Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://hea-www.harvard.edu/~kim/pap/5ETG_1997.pdf Äàòà èçìåíåíèÿ: Mon Dec 6 20:20:27 2010 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 23:48:34 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: virgo cluster

Astron. Astrophys. 318, 361­375 (1997)

ASTRONOMY AND ASTROPHYSICS

ROSAT PSPC observations of 5 X-ray bright early type galaxies
G. Trinchieri1,2 , G. Fabbiano3 , and D-W. Kim4
1 2 3 4

Osservatorio Astronomico di Brera, Via Brera 28, I-20121 Milano, Italy ¨ ¨ Visiting Astronomer, Max-Planck Institut fur extraterrestrische Physik, Garching bei Munchen, Germany Harvard-Smithsonian Center for Astrophysics 60 Garden St., Cambridge, MA 02138, USA Chungnam National University, Dept. of Astron. and Space Sci., 305-764 Daejeon, Korea

Received 4 March 1996 / Accepted 1 July 1996

Abstract. We report on the ROSAT PSPC observations of 5 Xray bright early-type galaxies. Their X-ray morphology is more complex than E instein data had shown, ranging from the ellipsoidal shape of NGC 533 to the 200 kpc tail of NGC 7619. Hot gas at an average temperature of 0.8-1 keV dominates their X-ray emission. The spectral analysis is based on the assumption of a hot thin plasma at cosmic abundances, since the spectral resolution of the PSPC does not allow unambiguous measures of the model parameters (temperatures, low energy cut-off, metal abundance). However, the estimated temperature are not strongly affected by this choice. A temperature distribution of the hot interstellar medium is also derived. Higher temperatures are in some cases observed at larger radii, while the innermost 1 region is cooler in all of the objects studied. Due to the large range in distances however this corresponds to significantly different galaxy radii, from 5 kpc in NGC 4649 to 30 kpc in NGC 533. The temperature and density distributions derived are used to estimate the total mass of these systems, which span from MT 1012 - 1013 M and corresponding mass-to-light ratios M/L 10-150 in solar units. Formal errors on the derived masses are estimated to be > 25%. Care must be taken in comparing results for different systems, since some of the emission attributed to these objects could suffer from the contribution for the groups and clusters they belong to. Accordingly, our mass estimates could reflect the effect of the cluster/group potential.

1. Introduction E instein observations of early type galaxies have established that a hot interstellar medium (ISM) dominates the emission in the X-ray brightest objects. This discovery is very important for many reasons: 1) it has indicated the presence of substantial amounts of ISM in these objects; 2) it has provided a potentially powerful tool for estimating total galaxy masses at large radii; 3) it has provided a new wealth of parameters for the understanding of the enrichment of the intercluster medium. However, with E instein data, we had not been able to achieve a good understanding of the detailed properties of the hot ISM. Average spectral parameters, over the entire source extent, could be estimated, albeit with large uncertainties. The morphological appearance and total extent of the hot ISM could only be roughly estimated. The improved spatial and spectral resolution of the PSPC (Position Sensitive Proportional Counter) instrument on board ¨ the ROSAT satellite (Trumper 1983; Pfeffermann et al 1987) has proven to be crucial in furthering our understanding of the properties of the X-ray emitting gas in early type, normal, nearby galaxies. Its softer energy band ( 0.1 - 2 keV) coupled with the 0.5 spatial resolution is well suited for studying the spectral parameters of the gas in 1 annular regions. This approach has already been successfully applied to the data of other Xray bright early type galaxies (e.g . NGC 4636, Trinchieri et al 1994; NGC 507, Kim & Fabbiano 1995 among others). This work suggests that that the hot ISM is not isothermal over the entire source, but can show a slow temperature increase over a large region ( 40 kpc in NGC 4636), and that its morphology is far from smooth and regular, but can show asymmetries and inhomogeneities (e.g . NGC 507). In order to better understand the characteristics of the ISM in early type galaxies we have observed 5 X-ray bright early type galaxies for which E insetin data had already established the presence of a dominating hot gas component in their X-ray emission. The environment in which the galaxies sit spans a large range in richness, from the low density environment of NGC 533 (in the vicinity of A193, it is at the same recessional

Key words: galaxies: individual: NGC 533 ­ galaxies: individual: NGC 2563 ­ galaxies: individual: NGC 4649 ­ galaxies: individual: NGC 7619 ­ galaxies: individual: NGC 7626 ­ Xrays: galaxies

Send offprint requests to: G. Trinchieri


362

G. Trinchieri et al.: ROSAT PSPC observations of 5 X-ray bright early type galaxies Table 2. Parametric Representation of Radial Distribution of the X-ray Surface Brightness outside r 2 Obs. Dates Beginning End 14/07/93 27/07/93 11/10/93 13/10/93 30/04/94 02/05/94 21/12/91 27/12/91 30/05/92 11/06/92 On Time (s) 13054 27174 NGC 7619 22873 NGC 7626 14301
a b c d

Table 1. Log of the Observations Field RA Dec (J2000) 1:25:31.2 1:45:36.0 8:20:36.0 21:04:12.0

NGC 533 NGC 2563

Galaxy NGC 533 NGC 2563 NGC 4649

R

a max

16 14 13 7 22 3. 5

-1.95±0.14 -1.18±0.07 -2.51±0.18 -3.39±0.25 -1.45±0.06 -1.28±0.17 -2.87±0.05

b

b c d

NGC 4649 NGC 7619/7626 (Pegasus I)

12:43:43.2 11:33:36.0 23:20:31.2 8:12:36.0

18234

Maximum radius used in the fit For the azimuthally averaged profile From 4 to 22 , in the 250o pie region that excludes the X-ray tail From 1.5 to 3. 5 only.

velocity but it is located at > 3 , or > 6 Mpc, from the cluster center, and it is not considered as a possible member, Chapman et al 1988; it is instead one of 4 members of group GH14, Geller & Huchra 1983) to galaxies in groups and poor clusters (NGC 7619 and NGC 7626 in Pegasus I group; NGC 2563 in group A of the Cancer cluster) and galaxies in richer clusters (NGC 4649 in the Virgo cluster, at 1 Mpc from the cluster center). The five early type galaxies have been observed by the ROSAT PSPC as summarized in Table 1. The two observations of NGC2563 have been merged to improve on the statistical significance of the data, after checking that the single observations yielded consistent results. The data have been analyzed using primarily the PROS software available under IRAF. The details of the spatial and spectral analysis are reported in Sect. 2. Derived quantities are discussed in Sect. 3. Sects. 4­6 are devoted to the results for the Pegasus I and Cancer groups. The general results of these observations are summarized in the conclusions. 2. Data analysis The early type galaxies considered here all show extended emission in X-rays. Contour plots of their smoothed images in the energy band 0.14-2.0 keV are shown in Fig. 1, 8 and 9. In this energy band the expected effect of the vignetting correction is well represented by the exposure map (a model field containing information about the effective exposure moduled by the expected vignetting, see PSPC documentation for more details) provided with the data for each observed field (see also Trinchieri et al 1994, Kim & Fabbiano 1995). We can therefore use this map to flat field the data. The raw images have been normalized to the exposure maps, and smoothed with a Gaussian function, as specified in the figure caption. Several additional sources are visible, often embedded in the X-ray emission of the target galaxy. These possible interlopers were conservatively excluded from the subsequent analysis, unless otherwise noted, by masking out circles at the source positions with radii chosen to be comparable to the outer contour (typically at 2 ) on the contour maps. The presence of

several, possibly unrelated, sources in the field is not surprising. From the LogN-LogS function derived from ROSAT data (Hasinger et al 1993) we can estimate that 6 sources will be present in the inner 0.3 square degree field at fx 4 â10-14 erg cm-2 s-1 , and 20 if the flux limit is lowered to 2 â10-14 erg cm-2 s-1 . The NGC2563, NGC4649 and NGC7626 images show extended and azimuthally symmetric emission. The external X-ray isophotes of the NGC533 map instead show a more elliptical shape. A tail of emission is observed to the SW of NGC7619, extending for about 10 away from the X-ray peak. 2.1. Radial count distribution To estimate the size of the X-ray sources associated with these galaxies, we have produced plots of the radial distribution of the X-ray emission and compared them with the expected background, estimated from the properly normalized corresponding exposure map (Fig. 2). We have used concentric annuli, centered at the source peak position, with width varying from 0.25 to 2 - 3 depending on the counting statistics. The profile of the Pegasus I group is centered on NGC 7619. It appears that the profile of the NGC4649 field has the same shape as the exposure map from r 10 . However, two different normalizations are necessary to match the level of emission, one in the 10 - 16 annulus and a lower one outside of 25 (both shown in the figure). A more careful determination of the shape and level of the background will be necessary for this object; however, in this work we will consider both possibilities and will discuss the differences in the two cases. As can be seen from Fig. 2, the exposure map represents a good fit to the emission in the region outside 20 from the field center (e.g . outside of the ring of the support structure for the PSPC that has a shape reminiscent of a "wagon wheel" and that can be recognized as a "dip" in both source data and exposure map profiles). The net count distributions are plotted in Fig. 3 to 5 for each galaxy. Two Point Spread Functions (PSF), one appropriate for on-axis sources with temperatures of 0.2 keV (the softest


G. Trinchieri et al.: ROSAT PSPC observations of 5 X-ray bright early type galaxies

363

Fig. 1. Isointensity contour plots of NGC 533 and NGC 4649. Data in the 0.14-2.0 keV energy range have been normalized by the exposure map, and smoothed with a Gaussian function with = 30 . A constant background value has been subtracted. The dashed contour is for 0 counts. Other contours are at 2 , 3 , 4 , 5 , and higher.

Fig. 2. Radial surface brightness distribution of the raw counts, in concentric annuli centered at the X-ray peak position (coincident with the optical position of the galaxy's nucleus). The solid line represents the background model we have assumed for each field, derived from the exposure map properly rescaled to the data. The two lines shown for NGC 4649 represent two different normalization values (see text). The profile of NGC 7619 & NGC 7626 is centered on NGC7619, offset from the field's center. Consequently, the "dip" in the exposure map profile here is not as pronounced as in the other cases.

PSF available with the PROS software) and one for a kT corresponding to the best fit temperature of the innermost circle (see Sect. 2.2), are also plotted for comparison, normalized at the central bin. In all cases the emission is clearly larger than either PSF, although in most cases the 0.2 keV PSF is a reasonable representation of the data for r 1 . The spectral results however indicate significantly harder emission in that region (cf. Sect. 2.2). Fig. 4 and 5 show both the azimuthally averaged profiles of NGC533 and NGC7619 and the comparison of the emissions from different pie regions. In NGC 533 there is an indication of a significant change of slope at r 1 . This could be interpreted as due to the presence of an unresolved, very soft core, since the emission within 1 radius is well approximated by the 0.2

keV PSF. However, this is in contrast with the spectral results (Sect. 2.2), that indicate an average kT of 0.9 keV in the inner 1 circle. More likely, the difference is due to the dominating contribution of the galaxy relative to the group emission (1 30 kpc at the galaxy's distance). Fig. 4b compares the NE-SW sectors averaged together (i.e. a 70 cone along position angle PA=35 , counterclockwise from North) to that in the NW-SE ones (PA=125 ). Outside of r 5 - 6 , the surface brightness along PA=35 is flatter and significantly higher than that from the complementary regions. In Fig. 5b, the region SW of NGC7619 (from 170 to 280 , corresponding to the `tail' visible in the contour plot of Fig. 8), and the complementary region are compared. The SW profile appears significantly higher than in the other angular sector for


364

G. Trinchieri et al.: ROSAT PSPC observations of 5 X-ray bright early type galaxies

Fig. 3. Radial surface brightness profile of the net X-ray emission, in concentric circular annuli, for NGC 2563, NGC 4649 and NGC 7626. The power law fits of Table 2 are reported on the data.

Fig. 4a and b. Radial surface brightness profile of the net X-ray emission of NGC533, azimuthally averaged a, and in complementary pie regions b, as described in the text. In the inner 2 the profile in b) is azimuthally averaged as in a).

r> 2 , although it is not a simple radial function. This is consistent with a closer inspection of the tail in the X-ray map that shows a highly irregular morphology (see Fig. 8). More detailed discussion of the properties of this galaxy will follow. Although surface brightness profiles have traditionally been represented by a two parameter model (a core radius rx and a radial dependence parameter ), the core radii in these galaxies are smaller than the PSF of the instrument, as suggested by the higher spatial resolution data of some of these (see for example Trinchieri et al 1986), so that determining rx is not extremely meaningful. We have therefore fitted the radial dependence of

the profile at large radii with a simple power law fit (Table 2). Outside of a radius of 2 the influence of the PSF and of rx are negligeable. There is a large variety of radial dependencies of x , spanning from the flattest distribution of NGC2563 (x r-1.2 ) to the steepest NGC 4649 (x r-3.4 ). 2.2. Spectral data As already shown by previous studies (Trinchieri et al 1994; Buote & Canizares 1994; Fabbiano et al 1994; Bauer & Bregman 1996) the spectral capabilities of ROSAT do not allow us


G. Trinchieri et al.: ROSAT PSPC observations of 5 X-ray bright early type galaxies

365

Fig. 5a and b. Radial surface brightness profile of the net X-ray emission of NGC7619, azimuthally averaged a, and in complementary pie regions b. In the inner 1 the profile in b) is azimuthally averaged as in a).

to measure the metallicity of the gas, since the results obtained from models of varying metallic contents are at best ambiguous. We have therefore assumed a thermal plasma model with low energy absorption and fixed the gas abundances at the cosmic value. We have however checked how much different abundances would affect the results obtained and discuss the results at the end of this section. We have based the spectral analysis primarily on the IRAF/PROS software, which uses the same 34 channels of pulse height defined by the `SASS' (see ROSAT Data Production Guide, by Downes, White and Reichert). To obtain the temperature in different regions, the data are extracted in concentric annuli of varying widths. The background is taken in a region considered free of emission, from the same field to ensure that its spectral distribution is the correct one for that particular observation, and is then corrected for differential vignetting before subtraction, as described in the PROS documentation. In general, the annulus at 25 - 27 from the field center has been used to estimate the background. Other choices made for NGC 4649 and the Pegasus field are discussed below. An average temperature, over the entire source, is also derived for completeness. Tables 3a through 3e summarize the spectral results for each galaxy separately, graphically shown in Fig. 6. A cooler inner region, and a general tendency of increasing temperatures with galactocentric radii are seen, with the exception of NGC 2563, for which the temperature outside the inner 1 is isothermal at 1.1 keV. For NGC 533 a drop in the temperature is also observed outside of a 6 radius, although with large errors. There is no clear trend of varying low energy absorption with radius, with the exception of the central bin of NGC 4649, where the fitted values of NH is higher than outside. Often however NH is not very well constrained. Briefly for each galaxy: NGC 533 : The average gas temperature within r 6 , i.e. before the radial profile becomes asymmetric, is 1 keV, in the assumption of a 1-Temperature model. This however gives a relatively high 2 value of 42.9 for 25 degrees of freedom min (DOF), that is significantly lowered (2 19.6) by adding a min

second temperature in the model. The two best fit temperatures become kT1 =0.30-0.90 [at the 90% confidence limit] and kT2 > 1.2 keV. The suggestion that at least 2 temperatures are present in the data is well justified by the temperature profile obtained at different radii. In fact, when the emission is divided in 3 concentric annuli, the temperature increases from 0.9 [0.840.94] at the center to > 1.1 keV in the outermost bin. If the whole 17 region is considered, the average temperature also is around 1 keV, however the errors become larger and the 2 value does not require a two-temperature fit as in min the case above.

NGC 2563 : The spectral distribution of the data obtained in the two observations have been compared to assure ourselves of their consistency, before merging the two data sets for the following analysis. The average gas temperature within r=15 is kT 1.06 keV, with an acceptable 2 min = 24 for 23 DOF. The temperature profile obtained in 7 concentric annuli indicates a cooler core in the inner 1 radius, and a constant 1.1 keV value outside, although a possible indication of a temperature decrease with radius is visible in the data (see Fig. 6).

NGC 4649 : Given the uncertainty in the normalization of the background template, we have considered both the 25 - 27 annulus used for all other fields, and an annulus at 11 - 13 for background estimates. This latter choice has the advantage that the vignetting corrections are reduced. Moreover, if the Virgo cluster is responsible for the "extra" emission outside of 7 , we would subtract it from the galaxy data. A comparison of the spectral distribution of the data in the two regions (11 - 13 vs 25 - 27 ) however shows that they are consistent within the errors. In the present analysis we have chosen the 11 - 13 annulus as background. The one temperature fit for the whole source within r 6 gives a best fit kT=0.85, with a 2 of 47.9 for 25 DOF. The min introduction of a second model component lowers the 2 to min 30.5 for 23 DOF, and gives kT1 = 0.64 - 0.84 keV, and kT2 > 1.2


366

G. Trinchieri et al.: ROSAT PSPC observations of 5 X-ray bright early type galaxies

Table 3. Spectral Results for each Galaxy in different regions Table 3a. NGC 533 Ann. Net counts () 0-1 1-3 3-6 6-10 10-17 1334.4±41.6 812.7±36.9 897.8±46.1 985.8±63.9 353.6±41.3 3093.0±70.7
2 min

Chan.

kT (keV) 0.904 1.074 1.299 0.941 1.024

5-28 5-30 5-28 5-28 5-8 10-25 5-32

90% confidence region 0.84-0.94 0.99-1.22 >1.07 0.83-1.08 >0.80

NH (cm

-2

)

20.41 20.23 20.22 19.74 19.60

90% confidence region 20.25-20.55 20.05-20.40 <20.45 <20.20 <20.20



Notes

23.5 16.2 18.9 9.6 12.2

21 23 21 21 17

0-6

1.028 0.552 1.228

0.98-1.07 0.30-0.90 >1.2

20.26 20.33

20.20-20.35

42.9 19.6

25 23

1

Table 3b. NGC 2563 Ann. Net counts () 0-1 1-3 3-5 5-7 7-9 9-12 12-15 1143.5±40.5 1302.8±53.0 1176.3±60.1 1308.1±72.1 1269.0±82.6 1867.4±116.6 1149.6±127.7 9050.1±312.0

Chan.

kT (keV) 0.848 1.177 1.244 1.192 1.025 1.075 1.014

5-30 5-30 5-30 5-30 5-30 5-10 13-30 5-10 13-26 5-30

90% confidence region 0.80-0.92 1.07-1.38 1.10-1.55 1.08-1.38 0.91-1.15 0.97-1.27 0.83-1.35

NH (cm

-2

)

20.40 20.40 20.40 20.38 20.22 20.00 19.60

90% confidence region 20.20-20.60 20.20-20.60 20.15-20.70 20.15-20.55 <20.70 <20.35 <20.60



2 min



Notes

17.8 22.1 19.4 19.1 20.7 17.5 15.3

23 23 23 23 23 21 17

0-15

1.058

1.01-1.14

20.20

20.00-20.40

23.9

23

Table 3c. NGC 4649 Ann. Net counts () 0-1 3670.0±66.1 1237.2±44.0 463.8±34.8 388.7±61.7 5741.2±112.9

Chan.

kT (keV) 0.814 0.756 2.91 0.898 0.923 1.315

5-30

1-2 2-3 3-6

5-28 5-7,9-29 6-9,11,12 14-26 5-32

90% confidence region 0.78-0.83 0.46-0.80 >1 0.83-0.96 0.78-1.08 >0.93

NH (cm

-2

)

20.28 20.36 19.99 19.74 19.60

90% confidence region 20.25-20.35 -- 19.75-20.20 <20.25 < 20.60



2 min



Notes

27.2 14.1 15.3 14.0 15.8

23 21 21 21 16

1

0-6

0.853 0.793 3.000

0.82-0.90 0.64-0.84 > 1.2

20.10 20.20

19.95-20.30 --

47.9 30.5

25 23

1

Table 3d. NGC 7619 Ann. Net counts () 0-1 1-3 3-10 0-10 972.4±36.8 858.3±39.6 1073.6±49.1 2886.4±72.6

Chan.

kT (keV) 0.775 0.885 0.948 0.880 0.812 3.00 1.267

5-30 5-29 6,7,9-30 5-29

90% confidence region 0.70-0.82 0.82-0.94 0.88-1.02 0.84-0.92 0.60-0.88 > 1.1 1.10-1.90

NH (cm

-2

)

20.59 20.55 20.41 20.58 20.67

90% confidence region 20.40-20.85 >20.25 20.10-21.10 20.40-20.85 --



2 min



Notes

12 17.6 19.2 27.8 19.5

23 22 21 22 20

2 2 1

"group"

2150.1±143.6

5-7,9 11-29

20.04

19.50-20.40

13.3

20 3


G. Trinchieri et al.: ROSAT PSPC observations of 5 X-ray bright early type galaxies Table 3. (continued) Table 3e. NGC 7626 Ann. Net counts () 0-1 0-3 404.4±26.1 613.5±34.1
2 min

367

Chan.

kT (keV) 0.737 0.826

6-30 8-28

90% confidence region 0.58-0.88 0.68-0.90

NH (cm

-2

)

20.48 20.50

90% confidence region >20.20



Notes

18.9 25.2

22 18

Notes to the table. The line-of-sight NH value is: 20.5 for NGC 533; 20.7 for NGC 2563, NGC 7619 and NGC 7626; 20.4 for NGC 4649 (Stark et al 1992). 1 A two temperature fit is used. kT1 and kT2 are given on successive lines. The errors are derived for kT only 2 Limited to the 170 -280 sector in the 3 - 10 annulus. 3 "group" is the region from 2 to 16 , centered on NGC 7619, and excluding the 170 -280 sector.

Fig. 6. Summary plot of the temperature parameters derived for different annular regions in the sample galaxies.

keV at the 90% confidence level. No further reduction of 2 min is achieved with the addition of a third component. A slow increase of T towards large radii, and a higher value of the absorbing column in the innermost 1 circle are suggested by the data. Pegasus I group : The contour map and the radial profile show that the emission from the Pegasus I group field covers the inner 20 circle, and it is likely to extend further out, with the possible exception of a region north of the NGC7619 galaxy. We have

therefore tried both this region of low emission and the annulus at 25 - 27 for background estimates. The former choice would have the advantage of smaller corrections, but with a larger statistical uncertainty due to the smaller area covered. As for the case of NGC4649 however, a comparison between the spectral distribution of the data in the two regions shows consistency within the statistical errors. We have therefore used the outer annulus as background in the following analysis. The source associated with NGC7619 has been divided into two annuli within r=3 , and the "tail", in the SW quadrant, from


368

G. Trinchieri et al.: ROSAT PSPC observations of 5 X-ray bright early type galaxies

3 - 10 . Table 3d indicates that there is a significantly cooler 1 core, and that the tail is harder than the galaxy. The group gas (in the 2 - 16 region, excluding the "tail") is also harder than the tail. The low energy absorption is not well constrained, and is in general consistent with the line of sight value. NGC7626 extends only out to r3 , and the total number of counts detected does not allow us to examine the spectral distribution of the data in different regions. The average temperature over the whole 3 circle is kT0.83 keV with a 2 of 25.2, min which is entirely consistent with the temperature obtained from just the innermost 1 circle. We have also examined the spectral distribution of the data of the galaxies above the possible cluster emission by assuming as background the annulus at 4 - 10 from the field center, with the exclusion of a 3 circle at the position of the two galaxies. The resulting best fit parameters are consistent with those reported in Table 3.

3.2. Density profiles The radial profiles, together with the spectral results, have been used to derive the radial dependence of intrinsic gas parameters: electron density, cooling time and total gas mass. Since the observed surface brightness profile is the integral along the line of sight, a deprojection of the observed quantity is needed to obtain the intrinsic distribution (Kriss et al 1993). However, the temperature of the gas must be known to derive its density, and the preceding analysis has only provided us with the projected values. A proper deconvolution of the spatial/spectral data is needed to derive this quantity. We have however approximated this procedure with the following assumptions: 1) the gas is assumed homogeneous, i.e. the gas has only one temperature at any given radius; 2) the temperature is constant in any given annulus; 3) the observed projected value of the temperature is not significantly different from the de-projected intrinsic value. This last assumption is supported by the results of the procedure used in the analysis of NGC4636 data. There Trinchieri et al (1994) had applied the same de-projection used to derive the intrinsic surface brightness distribution to the spectral data as well, and have derived the best fit parameters and errors based on the deprojected distribution. As can be seen from the comparison of their Tables 1, 2 and 5, the temperatures are consistent, within the errors. Moreover, it should be remembered that the density depends weakly on the temperature, therefore small changes in kT should not affect the results significantly. Finally, while assumption 1) is probably strictly incorrect, as the temperature is in most cases an increasing function of the radius, the increase is small and the data quality is such that this approximation should not be a serious source of error. Moreover, no correction for the effect of vignetting at large radii is applied (however the error introduced by this should be small). T he case of N GC 4649: The density profile has been derived in the assumption that the emission outside of 7 should be considered background emission relative to the galaxy. This assumption is justified by its shape (see Sect. 2) and also by the fact that the galaxy is located in the Virgo cluster, relatively close to M87 (< 1 Mpc, for D=17 Mpc) where the cluster emission is still relatively strong (Boehringer et al 1994). The radial distribution of the net emission shown in Fig. 3 (panel b) also indicates that the profile flattens quite significantly outside of 7 , when the background is taken far from the galaxy. This could be justified again by the presence of a low surface brightness emission such as that of a cluster gas. T he tail of N GC 7619: An asymmetric feature is present to the SW of NGC7619. Even though the asymmetry is already present at a radius r 2 , we have derived the density profile within a radius of 4 assuming azimuthal symmetry. This radius is estimated from the radial profile of the region complementary to the tail, and corresponds to a possible flattening of the profile. The density profiles are shown in Fig. 7. Their parameterization is given in Table 5. Central densities span a relatively large range of value, 5-15 â10-3 cm -3 [Note that these are probably underestimated due to the influence of the PSF which has not been removed]. This spread however could be due to

Metal Abundance In order to explore the effect of different abundance models on the temperatures derived, we have redone the spectral fits for a few of the cases presented in Table 3, where the statistics is higher, fixing the abundance parameter at 20% the cosmic value. The absolute values of the temperatures are smaller (< 0.1 keV in all cases examined, and typically is 0.05 - 0.06) and the low energy cutoff values are higher (typically, 0.3 in log NH ) in the case of 20% abundances, but the trends and the uncertainties on T are preserved. The values of 2 are also min consistent with one notable exception, given by the global spectrum of NGC533 (in the 0 - 6 annulus). The 20% abundance model gives an acceptable 2 =24 for kT=0.95 and NH =20.6 min (2 = 43 for the 1-temperature model at 100% cosmic abunmin dance, Table 3). To obtain a similar 2 , a 2 temperature model min must be assumed for 100% cosmic abundances. This is however an indication of the ambiguity with which the spectral parameters can be derived from the limited spectral resolution of the ROSAT PSPC (cf. Fig. 7 in Trinchieri et al. 1994).

3. Derived quantities 3.1. Fluxes and luminosities Table 4 summarizes the resulting fluxes and luminosities for the sample galaxies within the given radius Rm . The counts to flux conversion is derived from the parameters of the best fit spectral model of the source as a whole. However, the fluxes are entirely consistent with those obtained by summing the contribution of each annulus, accounting for the different spectral parameters obtained for each of them. Luminosities are in the range Lx 2 - 60 â 1041 erg s-1 , at the top of the distribution of X-ray luminosities of early type galaxies (see Fabbiano et al 1992).


G. Trinchieri et al.: ROSAT PSPC observations of 5 X-ray bright early type galaxies Table 4. Global Fluxes and Luminosities Galaxy Ra m (kpc) 18(565) 20(560) 7(35) 17(84) 10(225) 3.5(75) Counts
a

369

Flux erg cm s-
-2

b 1

Lb x erg s-1 5.90â10 2.53â10 1.61â10 2.02â10 1.51â10 2.83â10
42 42 41 41 42 41

D Mpc 108 96 17 17 75 75

NGC NGC NGC NGC NGC NGC
a

533 2563 4649c 4649d 7619e 7626

4321±115 8612±220 5890±103 7418±197 2980±60 623±34