 9 & 16) as well as the region northeast to that (region 13) allow a lower limit estimate of the absorption through the disc. Here an additional absorption component was required to achieve a good spectral ?t. The required column densities range between ?0.5×1022 cm?2 and ?0.9×1022 cm?2 .

In a few cases, the two thermal plasma plus a power law component model did not give the best ?t. The southern region on the northeastern end of the disc (region 12) did not require a power law component. It does not cover a spiral arm of NGC 253, thus the contribution from point sources below the detection limit may not be signi?cant. The region furthest to the south in the disc (region 21) was well ?t with one thermal plasma and a power law. A second thermal plasma was not required. We found that spectra are harder in regions which cover spiral arms. The northeastern regions of the disc (regions 11 & 12) are the best example for this. The region which covers 0.05 the spiral arm (region 11) shows temperatures of 0.18+ ?0.04 and +0.16 0.58?0.18 keV, whereas the region adjacent to the spiral arm (region 12) is signi?cantly cooler with temperatures of 0.07 ± 0.01 0.04 and 0.25+ ?0.03 keV. The latter spectrum is actually more similar to the typical halo spectrum (see next section).

3.2. Halo diffuse emission
The halo shows emission only below ?1 keV. Its projected maximum extent is ?9.0 kpc to the northwest, and ?6.3 kpc to the southeast, perpendicular to the major axis. The general shape resembles a horn structure. This was already seen with ROSAT (e.g. Pietsch et al. 2000), and Chandra (e.g. Strickland et al. 2002). In the northwestern halo, the EPIC PN images only show the eastern horn. In the southeastern halo, both the eastern and the western horn are visible in the energies between 0.2 and 0.5 keV. At higher energies, the western horn is not visible. On smaller scales the halo emission seems not to be uniformly distributed. It shows a ?lamentary structure, as was seen before in the ROSAT data. One notable feature is a brighter knot, which coincides with the nuclear out?ow axis in the northwestern halo at a height of about 3.5 kpc above the disc. It is bright in energies between 0.2 and 1.0 keV and appears yellow in the false-colour image (Fig. 3). We checked whether any of the detected structures coincide with chip gaps of the detector, and could therefore be arti?cial, but no correlation was found. The spectral properties in di?erent regions in the halo are summarised in Table A.1. To ?t the spectra, we applied the same approach as for the disc. Again, simple models cannot describe the spectra. A model with two thin thermal plasmas gave a good ?t in all regions in the halo. Unlike in the disc, no power law component was necessary in addition to the thermal components. Fig. 6. Representative spectra of a region in the disc (top, region 14) and of a region in the halo (bottom, region 7). The red and the black data points and model ?ts are from observations 1 and 4, respectively (see Table 1). The lower panel shows the residuals of the ?ts. The southeastern halo is softer than the northwestern halo, which results in redder colours in the southeastern halo in the EPIC PN false-colour image (Fig. 3), and also in lower values in HR1. A ?t to the spectrum of the whole northwestern halo 0.02 gave temperatures of 0.10 ± 0.01 and 0.33+ ?0.01 keV. The spectrum in the southeastern halo is similar, with temperatures of 0.03 0.09 ± 0.02 and 0.29+ ?0.04 keV. The di?erence in hardness is because the two plasma components contribute di?erent amounts. Compared to the normalisation of the hotter plasma, the normalisation of the cooler plasma is about 1.5 times stronger in the southeastern halo, with respect to the northwestern halo. The good statistics of the EPIC PN data allowed a further subdivision of the halo into smaller regions. A representative example (region 7) of one of these halo spectra is shown in Fig. 6. The oxygen lines at 0.57 keV (O VII) and 0.65 keV (O VIII) are prominent. Also visible is the iron line at ?0.8 keV (Fe XVII). The halo is not uniform in its spectral properties on smaller scales. The northwestern halo is softer in the east than in the west, while the southeastern halo is softer further away from

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

7

Fig. 5. Hardness ratio maps: The map was binned to the same regions as in Fig. 4. The higher the index, the harder the spectrum. The background colour corresponds to an arti?cially set value. We also show the inclination corrected optical D25 ellipse. (left): HR1=(B2 -B1 )/(B2 +B1 ), where B1 and B2 are the count rates in the energy bands 0.2–0.5 keV and 0.5–1.0 keV, respectively. (right): HR2=(B3 -B2 )/(B3 +B2 ), where B2 and B3 are the count rates in the energy bands 0.5–1.0 keV and 1.0–2.0 keV, respectively. the disc (see HR1 map in Fig. 5). Additionally the emission hardens along the direction of the northwestern out?ow (regions 3 & 6). However, the signi?cance of this di?erence is only 1.9 σ. The total intrinsic luminosity for the di?use emission, corrected for the area of the removed point sources, in the northwestern halo is 8.4×1038 erg s?1 (0.2–1.5 keV), compared to 2.1×1038 erg s?1 in the southeastern halo. To calculate electron densities, we assumed a volume for the emitting region. We modeled the northwestern halo with a cylinder with a radius of 4.3 kpc and a height of 4.2 kpc, plus a cylindrical segment with a height of 4.2 kpc, a radius of 4.3 kpc and a width in the southeast-northwest direction of 3.0 kpc (to model region 1). This gives a volume of 298 kpc3 or 8.7 × 1066 cm3 . For the southeastern halo we assumed a cylinder with a radius of 3.5 kpc and a height of 2.0 kpc, plus a cylindrical segment with a height of 2.0 kpc, a radius of 3.5 kpc and a width in the southeast-northwest direction of 3.0 kpc (region 25), resulting in a volume of 113 kpc3 or 3.3 × 1066 cm3 . To calculate densities and the total mass in the emission regions, we corrected the volumes for the cut-out point sources. Using the emission measure of the ?t (cf. the documentation of the apec model in XSPEC), the resulting densities are 3.2 η?0.5 × 10?3 cm?3 and 4.7 η?0.5 × 10?3 cm?3 for the northwestern and southeastern halo, respectively. η is the volume ?lling factor (η ≤ 1). With solar abundances from Wilms et al. (2000), this implies total masses of 3.3 η?0.5 × 107 M⊙ and 1.8 η?0.5 × 107 M⊙ for the northwestern and southeastern halo, respectively.

4. Discussion

4.1. The extent of the diffuse emission of NGC 253
Extended emission from the soft northwestern halo was ?rst reported from Einstein observations (Fabbiano 1988). Later, observations with ROSAT also discovered the southeastern halo in X-rays (e.g. Pietsch et al. 2000). The ROSAT images in the soft band trace the emission in the outer halo to projected distances of up to 9 kpc, both in the northwest and the southeast direction. With XMM-Newton, the emission is detected out to 9.0 kpc to the northwest and 6.3 kpc to the southeast. This di?erence in the southeastern halo can be explained by the high ROSAT sensitivity extending down to 0.1 keV. The useful XMM-Newton EPIC PN range is limited to 0.2 keV. This makes a big di?erence as there are many strong lines from O IV, Ne VIII, Mg IX, Mg X, Si IX, and Si X in the energy band between 0.1 and 0.2 keV. For a thermal plasma at a temperature of ?0.1 keV, these lines are even stronger than the O VII and O VIII lines, and about 60% of the total ?ux in the energy band from 0.1 to 2.0 keV originates from lines below 0.2 keV. The southeastern halo shows softer emission than the northwestern halo, therefore, the e?ect is strongest in the southeastern halo. Also in the disc the extent of the emission is di?erent. The ROSAT images trace the soft emission ?6.8 kpc towards the

8

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

northeast and ?5.3 kpc towards the southwest. With XMMNewton, the disc emission has an extent of ?7.2 kpc and ?6.3 kpc to the northeast and southwest, respectively. The disc spectra are harder than the halo spectra, and therefore the higher XMM-Newton sensitivity at energies >0.4 keV comes into play.

4.2. Is the diffuse emission in the disc really from hot interstellar gas?
The cumulative emission of a large population of weak stellartype X-ray sources can mimic the characteristics of a hot interstellar gas component. This was ?rst discovered in the Milky Way’s ridge X-ray emission (e.g. Revnivtsev et al. 2006), who found evidence that the bulk of the Galactic ridge X-ray emission is composed of weak X-ray sources, mostly cataclysmic variables and coronally active stars in binary systems, with a luminosity of most of these sources of less than 1031 erg s?1 . Also in other galaxies, Revnivtsev et al. (2007) found that the apparently di?use emission is consistent with the emission from an old stellar population like in the Milky Way. Can this also explain the extended X-ray emission in the disc of NGC 253? Following the method by Revnivtsev et al. (2007), we used K-band observations, to infer the emissivity of the di?use X-ray component per unit stellar mass. We derived the near-infrared luminosity and stellar mass of NGC 253, using the total K-band magnitude of 3.772 (Jarrett et al. 2003), the distance modulus of 27.06, corrected for interstellar extinction of 0.007 (Schlegel et al. 1998), and the colourdependent K-band mass-to-light ratio from (Bell & de Jong 2001, log( M? /LK ) = ?0.692 + 0.652 × ( B ? V )), with ( B ? V ) = ?0.16 (Comer? on et al. 2003). This yielded a total Kband luminosity LK = 1.6 × 1011 L⊙ and a total stellar mass M? = 2.6 × 1010 M ⊙ . With a X-ray luminosity of NGC 253 of 2.0×1039 erg s?1 (0.5–10 keV), the emissivity of the di?use X-ray component per unit stellar mass then resulted in L0.5?10 keV 0.9 28 ?1 = 7 .6 + erg s?1 M⊙ . ?0.3 (±2.3) × 10 M? The errors are statistical errors on the measured X-ray ?ux. Additionally, we assumed an uncertainty of ?30% (given in parentheses), which might be associated with the LK to M? conversion (Bell et al. 2003). The emissivity of NGC 253 should only be considered a lower limit. We cut out a quite large region in the centre of NGC 253 and corrected for this by ?lling the hole with the average ?ux of the disc. Therefore, the obtained X-ray luminosity as well as the emissivity are probably too small. From the luminosity and other properties of the Galactic ridge X-ray emission (e.g. Revnivtsev et al. 2006) and from direct measurements of the luminosity function of sources in the solar neighbourhood (Sazonov et al. 2006), the combined 0.5–10 keV emissivity of cataclysmic variables and coronally active stars has been estimated as LX /M? ? 1.2 ± 0.3 × ?1 1028 erg s?1 M⊙ . The value derived for NGC 253 is larger than the value for the Milky Way, indicating the presence of a hot gaseous component.

An even stronger argument is the following: if the di?use X-ray emission is produced by an old stellar population, then their morphologies should be similar. A comparision of the X-ray emission with the 2MASS K-band image (Jarrett et al. 2003) is shown in Fig. 7. We found that the X-ray morphology does not match the K-band morphology, therefore the di?use emission is indeed not simply due to an old stellar population, but has to have a truly di?use component.

4.3. Spectral ?ts and variations in the halo
As it was mentioned already in earlier publications, there is an ambiguity in the spectral ?ts between a pure multi-temperature thermal plasma model and a combination of thermal plasmas plus a power law component (e.g. Dahlem et al. 2000; Strickland et al. 2002). This ambiguity in the halo emission still exists with the XMM-Newton data. Fits to the halo spectra with a thermal plasma plus a power law model (see Table A.2) resulted in similar χ2 ν , as for a multi-temperature thermal plasma model. A power law component from point sources could be excluded, since we were careful to remove any point source contribution. Another source for non-thermal emission could be synchrotron emission from cosmic ray electrons that are advected with the superwind or are accelerated locally in internal wind shocks. A comparison of the X-ray emission to the 330 MHz and 1.4 GHz radio emission (Carilli et al. 1992) showed that the radio emission is more extended, and does not show the horn structure that we see in X-rays. Because of this inequality, we prefer the multi-temperature thermal plasma model for the X-ray halo emission at the moment. A currently ongoing analysis with non-equilibrium models (e.g. Breitschwerdt & Schmutzler 1999) might also be able to explain the observations. The northwestern halo shows signi?cant hardness variations in HR1, as opposed to the ?ndings by Strickland et al. (2002). We checked if this can be caused by a di?erent energy band selection, but the result is independent whether we use the bands from Strickland et al. (2002) (0.3–0.6 keV and 0.6– 1.0 keV) or our own. These hardness variations might also be a sign of non-equilibrium ionisation (NEI) X-ray emission.

4.4. Temperatures, abundances and column densities
The X-ray emission from NGC 253 has been observed before with several other X-ray observatories. Especially the early observatories did not allow to separate the point sources from the di?use emission since the point spread function was quite large. Hence, only a combined ?t of the emission from the halo, the disc, and the nuclear region was possible. Temperatures of multi-temperature models ranged between 0.1 and 0.3 keV for the low, and between 0.6 and 0.7 keV for the high temperature component (Dahlem et al. 1998; Weaver et al. 2000; Dahlem et al. 2000). Reported abundances were mostly highly subsolar and therefore unphysical for a supposedly metal enriched starburst galaxy plasma. Only the X-ray observatories like XMM-Newton, Chandra, and to some degree ROSAT allow us to separate the halo from

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

9

Fig. 7. Comparison between the K-band (left, 2MASS, Jarrett et al. 2003) and X-ray (right) morphology. The white contours overplotted on the X-ray image represent the K-band brightness levels. Both images are on a linear colour scale, and on the same spatial scale as indicated in the K-band image. the disc emission and to remove contribution by point sources via a spatial selection. From ROSAT data, Pietsch et al. (2000) inferred a foreground absorbed two-temperature thermal model with temperatures of 0.13 and 0.62 keV for the northwestern halo emission. No highly subsolar abundances were required. The disc emission could be explained by a 0.7 keV thermal plasma and an additional thermal plasma (kT=0.2 keV) in front of the disc coming from a coronal component. The ?rst XMM-Newton results by Pietsch et al. (2001) required a two-component model for the disc emission with temperatures of 0.13 and 0.4 keV plus residual harder emission, possibly from unresolved point sources. For the nuclear region three temperatures were needed (0.6, 0.9, and 6 keV). Both models used solar abundances. No analysis of the halo emission was presented in their paper. The best spatial resolution is provided by the Chandra observatory. Results on the di?use disc and halo emission were ?rst published by Strickland et al. (2002). For the halo emission they needed a multi-temperature model (apec) with at least two temperatures of 0.24 and 0.71 keV (the latter with quite large errors) and with a foreground absorption of 5.3 × 1020 cm?2 . A power law (Γ = 3.3) plus a thermal plasma (kT=0.24 keV) gave a similarly good ?t. A combination of other thermal models (mekal) or non-equilibrium models (vnei) did not result in better ?ts. The di?use emission from the disc was ?tted with the same models, however, the temperatures were lower than in the halo, with 0.17 and 0.56 keV, respectively. The foreground absorption yields 4.7 × 1020 cm?2 . In all cases unphysically sub-solar abundances had to be assumed. The temperature values for the halo emission, as found by our analysis, are lower than the ones from previous observations. Our soft component is about 0.10 keV, which is still compatible with the ROSAT results. However, the hard component is only ?0.32 keV for the northwestern halo and ?0.29 keV for the southeastern halo. A higher temperature was not necessary in any of our ?ts. A possible explanation could be the way the spectra were background subtracted. We used a sophisticated method (see App. C) that uses the local background at the border of the ?eld of view, where no emission from NGC 253 is expected, while other authors used e.g. blank-sky observations (Strickland et al. 2002). Using a background from di?erent times and di?erent ?elds on the sky can lead to systematic e?ects in the background substraction. A background region with a higher contribution of the local bubble could, for example, lead to an over-correction, especially at very soft energies (<0.5 keV). In the disc we found temperatures between 0.1 and 0.3 keV and between 0.3 and 0.8 keV, for the soft and the hard component, respectively. This is consistent with earlier results. We also tried to constrain the metal abundances in our ?ts. However the errors on the obtained values are so large, that we are not able to give well constrained abundances (northwest4.7 +4.6 ern halo: Z = 0.3+ ?0.2 Z ⊙ , southeastern halo: Z = 0.4?0.3 Z ⊙ , 0.9 Z ). Since we do not expect highly subsodisc: Z = 1.0+ ?0.7 ⊙ lar abundances in an environment which is enriched with metals from the starburst via the superwind and galactic fountains, we ?xed the abundances in our analysis to solar. This is very well consistent with the above values. A reason for the low abundances, found with di?erent instruments, could be that due to a lower spatial resolution and or/and sensitivity more point sources contribute to the ?nal spectrum, increasing the continuum ?ux. The ratio of line emission to continuum ?ux is therefore decreased, which mimics the spectral shape of a plasma with low metal abundances. A similar e?ect can be achieved when a NEI spectrum is ?tted with CIE models. Also, a too

10

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

Fig. 8. Two-colour UV image of NGC 253 with NUV (175– 280 nm) in yellow and FUV (135–175 nm) in blue. The intensity was rescaled to emphasise the faint, di?use emission. Overplotted are the 0.2–0.5 keV X-ray contours.

simplistic model could be the reason, combining regions with di?erent temperatures. In the disc, three of the regions required an extra absorption component in the spectral model. The additional column densities range between 0.5×1022 cm?2 and 0.9×1022 cm?2 . Direct radio measurements of the H I column density showed lower values than we derived from the X-ray data. An interpolation of the H I maps by Puche et al. (1991) and Koribalski et al. (1995) resulted in ?2.4, 3.9, and 3.4×1021 cm?2 for regions 9, 13, and 16, respectively. However, the H I value for the region including the nucleus of NGC 253 (region 16) is a?ected by H I absorption, so the resulting column density can only be considered as a lower limit. Additional absorption is expected from molecular hydrogen. Mauersberger et al. (1995) derived the H2 column density in the direction of the nucleus of NGC 253 to 3.7×1023 cm?2 . Taking this value as an upper limit for the column density in the disc regions around the nucleus, the column densities derived from X-ray spectra are within the limits from radio observations.

horn structure, where the UV horn extends to about 7 kpc away from the disc. However, the FUV horn is slightly o?set by ?700 pc to the northeast with respect to the X-ray horn. Images obtained with the Optical Monitor onboard XMM-Newton are not sensitive enough to show the extraplanar UV emission. Hoopes et al. (2005) proposed the following model for the origin of the UV emission: Since the UV luminosities are too high to to be produced by continuum and line emission from photoionized or shock-heated gas, the UV emission could be explained by dust in the out?ow that scatters the stellar continuum from the starburst into our line of sight. They also found that the UV halo emission, as seen with GALEX, correlates with the Hα emission, which could originate from gas that is photoionized by UV photons from the starburst. The UV and Hα emission would originate in the same cold regions in the halo. How does the warm gas that is responsible for the Hα and UV emission get out into the halo? There are two possibilities: either it has already been there from the beginning in the form of a cold and maybe clumpy halo component, or it was transported by the superwind and galactic fountains from the disc out into the halo. There are models, where it is possible to drag up clouds of cold gas into the halo (e.g. model 3 of Strickland et al. 2002). In a sheet surrounding these clouds, X-ray emission could be produced by shocks or in conductive or turbulent mixing interfaces on the cloud surface. This model would also account for the non-uniformity of the X-ray emission as seen in the XMMNewton images (Fig. 1 and 3). However, the model cannot explain the displacement of the UV emission in the southeastern halo, since in the model the clouds are located within or at the inner border of the superwind. Could the dust even survive this transport from the disc into the halo embedded in a hot plasma environment? Draine & Salpeter (1979) give the sputtering time for a spherical dust grain of radius a embedded in a plasma of hydrogen with temperatures between 106 and 109 K and the density nH as n H ?1 a yr. (1) tsput ? 106 ?m cm?3 For nH between 2.5 × 10?2 cm?3 in the out?ow close to the centre (Bauer et al. 2007) and 3.2 × 10?3 cm?3 out in the northwestern halo, and a grain size of a=0.1 ?m, tsput varies between 4.0 and 31 Myr. So to reach a height above the disc of 7.5 kpc in less than 31 Myr, an average velocity of at least 240 km/s is required. This is well compatible with measurements of out?ow velocities in di?erent wavelengths, that range from 260 km/s (Na D absorption, Heckman et al. 2000) to about 400–600 km/s (Hα, N II, S II, and O II emission, Ulrich 1978; Demoulin & Burbidge 1970). Therefore it is quite possible that the dust survives the transport from the disc out into the halo. Another model to explain the UV and X-ray morphology (e.g. model 5 of Strickland et al. 2002) requires a thick disc component, through which the superwind emerges into the halo. On the contact surface between the hot superwind ?uid and the cold thick disc material we get a heated layer through

4.5. X-ray versus UV morphology
Fig. 8 shows the X-ray contours from the energy band 0.2– 0.5 keV overplotted on a two-colour UV image, taken with the GALEX observatory (Galaxy Evolution Explorer, a UV space telescope) on 2003 October 13. For the northwestern halo there is quite a good agreement between the FUV and X-ray emission regions. The FUV emission traces the western horn to a distance of ?7.5 kpc above the disc, as well as the broad base emission in soft X-rays quite well. In the southeastern halo, again, the UV and the soft X-ray emission show the western

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

11

shocks and turbulent mixing where the X-rays are produced, surrounded on the outside by a colder layer where the UV emission originates. The thick disc component was originally created by lifting material up from the disc through the star formation activity (simulations by Rosen & Bregman 1995). This model would easily explain the UV displacement from the Xrays, however we would only get a hollow cone with X-ray emission. The latter is not what we see in the XMM-Newton observations. Though a mix of both models would be able to explain the observed morphology. The magnetohydrodynamics ISM simulations of de Avillez & Breitschwerdt (2005) also shows a clumpy halo structure, characterized by turbulent mixing layers, which could explain the UV and X-ray ?lamentary structure. In some regions, the magnetic ?eld forms loops surrounded by shells which may exhibit enhanced UV emission.

Science Institute (MAST). STScI is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555. Support for MAST for non-HST data is provided by the NASA O?ce of Space Science via grant NAG5-7584 and by other grants and contracts. MB acknowledges support from the BMWI/DLR, FKZ 50 OR 0405.

References
Bauer, M., Pietsch, W., Trinchieri, G., et al. 2007, A&A, 467, 979 Bell, E. F. & de Jong, R. S. 2001, ApJ, 550, 212 Bell, E. F., McIntosh, D. H., Katz, N., & Weinberg, M. D. 2003, ApJS, 149, 289 Breitschwerdt, D. & Schmutzler, T. 1999, A&A, 347, 650 Cappi, M., Persic, M., Bassani, L., et al. 1999, A&A, 350, 777 Carilli, C. L., Holdaway, M. A., Ho, P. T. P., & de Pree, C. G. 1992, ApJ, 399, L59 Comer? on, F., G? omez, A. E., & Torra, J. 2003, A&A, 400, 137 Dahlem, M., Parmar, A., Oosterbroek, T., et al. 2000, ApJ, 538, 555 Dahlem, M., Weaver, K. A., & Heckman, T. M. 1998, ApJS, 118, 401 de Avillez, M. A. & Breitschwerdt, D. 2005, A&A, 436, 585 Demoulin, M. H. & Burbidge, E. M. 1970, ApJ, 159, 799 den Herder, J. W., Brinkman, A. C., Kahn, S. M., et al. 2001, A&A, 365, L7 Dickey, J. M. & Lockman, F. J. 1990, ARA&A, 28, 215 Draine, B. T. & Salpeter, E. E. 1979, ApJ, 231, 77 Fabbiano, G. 1988, ApJ, 330, 672 Fabbiano, G. & Trinchieri, G. 1984, ApJ, 286, 491 Freeman, P. E., Kashyap, V., Rosner, R., & Lamb, D. Q. 2002, ApJS, 138, 185 Freyberg, M. J., Briel, U. G., Dennerl, K., et al. 2004, in XRay and Gamma-Ray Instrumentation for Astronomy XIII. Edited by Flanagan, Kathryn A.; Siegmund,Oswald H. W. Proceedings of the SPIE, Volume 5165., ed. K. A. Flanagan & O. H. W. Siegmund, 112–122 Gehrels, N. 1986, ApJ, 303, 336 Heckman, T. M., Lehnert, M. D., Strickland, D. K., & Armus, L. 2000, ApJS, 129, 493 Hoopes, C. G., Heckman, T. M., Strickland, D. K., et al. 2005, ApJ, 619, L99 Jansen, F., Lumb, D., Altieri, B., et al. 2001, A&A, 365, L1 Jarrett, T. H., Chester, T., Cutri, R., Schneider, S. E., & Huchra, J. P. 2003, AJ, 125, 525 Koribalski, B., Whiteoak, J. B., & Houghton, S. 1995, Publications of the Astronomical Society of Australia, 12, 20 Mason, K. O., Breeveld, A., Much, R., et al. 2001, A&A, 365, L36 Mauersberger, R., Henkel, C., & Chin, Y.-N. 1995, A&A, 294, 23 Monet, D. G., Levine, S. E., Canzian, B., et al. 2003, AJ, 125, 984 Pence, W. D. 1980, ApJ, 239, 54 Pietsch, W. 1992, in Physics of Nearby Galaxies: Nature or Nurture?, ed. T. X. Thuan, C. Balkowski, & J. Tran Thanh

5. Summary
We have characterised the di?use emission in NGC 253. The disc extends 13.6 kpc along the major axis and shows emission between 0.2 and 10 keV. The spectrum could be modelled with two thermal plasmas (T cold = 0.1 ? 0.3 keV and T hot = 0.3 ? 0.8 keV) with solar abundances plus a power law component and galactic foreground absorption. The power law component may indicate an unresolved contribution from X-ray binaries in the disc. The total luminosity of the di?use emission in the disc is 2.4×1039 erg s?1 (0.2-10.0 keV). We found clear evidence for hot plasma in the disc. The di?use emission does not originate completely from an old stellar population. The halo resembles a horn structure which reaches out to a projected height of ?9 kpc perpendicular to the disc. The halo emission on smaller scales seems not to be uniformly distributed, but shows a ?lamentary structure. The southeastern halo is softer than the northwestern halo. To model the spectra in the halo we needed two thermal plasmas (T cold ? 0.1 keV and T hot ? 0.3 keV) with solar abundances plus galactic foreground absorption. The total luminosity of the di?use emission is 8.4×1038 erg s?1 and 2.1×1038 erg s?1 (0.2-1.5 keV) in the northwestern and southeastern halo, respectively. Densities computed to 3.2 η?0.5 × 10?3 cm?3 and 4.7 η?0.5 × 10?3 cm?3 , with the volume ?lling factor η. With solar abundances this implies total masses of 3.3 η?0.5 × 107 M⊙ and 1.8 η?0.5 × 107 M⊙ for the northwestern and southeastern halo, respectively. A comparison between X-ray and UV emission showed that both originate from the same regions. The UV emission is more extended in the southeastern halo, where it seems to form a shell around the X-ray emission.
Acknowledgements. The XMM-Newton project is supported by the Bundesministerium f¨ ur Wirtschaft und Technologie/Deutsches Zentrum f¨ ur Luft- und Raumfahrt (BMWI/DLR, FKZ 50 OX 0001), and the Max-Planck Society. This research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. This research has made use of the SIMBAD database, operated at CDS, Strasbourg, France. The GALEX data presented in this paper were obtained from the Multimission Archive at the Space Telescope

12

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

van, 67 Pietsch, W., Roberts, T. P., Sako, M., et al. 2001, A&A, 365, L174 Pietsch, W., Vogler, A., Klein, U., & Zinnecker, H. 2000, A&A, 360, 24 Ptak, A., Serlemitsos, P., Yaqoob, T., Mushotzky, R., & Tsuru, T. 1997, AJ, 113, 1286 Puche, D., Carignan, C., & van Gorkom, J. H. 1991, AJ, 101, 456 Read, A. M., Ponman, T. J., & Strickland, D. K. 1997, MNRAS, 286, 626 Revnivtsev, M., Churazov, E., Sazonov, S., Forman, W., & Jones, C. 2007, ArXiv Astrophysics e-prints Revnivtsev, M., Sazonov, S., Gilfanov, M., Churazov, E., & Sunyaev, R. 2006, A&A, 452, 169 Rosen, A. & Bregman, J. N. 1995, ApJ, 440, 634 Sazonov, S., Revnivtsev, M., Gilfanov, M., Churazov, E., & Sunyaev, R. 2006, A&A, 450, 117 Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525 Smith, R. K., Brickhouse, N. S., Liedahl, D. A., & Raymond, J. C. 2001, ApJ, 556, L91 Snowden, S. L., Collier, M. R., & Kuntz, K. D. 2004, ApJ, 610, 1182 Strickland, D. K., Heckman, T. M., Colbert, E. J. M., Hoopes, C. G., & Weaver, K. A. 2004a, ApJS, 151, 193 Strickland, D. K., Heckman, T. M., Colbert, E. J. M., Hoopes, C. G., & Weaver, K. A. 2004b, ApJ, 606, 829 Strickland, D. K., Heckman, T. M., Weaver, K. A., Hoopes, C. G., & Dahlem, M. 2002, ApJ, 568, 689 Str¨ uder, L., Briel, U., Dennerl, K., et al. 2001, A&A, 365, L18 Turner, M. J. L., Abbey, A., Arnaud, M., et al. 2001, A&A, 365, L27 Ulrich, M.-H. 1978, ApJ, 219, 424 Vogler, A. & Pietsch, W. 1999, A&A, 342, 101 Weaver, K. A., Heckman, T. M., & Dahlem, M. 2000, ApJ, 534, 684 Weaver, K. A., Heckman, T. M., Strickland, D. K., & Dahlem, M. 2002, ApJ, 576, L19 Wilms, J., Allen, A., & McCray, R. 2000, ApJ, 542, 914

Appendix A: Spectral ?ts

Table A.1. Spectral ?ts (multi-temperature thermal plasma, plus power law component in the disc) and hardness ratios in the extraction regions.
χ2 kT norm NH kT norm Γ norm HR1 HR2 ν (χ2 /ν) (keV) (10?5 ) (1022 cm?2 ) (kev) (10?5 ) (10?5 ) +0.02 +8.5 +0.06 +1.6 1 1.0 (21.1/21.0) 0.10? 19.4? ... 0.35? 5.6? ... ... 0.03 ± 0.04 ?0.45 ± 0.03 0.02 8.3 0.05 1.5 +1.0 +0.06 +4.2 +0.03 ... ... ?0.02 ± 0.04 ?0.44 ± 0.04 3.0?0.9 ... 0.32?0.05 8.3?3.8 2 0.8 (14.6/18.0) 0.10?0.02 +0.9 +0.01 +6.0 +0.05 ... ... 0.12 ± 0.03 ?0.57 ± 0.02 3 0.9 (35.3/39.0) 0.09? 11.6? ... 0.34? 4.6? 0.02 0.03 0.9 5.0 +0.04 +3.9 +0.07 +0.9 4 0.7 (16.5/25.0) 0.11? 5 . 7 . . . 0 . 33 3 . 3 . . . . . . 0 .15 ± 0.04 ?0.49 ± 0.03 0.03 ?2.9 ?0.04 ?0.9 +0.6 +0.04 +4.1 +0.02 . . . . . . ? 0 . 02 ± 0.03 ?0.57 ± 0.02 3 . 4 . . . 0 . 30 11 . 5 5 0.8 (31.1/41.0) 0.10? ?0.9 ?0.04 ?2.7 0.02 +0.9 +0.01 +3.3 +0.03 ... ... 0.05 ± 0.02 ?0.61 ± 0.03 6 0.9 (82.5/90.0) 0.10? 15.4? ... 0.32? 6.0? 0.01 4.1 0.02 0.8 +0.9 +0.03 +0.02 +3.7 ... 0.31? 5.0? ... ... 0.13 ± 0.03 ?0.57 ± 0.02 8.2? 7 0.9 (52.0/59.0) 0.11? 1.1 0.03 2.8 0.03 8 ... ... ... ... ... ... ... ... 0.38 ± 0.04 ?0.24 ± 0.04 +0.08 +1.2 +0.391 +0.08 +4.7 +0.54 +0.4 9 0.8 (36.4/47.0) 0.17? 1.1? 0.496? 0.32? 9.2? 1.12? 0.9? 0.44 ± 0.04 ?0.30 ± 0.03 0.07 0.9 0.07 4.9 0.43 0.6 0.169 10 ... ... ... ... ... ... ... ... 0.55 ± 0.06 ?0.15 ± 0.04 +0.05 +1.4 +0.16 +0.8 +0.53 +0.8 11 0.6 (47.0/78.0) 0.18? 3.4? ... 0.58? 1.9? 2.54? 1.2? 0.14 ± 0.03 ?0.52 ± 0.02 0.04 1.3 0.18 0.8 1.2 0.76 +0.01 +12.0 +0.04 +0.6 12 1.4 (43.3/32.0) 0.07? 17 . 2 . . . 0 . 25 2 . 7 . . . . . . ? 0 .05 ± 0.04 ?0.52 ± 0.03 0.01 ?9.6 ?0.03 ?0.6 +0.07 +0.7 +1.095 +0.91 +4.8 +0.78 +0.7 13 0.8 (41.1/53.0) 0.30? 1 . 9 0 . 892 0 . 77 4 . 2 0 . 67 0 . 7 0 .46 ± 0.04 ?0.22 ± 0.04 ?1.0 ?0.408 ?3.8 0.06 ?0.45 ?0.67 ?0.6 +0.05 +1.0 +0.06 +1.1 +0.23 +0.6 14 0.5 (149.0/275.0) 0.20? 4.3? ... 0.59? 5.7? 2.12? 2.2? 0.27 ± 0.02 ?0.57 ± 0.02 0.31 0.05 1.6 0.05 1.6 0.5 +0.9 +0.31 +0.5 +0.15 +0.05 +0.9 0.03 ± 0.02 ?0.65 ± 0.02 1.0? ... 0.59? 1.6? 3.13? 2.8? 15 0.7 (80.2/116.0) 0.20? 0.32 0.4 1.4 0.06 1.3 0.15 +0.4 +0.239 +0.11 +3.7 +0.19 +0.08 +2.1 2 . 0 0 .43 ± 0.02 ?0.24 ± 0.02 16 (NW out?ow) 0.8 (126.1/150.0) 0.25? 1 . 8 0 . 542 0 . 59 9 . 1 1 . 38 ?0.12 ?0.18 0.07 ?1.0 ?0.179 ?0.5 ?5.2 +0.06 +1.6 +0.08 +1.2 +0.16 +0.5 17 (Centre) 0.8 (187.1/249.0) 0.24? 2 . 6 . . . 0 . 58 2 . 8 1 . 73 3 . 3 0 . 26 ± 0.02 ?0.43 ± 0.02 0.04 ?1.4 ?0.12 ?1.1 ?0.11 ?0.6 +0.04 +1.2 +0.19 +1.0 +0.27 +0.5 18 (SE out?ow) 0.5 (82.1/152.0) 0.24? 2.9? ... 0.66? 1.2? 2.28? 1.8? 0.14 ± 0.02 ?0.53 ± 0.02 0.04 1.3 0.22 0.8 0.33 0.5 +0.05 +0.8 +0.10 +0.8 +0.55 +0.5 1.09? 1.1? 0.34 ± 0.03 ?0.51 ± 0.02 19 0.7 (84.5/114.0) 0.19? 2.7? ... 0.59? 2.4? 0.11 0.9 0.32 0.8 1.0 0.05 +0.05 +1.1 +0.16 +0.8 +0.36 +0.7 20 1.0 (73.7/77.0) 0.21?0.05 3.0?1.8 ... 0.57?0.47 1.7?1.6 1.51?0.28 2.2? 0.27 ± 0.03 ?0.42 ± 0.03 0.7 +0.07 +0.7 +0.64 +0.8 21 0.8 (32.3/39.0) 0.20? 1 . 7 . . . . . . . . . 1 . 39 2 . 1 0.19 ± 0.04 ?0.48 ± 0.02 0.07 ?1.1 ?0.76 ?0.7 +0.02 +2.3 +0.06 +0.5 22 0.9 (45.3/48.0) 0.10? 8.9? ... 0.32? 1.9? ... ... ?0.05 ± 0.04 ?0.53 ± 0.03 0.02 2.4 0.8 0.05 +0.02 +3.4 +0.04 +0.8 23 0.9 (54.0/62.0) 0.09? 13.5? ... 0.30? 3.1? ... ... ?0.07 ± 0.03 ?0.53 ± 0.03 0.01 3.9 0.03 0.7 24 ... ... ... ... ... ... ... ... ?0.06 ± 0.05 ?0.30 ± 0.04 +0.01 +20.6 +0.03 +1.0 23+24 1.1 (82.5/75.0) 0.07? 46.7? ... 0.26? 6.2? ... ... ?0.07 ± 0.03 ?0.45 ± 0.02 0.01 0.02 1.0 15.2 +0.03 +3.3 +0.08 +1.2 25 0.4 (8.5/20.0) 0.09?0.03 8.5?4.1 ... 0.24?0.06 1.3?0.8 ... ... ?0.19 ± 0.05 ?0.37 ± 0.04 +0.01 +11.6 +0.02 +2.7 1. . . 7 (NW halo) 0.8 (209.4/261.0) 0.10? 97.5? ... 0.33? 34.7? ... ... 0.04 ± 0.01 ?0.55 ± 0.01 0.01 12.4 0.01 3.6 +9.2 +0.03 +2.0 +0.02 . . . 0 . 29 7 . 5 . . . . . . ? 0 .08 ± 0.02 ?0.46 ± 0.02 22. . . 25 (SE halo) 1.1 (82.4/76.0) 0.09? 34 . 9 ?5.8 ?0.04 ?1.8 0.02 +0.01 +8.6 +0.03 +1.8 2. . . 4 1.0 (97.8/99.0) 0.10? 26 . 6 . . . 0 . 33 11 . 1 . . . . . . 0 . 09 ± 0.02 ?0.51 ± 0.02 0.01 ?7.8 ?0.03 ?1.6 +1.7 +0.02 +6.1 +0.01 ... ... 0.06 ± 0.02 ?0.59 ± 0.01 14.8? ... 0.32? 36.8? 5. . . 7 0.9 (157.8/168.0) 0.10? 1.9 0.02 7.1 0.01 +0.02 +6.7 +0.03 +1.7 22. . . 24 1.1 (91.1/81.0) 0.09? 29.0? ... 0.29? 6.7? ... ... ?0.06 ± 0.02 ?0.48 ± 0.02 1.4 0.02 7.6 0.03 NOTE: All errors are 90% con?dence for a number of interesting parameters equal to the number of free parameters in the model (3, 4, 5, 6, 7, or 9 free parameters depending on the model). Thin thermal plasma models are apec, absorbtion models are tbabs. All models have a ?xed foreground absorption of 1.3 × 1020 cm?2 . For region 8, 10 and 24 no spectral ?t was attempted due to low statistics. The hardness ratios in these regions do give meaningfull values though. Region

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253 13

14

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

Table A.2. Spectral ?ts in the halo extraction regions with a thin thermal plasma plus power law component, as opposed to a multi-temperature thermal plasma model in Table A.1.
χ2 kT norm Γ norm ν (χ2 /ν) (keV) (10?5 ) (10?5 ) +0.21 +2.9 +0.64 +1.8 1 1.1 (22.3/21.0) 0.19? 1.3? 1.90? 5.6? 0.19 1.3 0.61 2.0 +0.08 +1.5 +0.60 +0.9 2 0.7 (12.5/18.0) 0.24? 2 . 0 2 . 48 1 . 5 ?1.4 ?1.0 0.06 ?0.86 +0.06 +1.4 +0.34 +0.5 3 1.1 (41.2/39.0) 0.32? 3.8? 3.09? 0.9? 0.04 1.1 0.7 0.57 +0.05 +1.3 +0.56 +0.6 4 1.0 (24.4/25.0) 0.29?0.05 3.1?1.3 2.87?1.11 0.7?0.7 +0.04 +1.2 +0.43 +0.6 5 0.8 (32.9/41.0) 0.21?0.04 2.5?1.2 2.60?0.53 1.5?0.5 +0.03 +1.4 +0.30 +0.6 6 0.8 (74.9/90.0) 0.25? 4 . 2 2 . 41 2 .6? 0.03 ?0.32 0.7 ?1.5 +0.04 +1.8 +0.47 +1.1 7 0.7 (41.7/59.0) 0.22?0.04 2.9?1.5 1.74?0.62 3.5?1.5 +0.06 +1.3 +0.64 +0.4 22 0.9 (44.0/48.0) 0.16? 1.8? 2.40? 1.4? 1.2 0.4 0.05 0.66 +0.03 +1.0 +0.35 +0.3 23 0.9 (52.8/62.0) 0.24? 2 . 7 3 . 25 0 . 8 0.03 ?0.9 ?0.37 ?0.3 +0.05 +2.0 +0.66 +0.2 25 0.5 (10.4/20.0) 0.18? 1.9? 3.80? 0.3? 1.0 0.84 0.3 0.05 +0.07 +4.7 +0.28 +0.3 23+24 1.3 (98.6/75.0) 0.22?0.08 0.9?0.5 2.49?0.35 3.5?1.9 +0.01 +3.9 +0.16 +1.5 1. . . 7 (NW halo) 0.8 (201.6/261.0) 0.26? 29.1? 2.84? 10.2? 0.01 4.4 1.3 0.16 +0.04 +2.6 +0.37 +0.8 22. . . 25 (SE halo) 1.1 (81.2/76.0) 0.21? 6 . 2 3 . 03 2 . 7 0.03 ?2.4 ?0.43 ?0.8 +0.04 +2.9 +0.29 +1.4 2. . . 4 1.0 (101.6/99.0) 0.28? 7.5? 2.46? 4.5? 0.04 2.5 0.48 1.5 +0.02 +2.4 +0.23 +1.0 5. . . 7 0.9 (151.8/168.0) 0.25?0.02 12.4?2.6 2.65?0.24 4.8?1.1 +0.6 +0.33 +2.0 +0.03 1.9? 3.14? 5.7? 22. . . 24 1.2 (93.9/81.0) 0.23? 0.37 1.8 0.03 0.6 NOTE: All errors are 90% con?dence for a number of interesting parameters equal to the number of free parameters in the model. Thin thermal plasma models are apec. All models have a ?xed foreground absorption of 1.3 × 1020 cm?2 . Region

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

15

Appendix B: EPIC PN images
We developed an algorithm to create vignetting corrected and adaptively smoothed EPIC PN images. This procedure is based on the observation itself. Speci?cally, closed ?lter observations were not used in addition. In the following we will describe the algorithm step-by-step. The basis for this procedure is a cleaned event ?le and an out-of-time event ?le. This cleaning included screening for high background and also removing bad pixels and bad columns. If we want to use more than one observation, all steps have to be performed for all observations separately, before the products are merged. We will here concentrate on the creation of three images, which can be combined to a RGB colour image at the end of the procedure. In the following all steps are to be done for all three energy bands, unless stated otherwise. For the desired energy band, we extracted an image from the event ?le. This image was then corrected for out-of-time events, via subtraction of an image that was extracted from the out-of-time event ?le and rescaled with the out-of-time event fraction. Next, we corrected for the detector background (electronic noise, high energy particles) by subtracting the detector background surface brightness from the image. This value was determined from the corners of the detector which are outside of the ?eld of view of the telescope. We here assumed that the detector background is uniform across the whole detector (for energies above ?7.2 keV this is no longer a good approximation, see Freyberg et al. 2004). We created vignetting corrected exposure maps and masks using the SAS-tasks eexpmap and emask, respectively, which will be used to account for di?erences in the exposure times in the images, and to mask the images to regions with an acceptable minimum exposure time. Before we could smooth the image adaptively, we had to create a template with smoothing kernels. This template guarantees that images in di?erent energy bands are smoothed with the same kernel size. Therefore we added up the images in the di?erent energy bands. With the task asmooth, we created the template using the merged image, the merged mask, and an exposure map. Now we have all the necessary products to smooth the single images. We smoothed the images in the di?erent energy bands with the smoothing template, the vignetting corrected exposure map, and with the corresponding mask. This step included the vignetting correction via the exposure map, and a masking of the image to a region with an acceptable minimum exposure time. In a ?nal step we used ds9 to create the RGB colour image. The resulting images of NGC 253 and a combined RGB colour image of the lowest three energy bands is shown in Fig. 1, 2, and 3.

Table C.1. Rejected CCD rows due to MIPs per time unit in the used observations.
Obs ID 0122320707 0125960101 0152020101 0160362801 Filter Closed Medium Thin Closed rejected line counter value 181.7 190.2 141.4 120.4

Appendix C: Background spectra
The conventional way to create a background spectrum is to select a region from the same observation where there is no emission from the source. Additionally, the region should be close to the source. This way, the spectral background should

have the same characteristics as the background at the source region. In NGC 253, a region which su?ces the ?rst criterion can be found at the border of the ?eld of view in the southwestern part of the detector. The second criterion, however, is not satis?ed. The background region may show a di?erent detector background, and additionally the vignetting is di?erent. Since we were interested in determining the characteristics of emission with low surface brightness, that extends over a large region, where the background is (probably) the dominant component, we needed a very accurate estimate of the background. Given the very soft nature of the emission, we cannot use blank sky observations that were taken in regions of the sky where the foreground NH is di?erent (not to mention other uncertainties due to di?erent detector settings, particle radiation levels, etc.). Here we describe a method to use a local estimate of the sky background that takes properly into account vignetting and detector background issues. To remove the detector background, we used archival observations which were taken in the same mode as the NGC 253 observations, but where the ?lter wheel was closed. To avoid e?ects due to changes in the detector settings, or changes of the detector performance due to other reasons, we chose the closed observations to be as close as possible in time to the NGC 253 observations. The closed observations we used for observation 1 and 4 are: revolution 59, obs. id. 0122320701, exposure S003 (50.5 ks) and revolution 732, obs. id. 0160362801, exposure S005 (38.6 ks), respectively. To ensure, that there are as little as possible di?erences between the source observation and the closed observation, we removed bad columns and bad pixels both in the NGC 253 and closed observation. Additionally, the closed observations may have been taken when the spacecraft was exposed to a di?erent particle radiation level than the one present during observations 1 or 4. The XMM-Newton house keeping ?le contains information on how many CCD rows per time unit were rejected due to a possible minimum ionising particle (MIP) event, which is a direct estimator of the average radiation level. We used these values (see Table C.1), to rescale the count rate of the closed observations. We used Out-of-Time spectra from the source and background region to correct for contribution from Out-of-Time events. When one subtracts a closed observation spectrum from a Out-of-Time corrected spectrum, one actually removes the Out-of-Time spectrum of the detector background twice. This is corrected in our method by adding again the Out-of-Time spectra of the detector background. We corrected the background region spectrum for Out-ofTime events and the detector background and applied the vignetting correction in each energy bin as a function of o?-axis

16

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253 Robs tobs Robs tobs OOT ×f+ ? S det (E) Rdet tdet Rdet tdet
detector Out?of ?Time events

OOT B(E ) = S obs (E ) × f + S det (E ) Out?of ?Time events

detector background

? ? ? ? ? ? ? ? ? ? ? ? ? R t t R V ( E , θS ) A S ? ? ? obs obs obs obs ? ? OOT OOT ? ? B ×? × f + B ( E ) + ? ? obs ( E ) ? Bobs ( E ) × f ? Bdet ( E ) ? det ? V (E, θB ) AB ? R t R t ? ? det det det det ? ? ? ? ? ? Out?of ?Time events
detector background detector Out?of ?time events sky background

(C.1)

angle of the source and background spectrum. This gave us the sky background spectrum. In all of the above steps, di?erent exposure times and areas in the extraction regions have been accounted for. Since some of the components in the ?nal background spectrum do have low number statistics, we used the √ conservative approximation to Poissonian errors σN ≈ 1 + 0.75 + N (Gehrels 1986). To avoid unjusti?ed large errors, we roughly binned the spectrum before calculating errors. The resulting background subtracted spectrum then has a signi?cance in each bin of a least 3 σ. The errors were propagated properly and were included in the ?le with the ?nal background spectrum. This spectrum can be used with XSPEC as a background spectrum. The whole method can be summarised by Eq. C.1 with the following symbols: – B(E ) is the counts at energy E in the background spectrum – Bobs (E ) is the counts in the NGC 253 observation – S det (E ) is the counts from the detector background spectrum in the source region – Bobs (E ) is the counts in the detector background spectrum in the background region OOT – S obs (E ) are the counts in the Out-of-Time spectra in the source region – BOOT obs ( E ) are the counts in the Out-of-Time spectra in the background region – tobs is the exposure time in the NGC 253 observation – tdet is the exposure time in the closed observation – Robs is the rejected line counter values (see Table C.1) in the NGC 253 observation – Rdet is the rejected line counter values (see Table C.1) in the closed observation – AS is the area in the source region – AB is the area in the background region – V (E , θS ) is the vignetting value in the source region, depending on the o?set angle θ and the energy E – V (E , θB ) is the vignetting value in the background region, depending on the o?set angle θ and the energy E – f is the fraction of Out-of-Time events in the corresponding mode of the observation A comparison between this new method and the conventional method, that does not use the vignetting correction nor the closed observations, is shown in Fig. C.1 for two example spectra, both in observations 1 and 4. The single background components in the source and background region in observation 4 are shown in Fig. C.2. All ?gures show counts integrated over the extraction region. The counts in the background region

were rescaled to the source region area to be able to compare them to the source spectrum. Also, the counts in the closed observation were rescaled to the exposure time and radiation level in the source observation. The di?erences between the new and the conventional method in terms of the resulting best ?ts are the following: In the majority of the tested cases, an additional power law component with Γ ? 0 is required for the ?t in the spectrum, obtained with the conventional method. The temperatures are consistent between both methods, but the resulting ?ux levels in the conventional method are higher. Di?erences in total ?ux values range between 2% and 22%. The e?ect between the two methods is highest in regions with low surface brightness. Here the background dominates and a correct treatment is crucial. As an example, the di?erence in ?ux in region 7 (low surface brightness) is 15% and 22%, for observations 1 and 4, respectively. Whereas in region 14 (high surface brightness), the differences are 2% and 3%.

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

17

Fig. C.1. Comparison between the background substraction on two examples (same source regions as in Fig 6). (left): the spectrum of region 7, (right): the spectrum of region 14. The top panel shows the new method, as described in this paper, the bottom panel shows the conventional method, where the raw background spectrum is used, and a correction for Out-of-Time events has been applied. We only show the spectra of observation 4 here, since these have the better statistics.

18

M. Bauer et al.: XMM-Newton observations of the di?use X-ray emission in the starburst galaxy NGC 253

Fig. C.2. The single components that are part of the total background spectrum compared to the source spectrum. (left): Region 7, (right): Region 14, (top): components from the source region, (bottom): components from the background region. The single components were corrected for areas, exposure time, and radiation level, with respect to the source spectrum in the source region, but no vignetting correction was applied yet.


Xmm-Newton Observations of the Diffuse X-ray Backgr....pdf

Xmm-Newton Observations of the Diffuse X-ray Background We analyzed two XMM...ciently high to shield about 75% of the foreground emission in the 3/4...

Diffuse X-ray emission from the planetary nebula NG....pdf

The diffuse X-ray emission must originate in the shocked fast st...更多>>XMM-Newton EPIC observations of the planetary nebula (PN) NGC 7009, the ...

XMM-Newton observations of the Vela pulsar.pdf

XMM-Newton observations of the Vela pulsar_专业资料。We present spectral ...In the X-ray band, thermal emission was discovered by ROSAT (Ogelman et...

X-ray observations of the starburst galaxy NGC 253 ....pdf

XMM-Newton observations ... 暂无评价 18页 免费 ...X-ray observations of the starburst galaxy NGC ...use, ?lamentary X-ray emission. Seven sources ...

XMM-Newton Observation of Diffuse Gas and LMXBs in ....pdf

更多>>(Abridged) We present an XMM-Newton X-ray observation of the X-ray bright E2 elliptical galaxy NGC 4649. In addition to bright diffuse emission...

XMM-Newton Observations of the Black Widow Pulsar P....pdf

XMM-Newton Observations of the Black Widow Pulsar...The pulsar's X-ray emission is non-thermal and...elds in the range B ∝ (PP) ? 108 ? 1010 ...

INTEGRAL and XMM-Newton observations of the weak GR....pdf

XMM-Newton observations of the prompt γ-ray emission and the Xray ...ux of ?1.1 photons cm?2 s?1 in the 20-200 keV energy range. The ...

XMM-Newton observations of M31 X-ray properties of ....pdf

(2004) used XMM-Newton observations to study X-ray spectral properties of the 3 bright thermal SNR in the northern disk of M31. 2 Source identi?...

XMM-Newton observations of the microquasars GRO J16....pdf

XMM-Newton observations of the microquasars GRO J1...in the calibration of the Burst mode, with ...its thermal emission (kT ? 1.2 ? 1.3 keV)...

Properties of the diffuse X-ray background in a hig....pdf

Diffuse X-Ray Emission A... 23页 免费 Fluctuations in the diff... 暂无...The Galactic contributio... 暂无评价 4页 免费 Xmm-Newton Observations ......

XMM-Newton EPIC and OM Observations of Her X-1 over....pdf

XMM-Newton EPIC and OM Observations of Her X-1...star is weak or absent in the low state - in...The strong ?uorescence emission line at ? 6.4 ...

Diffuse X-Ray Emission from the Quiescent Superbubb....pdf

ROSAT observations of M17 detected diffuse X-ray emission...更多>>The ...eld in the ambient interstellar medium. Subject headings: ISM: bubbles ...

...XMM-Newton Spectra of Faint Sources in the Lockm....pdf

ed sources make up the bulk of the energy in the XRB. Acknowledgments. These observations were made using the XMM-Newton X-ray observatory, built, ...

Diffuse X-ray Emission in a Deep Chandra Image of t....pdf

We examine the spectrum of diffuse emission detected in the 17' by 17' field around Sgr A* during 625 ks of Chandra observations. The spectrum exhibits...

Diffuse light and building history of the galaxy cl....pdf

The analysis of archive XMM-Newton and Chandra observations shows X-ray emission elongated in the same direction. The X-ray temperature map shows the ...

XMM-Newton Observations of Galaxy Clusters The Radi....pdf

XMM-Newton Observations of Galaxy Clusters The ...(For example, in the sample of 106 clusters ...centred on the peak of the X-ray emission from...

...from the Optical Monitor on XMM-Newton_图文.pdf

The distribution of UV emission in the galaxy ...starburst ultraviolet: galaxies galaxies: evolution...observations obtained with XMM-Newton, an ESA ...

XMM-Newton observations of the supernova remnant IC....pdf

XMM-Newton observations of the supernova remnant IC 443 II. evidence of stellar...The dashed line traces the symmetry axis of the X-ray emission in the...

...NGC 2516 stars in the XMM-Newton observations_免....pdf

X-ray variability of NGC 2516 stars in the XMM-Newton observations We present the characteristics of the X-ray variability of stars in the cluster NGC25...

XMM-Newton observations of two X-ray bright galaxy ....pdf

XMM-Newton observations of two X-ray bright galaxy groups - pushing out to...the emission from galaxy groups is typically lost in the X-ray background...