The Diffuse and Compact X-ray Components of the Starburst Galaxy Henize 2-10 Henry A. Kobulnicky 0 Department of Physics & Astronomy 1 University of Wyoming 0 2 Laramie, WY 82070 n u [email protected] J 7 Crystal L. Martin ] O University of California, Santa Barbara C Department of Physics . h Santa Barbara, CA 93106 p - o [email protected] r t s a [ ABSTRACT 1 v Chandra X-ray Observatory imaging spectroscopy of the starburst galaxy 9 8 Henize 2-10reveals a strong nuclear point source and atleast two fainter compact 1 1 sources embedded within a more luminous diffuse thermal component. Spectral 6. fits to the nuclear X-ray source imply an unabsorbed X-ray luminosity L > 1040 x 0 erg s−1 for reasonable power law or blackbody models, consistent with accretion 0 1 onto a >50 M black hole behind a foreground absorbing column of N > 1023 ⊙ H : v cm−2 . Two of these point sources have L = 2−5×1038 erg s−1, comparable x i X to luminous X-ray binaries. These compact sources constitute a small fraction r (≤16%) of the total X-ray flux from He 2-10 in the 0.3–6.0 keV band and just a 31% of the X-rays in the hard 1.1–6.0 keV band which is dominated by diffuse emission. Two-temperature solar-composition plasmas (kT≃0.2 keV andkT≃0.7 keV) fit the diffuse X-ray component as well as single-temperature plasmas with enhanced α/Fe ratios. Since the observed radial gradient of the X-ray surface brightness closely follows that of the Hα emission, the composition of the X-ray plasma likely reflects mixing of the ambient cool/warm ISM with an even hotter, low emission measure plasma, thereby explaining the ∼solar ISM composition. Aperture synthesis 21-cm maps show an extended neutral medium to radii of 60′′ so that the warm and hot phases of the ISM, which extend to ∼30′′, are enveloped within the 8×1020 cm−2 contour of the cool neutral medium. This – 2 – extended neutral halo may serve to inhibit a starburst-driven outflow unless it is predominantly along the line of sight. The high areal density of star formation can also be reconciled with the lack of prominent outflow signatures if Henize 2-10 is in the very early stages of developing a galactic wind. Subject headings: galaxies: formation — galaxies: evolution — galaxies: funda- mental parameters — galaxies: abundances 1. Introduction 1.1. Goals of this Study The blue compact galaxy Henize 2-10 has been extensively studied as the prototype of starbursting galaxies containing large populations of Wolf-Rayet stars (Conti 1991). As a galaxy hosting recent intense star formation activity, He 2-10 is a natural target for investi- gating the impact of starbursts on interstellar medium. In Henize 2-10, supernova activity has apparently produced a hot X-ray emitting component and has arguably impacted the composition and kinematics of the warm and cold ISM components as well (M´endez et al. 1999; Johnson et al. 2000; Ott et al. 2005; Grimes et al. 2005; Schwartz et al. 2006). Our goal in this paper is to provide a systematic overview of the physical state of and the re- lation between these components using a panchromatic suite of X-ray, optical, and radio datasets. We use a 20 ksec imaging observation with the Chandra X-ray Observatory in order to characterize the distribution of the hot X-ray emitting plasma in relation to the ionized and neutral gas and provide a complete analysis of the X-ray point source population in and around He 2-10. Previous analyses of these Chandra data by Ott et al. (2005) and Grimes et al. (2005) have addressed the integrated X-rayproperties of He 2-10in the context of larger samples of star-forming galaxies with diffuse X-ray emission. These studies con- cluded that the hot ISM is mildly enriched in α-process elements, consistent with the mixing of fresh supernova ejecta from the recent starburst into the surrounding ISM. Our focus here is a more detailed look at both the diffuse and compact X-ray components of He 2-10 than previously reported. We show that α-element enrichment is not strictly required to fit the X-ray spectrum, we identify a hard ultraluminous X-ray source as a counterpart to a nuclear radio source, and we discuss the relative energetics and morphologies of the cold, warm, and hot phases of the ISM to understand the state of the starburst-driven outflow. First, we provide a brief review of the global characteristics of He 2-10 that will be relevant to later discussion. – 3 – 1.2. Henize 2-10: Content and Kinematics Morphological studies reveal at least three distinct starforming regions as shown in the images of M´endez et al. (1999). The most luminous central starburst region contains 6– 10 compact (<1′′) super starclusters which have diameters of <10 pc at the nominal 9.8 Mpc distance of He 2-10.1 These clusters have ages <10 Myr and masses up to 105 M ⊙ (Conti & Vacca 1994; Johnson et al. 2000; Vacca et al. 2002). Chandar et al. (2003) assign an age range of 4–5 Myr for the four most luminous clusters. Radiointerferometer observationsreveal thepresence ofatleastfourradio-bright“ultra- dense” H II regions powered by very young super starclusters (Kobulnicky & Johnson 1999; Johnson & Kobulnicky 2003). The Hα luminosity of Henize 2-10 (not corrected for extinc- tion) measured by M´endez et al. (1999) is 2.1×1041 erg s−1, leading to an implied current star formationrateof>1.8M yr−1 basedontheformulationofKennicutt (1989). Given the ⊙ relatively small 20′′×20′′ (0.89 kpc2) region containing all of this activity, the star formation rate per unit area is at least ≃ 2 M yr−1 kpc−2. The Spitzer Space Telescope 24 µm flux of ⊙ 4.9 Jy (Engelbracht et al. 2005) implies a much higher star formation rate of 6.8 M yr−1 ⊙ using the calibration of Calzetti et al. (2007), indicating that much of the star formation is heavily obscured (Vacca et al. 2002). He 2-10 contains an H I mass of 4.9 × 108 M⊙ (Kobulnicky et al. 1995) and molecu- lar mass of 2.9 × 108 M yielding a helium-corrected gas mass fraction of ≤0.20, consis- ⊙ tent with Henize 2-10 being a transition object between gas-rich dwarfs and larger, more gas-poor spiral galaxies. Beam-averaged H I column densities in the central region vary from few ×1020 cm−2 to as high as 2×1022 cm−2 (Baas et al. 1994; Kobulnicky et al. 1995; Meier et al. 2001; Santangelo et al. 2009) corresponding to extinctions up to A =20 mag V for typical Galactic gas-to-dust ratios (Bohlin et al. 1978). This is consistent with the lo- calized high extinctions of A =10–30 mag inferred from mid-IR silicate absorption and IR V Brackett emission-line methods (Phillips et al. 1984; Kawara et al. 1989; Beck et al. 1997) but much higher than the A =1–2 mag derived toward optically visible star clusters and V H II regions (Vacca & Conti 1992). These disparate results warn that the nuclear condi- tions are complex on small angular scales with large variations along the line of sight. This high dust and molecular content is consistent with the strong CO emission and with chem- 1 The observed optical and 21-cm radial velocities (Kobulnicky et al. 1995; M´endez et al. 1999; Schwartz 2005) corrected for Galactic rotation and Virgocentric infall yield a Hubble-flow velocity of ∼712 km s−1 , leadingtoadistanceof9.8MpcforH0 =72km s−1 Mpc−1. Thisimpliesanangularscaleof47pcarcsec−1 and a distance modulus of 29.95 mag. Throughout, we convert linear distances and implied luminosities fromprior worksself-consistently to this distance. Distances ranging from9 to 14 Mpc may be found in the literature. – 4 – ical abundance measurements showing a metallicity near the solar value (Kobulnicky et al. 1999). Darling et al. (2008) detect H O “kilomasers” in the nucleus, indicating the presence 2 of many massive star-forming regions. Single-dish radio continuum measurements over a range of frequencies show a dominant non-thermal component, indicating the presence of significant supernova activity (Allen et al. 1976). Hα images of Henize 2-10 show multiple ionized shells and filaments typical of actively starforming dwarf galaxies (Marlowe et al. 1995). High-resolution Hα spectroscopy reveals multiple kinematic components, with the fastest reaching ∆V =250 km s−1 (M´endez et al. 1999). Blueshifted interstellar absorption lines seen against the nucleus indicate an outflow of low-ionization material along the line of sight (Schwartz et al. 2006; Johnson et al. 2000). Published outflow speeds vary by a factor of two owing to uncertainties regarding the sys- temicvelocity, lowspectralresolution, andthewavelength ambiguitycausedbytheuncertain positioning of the continuum source within a large spectral aperture. Optical echelle spec- troscopy obtained with HIRES at the W. M. Keck Observatory offer the most accurate measurements to date (Schwartz 2005). Excited stellar Mg I triplet λλλ 5167.32, 5172.68, 5183.60 ˚A absorption yields a heliocentric systemic velocity of 869 ± 3km s−1 , consistent with the 21-cm and CO data. The narrow slit and seven km s−1 resolution reveal four blueshifted interstellar Na I λλ 5889.95, 5895.92 ˚A components in addition to a stellar ab- sorption component at the systemic velocity. The fastest (and weakest) of these reaches a maximum velocity of -180 km s−1 . The average outflow speeds of the four components are −34±2, −65±2, −93±4, and −128±2 km s−1 . The maximum H I rotation velocity observed by Kobulnicky et al. (1995) provides a lower limit on the dynamical mass of He 2-10. For the adopted distance of 9.8 Mpc and an inclination of 30 degrees, the dynamical mass within 2.1 kpc is M = 1.18×1010 M . dyn ⊙ The modified isothermal sphere model of Ferrara & Tolstoy (2000) for the mass distribution implies a halo scale radius of 4.8 kpc and a central density of 9.5×106 M kpc3. We estimate ⊙ the escape velocity at the disk radius of 2.1 kpc is at least 150 km s−1 for a disk inclination of 30 degrees. Hence little of the observed outflow exceeds the escape velocity. A decrease in the covering fraction of the low-ionization gas with radius (Martin & Bouch´e 2009) and/or increasing ionization fraction with radius, however, might render higher velocity gas difficult to detect. – 5 – 2. Data Acquisition and Reduction 2.1. Chandra Observations The Chandra X-ray Observatory observed Henize 2-10 with the Chandra CCD Imaging Spectrometer (ACIS) on 2001 March 23 for 20.02 ksec. The galaxy was placed on the back- illuminated chip ACIS-S3 (also known as chip 7). These data, as well as the fields on the ACIS-I and ACIS-S5 chips, are available in the archive under sequence number 600209. ThedatawereprocessedbytheChandra X-rayCenter(CXC)usingversionR4CU5UPD14.1 of their software. CIAO, version 2.3, with CALDB, version 2.4, was used for the majority of the data processing unless otherwise noted. We checked the positional accuracy of the chip 7 Chandra sources using our R-band image. For the 9 X-ray sources with optical detections, the rms positional discrepancy was found to be 0′.′5. The light curve was inspected, and no periods of high background occurred during the observation. No additional re-processing was required. Spectra were extracted using pulse invariant data values to account for gain variations between nodes. For each source region, we weighted the appropriate CXC spectral response files by the distribution of their areas within the aperture.2 The maximum difference be- tween ourarea-weighted response files andsingle response files extracted atthe flux-weighted centroid of the aperture was a few percent. Weighting the response files by the distribution of counts within the aperture produced results indistinguishable from the area-weighted re- sponses. Unbinnedimageswereextractedfromthetime-filteredevents fileinfourcarefullychosen bands: soft (S) 0.3–0.7 keV, medium (M) 0.7–1.1 keV, hard (H) 1.1–6 keV, and Total (T) 0.3–6keV. EmissioninthesoftbandissomewhatattenuatedbecausetheGalacticforeground column, N = 4.89 × 1020cm−2 (Hartmann & Burton 1996), corresponds to optical depth H unity at 0.35 keV. The galaxy was not detected at energies above 6 keV, the energy adopted for the hard band cutoff. Background images were extracted from the background events file in these bands. Each of the four X-ray images was smoothed using the adaptive smoothing algorithm of H. Ebeling & V. Rangarajan as implemented as CSMOOTH in CIAO. The smoothing scales are automatically adjusted to achieve a minimum S/N ratio of 1.8 and a maximum S/N ratio of 2.2 per pixel. Strong point sources are effectively unchanged by the smoothing 2Weusedthecalcrmf/calcarfsoftwarepackagecontributedbyJonathonMcDowellwhichisavailablefrom the CXC website – http://asc.harvard.edu/cgi-gen/cont-soft/soft-list.cgi. – 6 – process, whilethecontrastofweak, diffuseemissionisenhanced. Eachofthefourbackground imageswassmoothed(usingthesmoothingmapgeneratedbyCSMOOTHfromtheon-source images) and subtracted from the source images. Since adaptive smoothing does not preserve photon statistics in a straightforward man- ner, we used the adaptively smoothed images only to produce images for presentation and to define large apertures for the extraction of diffuse component spectra. All quantitative analyses were performed by extracting photons from the events file or unsmoothed images and by estimating their significance using Poisson statistics. 2.2. Point Source Identification The CIAO 4.1 detection algorithms CELLDETECT and WAVDETECT were used to identify compact X-ray sources on the ACIS S3 chip over the energy range 0.3 – 6 keV. We used WAVDETECT with wavelet scales between 1 and 32 pixels as the as the primary detection tool, and this yielded 29 probable X-ray sources, the weakest of which had a sig- nificance level of 1.7σ. Three sources (#’s 1,2,3) lie projected against the stellar component of He 2-10, while the remainder may be associated with field stars or background galaxies. CELLDETECT found a subset of these 29 sources, but also yielded four additional sources not identified by WAVDETECT (#’s 30 – 33). These four lie within 10′′ of the nucleus where the diffuse X-ray emission is strong, have sizes exceeding the nominal PSF size, and, we show later, are likely to be localized peaks in the diffuse thermal emission. Table 1 lists the positions of the 29+4 compact sources detected. Figure 1 shows these sources num- bered 1–29 overlaid on an optical 6550 ˚A image. Sources 1,2,3 and 30–33 are better seen in Figure 2 which shows an adaptively smoothed X-ray depiction (contours) of the Henize 2- 10 nucleus overlaid on a 400 s F814W image (grayscale) from the Hubble Space Telescope archive (Johnson et al. 2000). Figure 3 shows an unsmoothed 0.3 – 6 keV X-ray image of the ACIS S3 chip with the elliptical source regions, as in Figure 1. The inset in this Figure shows a zoomed view near the nucleus that includes sources 1,2,3 and 30–33. We measured counts in each of the four energy bands (Total, Soft, Medium, and Hard) using elliptical apertures containing probable source pixels as identified by WAVDETECT. Annuli surrounding each aperture were defined manually and used for measuring and sub- tracting the background. Table 1 lists the net counts detected in each band along with the ratio of the source size to the nominal PSF. Sources 1,2,3, 10 and 30–33 have PSF ratios near 2 or larger, suggesting that they may be either extended sources or blends of multiple point sources. – 7 – Our detection threshold of ∼4 counts in 20 ksec in the Total 0.3–6 keV band on the Chandra ACIS corresponds to an X-ray flux of 1.5×10−15 based on the WebPIMMS mission count rate simulator. At or above this flux level we would expect an X-ray source den- sity of ∼0.8 sources per square arcminute, based on the results of Mushotzky et al. (2000), Moretti et al. (2003), and Kim et al. (2007). Within the ∼0.25 square degree region en- compassing He 2-10, we would therefore expect, on average, 0.2 background X-ray sources. Statistically, the three (possibly seven) compact sources projected onto He 2-10 are likely to lie within the galaxy itself. 2.3. Optical imaging Optical images of He 2-10 were obtained the night of 2001 March 28 with the 3.5 m WIYN3 telescope and the MiniMosaic detector. Seeing was highly variable, and non- photometric conditions prevailed. Exposures of 60 s were obtained using a filter of width 385 ˚A centered at 6550 ˚A. Two exposures with the smallest point spread functions were combined to produce an image with a mean stellar FWHM of 1.2′′. A rough R-band flux calibration was performed on the image using the magnitudes of 8 stars in the field based on the Guide Star Catalog F-band magnitudes. The typical uncertainty is 0.5 mag. Poor weather and low target elevation prevented Hα narrow band exposures from being taken. We instead obtained a continuum-subtracted Hα image of Henize 2-10 which appeared in M´endez et al. (1999) (kindly given to us by C. Esteban.) The Hα image is flux calibrated using the calibration of M´endez et al. (1999). 2.4. Radio Continuum Observations Radio continuum observations of Henize 2-10 at 20 cm (1420 MHz) and 2 cm (15 GHz) were obtained with the Very Large Array in the B configuration and published previously in Kobulnicky et al. (1995) and Kobulnicky & Johnson (1999). Full details may be found in those papers. We used the 20-cm data mapped with a synthesized beam of 3′′ FWHM to search for radio sources in the field surrounding He 2-10 that correspond to the detected X-ray and optical sources. Other than Henize 2-10 itself, there are no 20-cm radio sources within the 8′ Chandra field to a 3σ limit of 0.6 mJy. 3The WIYN Observatory is a joint facility of the University of Wisconsin-Madison, Indiana University, Yale University, and the National Optical Astronomy Observatory. – 8 – 3. X-ray Properties 3.1. Nature of the Point Sources Sources 1–3 and 30–33 appear superimposed on diffuse emission near the nucleus of Henize 2-10, and all have PSF ratios between 2 and 4, while the remainder of the 26 compact sources further from the nucleus have PSF ratios not far from unity. These nuclear objects may either be genuine point sources (i.e., X-ray binaries or collections of X-ray binaries) projected onto regions of diffuse gas, or they may be local maxima in the diffuse emission. The second possibility better explains their larger sizes compared to the field sources. Their X-ray colors help to discriminate between these possibilities. Figure 4 shows an X-ray two-color diagram formed from ratios of count rates in the hard, medium, and soft bandpasses: (M − S)/(M + S) and (H −M)/(H + M). The left panels include only the seven sources projected against the nucleus, while right panels show the other 26 sources. Lines and filled circles illustrate the X-ray colors for various power law, thermal plasma (MEKAL), and blackbody spectral energy distributions for four different foreground absorbing columns: 0.0, 0.2×1022, 0.8×1022, and 1.6×1022 cm−2. Sources 1–3 areconsistent withΓ = 1.5powerlawspectrabehindmodestabsorbingcolumns. Thesethree objects are strongest in the hard band, consistent with heavily absorbed spectra. Sources 2 and 3 are too faint for meaningful spectral fits, but lower limits on their luminosities can be made assuming a minimum column density of N = 0.05×1022 cm−2 for the Galactic H foreground and a power law spectrum with photon index Γ = 1 where the number of photons per energy bin is given by N(E)∝ E−Γ. For these values, sources 2 and 3 have probable unabsorbed 0.3–8 keV luminosities of >3.4×1038 erg s−1 and >1.7×1038 erg s−1 respectively at 9.8 Mpc. Although not quite as luminous as the “superluminous” X-ray sources found in nearby galaxies (> 1039 erg s−1; e.g., Roberts et al. (2004)), these objects are likely to be powered by accretion onto stellar mass black holes. However, the uncertainties in the foreground column and the photon energy distribution would allow for intrinsic luminosities several times greater than these estimates if N and Γ were larger than the values adopted H here. We note that both of these objects appear to lie within the older starburst region “B” (Vacca & Conti 1992) several arcseconds to the east of the dominant starburst core region (“A”). By contrast with sources 2 & 3, sources 30–33 are strongest in the medium band, consistent with the color expected of soft thermal plasma. The upper right panel of Figure 4 plots the colors of sources 4–29. We tentatively identify these sources as either stars or active galactic nuclei (AGN) based on the ratio of their X-ray to optical fluxes, following Krautter et al. (1999). The ratio of X-ray to optical fluxes is f /f , where f (erg s−1 cm−2) X V X – 9 – = 3.6×10−16 ×C where C is the number of detected 0.3–6.0 keV X-ray photons in 20 tot tot ks, and f (erg s−1 cm−2)=1.7×10−6 × 2.5−R where R is the R-band magnitude. Table 1 V lists objects with log(f /f ) < −0.5 as probable stars and objects with log(f /f ) > −0.5 X V X V as probable AGN. In Figure 4, probable stars appear as filled squares and probable AGN appear as crosses. Solid dots connected by lines indicate the colors of MEKAL thermal plasmas with temperatures of 0.5, 1.0, and 2.0 keV attenuated by absorbing columns of 0.0, 0.2×1022, 0.4×1022, 0.8×1022, and 1.6×1022 cm−2. The candidate stars occupy a region of relatively low column density, as might be expected for Galactic foreground objects. The candidate AGN are distributed throughout the diagram, consistent with a range of power law slopes and absorbing columns. The strongest source in the field is object 7, the star GSC 06578-03010. It is a previously known X-ray source, RX J0836.0-2621. 3.2. Diffuse versus Point Source Flux Contributions The compact sources 1–3 represent onlya small fractionofthe X-rayflux of Henize 2-10. We performed photometry on the X-ray images using an 80′′ diameter aperture centered on the nucleus to obtain a total flux for the galaxy. The size of this aperture was chosen to include all of the diffuse emission seen in the smoothed, background-subtracted total band (0.3–6.0 keV) image. This aperture includes the seven compact sources in the main body of Henize 2-10 and none of the other 26 sources in the field. The background flux was estimated using an annular region extending 15′′ beyond the circular aperture. Table 2 lists the results of this photometry in each band. Table 2 also indicates the fraction of diffuse emission after subtraction of the flux from the compact sources 1–3. We find that 84±4% of the emission in the total 0.3–6.0 keV band comes from diffuse emission. In the hard 1.1–6.0 keV band, 69±9% of the emission is diffuse. In the soft and medium bands, the fraction of flux from the diffuse X-ray medium is over 90%. Thus, the X-ray flux of Henize 2-10 appears to be dominated by the diffuse plasma component even in the highest energy band. 3.3. X-ray Spectrum of the Nuclear Source: An Intermediate-Mass Black Hole? Johnson & Kobulnicky (2003) noted the non-thermal nature of the spectral energy dis- tribution of the pointlike radio source in the nucleus (source #3 of Kobulnicky & Johnson (1999)) but didnot discuss thenatureof thissource. Figure5 shows the X-raycontour image from Figure 2 overlaid on a grayscale representation of the 2 cm (15 GHz) radio map from Kobulnicky & Johnson (1999). Radio source KJ3, marked by a star, lies 0.25′′ south of the – 10 – brightest nuclear X-ray source, #1. Within the uncertainties of the radio and X-ray data, the two sources are coincident. The optical and mid-IR images of Cabanac et al. (2005) show that there is no bright optical or infrared source at this location.4 We extracted anX-rayspectrum ofthe regionwithin 1′′ (∼ 2 pix) of thedominant X-ray source, #1, usinganeighboringannularregiontoestimatethelocalX-raybackground. Other choices of local background yield similar results. The resulting spectrum, shown in Figure 6, was then analyzed in Xspec (version 11.3). No single-component spectral models produce acceptable fits to the data. Reasonable models require at least two spectral components to reproduce both the low-energy peak near 1 keV and the secondary peak near 3 keV. This secondary high-energy peak is most pronounced in small 2′′ diameter spectral apertures centered on the nucleus and becomes weaker relative to the rest of the spectrum as the aperture size is increased. This indicates that the high energy photons near 3 keV are concentrated near the nucleus, comprising a real spectral feature of the nuclear source. The observed 0.3–6.0 keV flux from the nuclear source is 7.6×10−14 erg s−1 cm−2. Given the high optical extinction toward the nucleus of Henize 2-10 (A =10–30 mag; Phillips et al. V (1984); Kawara et al. (1989)) X-rays are likely to provide the best estimate of the bolometric luminosity of the central source. Weperformed spectral fits tothe nuclear source assuming a minimum foregroundGalac- tic column of 0.05×1022 cm−2 of solar abundance (component WABS5 in Xspec). We fit the X-ray SED with spectral models that were linear combinations of a solar-abundance ther- mal plasma component (MEKAL with variable foreground, N , and variable temperature, H kT) superimposed on either a power law component (POW with variable N and photon H index, Γ) or a second thermal component (variable N and variable kT). In all cases, the H second component required very high foreground columns of more than 2 × 1022 cm−2 in order that it contribute at 3 keV but not at lower energies. Such high column densities are consistent with the beam-averaged molecular gas column densities toward the nucleus (Kobulnicky et al. 1995) and correspond to optical extinctions of A =20 for typical Galactic V ratios N /A = 9.4×1020 cm−2 mag−1 (Bohlin et al. 1978). H2 V The best fitting models have reduced χ2 ≃ 1 and involve an absorbed MEKAL ther- 4Cabanac et al. (2005) identify and correct erroneous astrometry in the HST images that were used to compare optical and radio/IR positions in Kobulnicky & Johnson (1999), Vacca et al. (2002), Johnson & Kobulnicky (2003), and in some earlier works. The net result is that the HST images should be shifted 1′′.2 west relative to the radio and IR images in Figures from those prior publications. 5Adopting a model for the absorbing ISM based on the more recent X-ray cross sections of Wilms et al. (2000) (TABS in Xspec) changes the best fit parameters negligibly.