A&A 378, 30-39 (2001)
H.-U. Zimmermann - Th. Boller - S. Döbereiner - W. Pietsch
Max-Planck-Institut für Extraterrestrische Physik, Postfach 1312, 85471 Garching, Germany
Received 5 February 2001 / Accepted 3 August 2001
In a correlation study of the ROSAT All-Sky Survey Bright Source Catalogue (RASS-BSC) with the Catalogue of Principal Galaxies (PGC) 904 X-ray sources were found that possess possible extragalactic counterparts within a search radius of 100arcsec. A visual screening process was applied to classify the reliability of the correlations. 547 correlations have been quoted as reliable identifications. From these, 349 sources are known as active galaxies. Although for the other sources no hints for activity were found in the literature, 69% of those for which we have distances show X-ray luminosities exceeding those of normal galaxies, a clear sign that these galaxies also own hitherto unreported X-ray active components. Some objects are located inside or in the direction of a known group or cluster of galaxies. Their X-ray flux may therefore be in part affected by hot gas emission. Luminosity and distributions are used to characterize different subsamples. Nuclei that are both optically and X-ray active are found predominantly in spirals. Two special source samples are defined, one with candidates for X-ray emission from hitherto unknown groups or clusters of galaxies, and one with high X-ray luminosity sources, that are likely candidates to possess hitherto unreported active galactic nuclei. Besides a compilation of X-ray and optical parameters, also X-ray overlays on optical images for all the objects are supplied as part of this work.
Key words: surveys - X-rays - galaxies: general
In 1990/91 the ROSAT observatory, using the PSPC detector in the focus of the X-ray telescope, scanned the whole sky during a half year period. It established the first all sky survey in the X-ray regime with an imaging telescope (Trümper 1983). Using a maximum likelihood source detection algorithm (Cruddace et al. 1987; Boese & Döbereiner 2001) about 80000 sources with a likelihood 10 have been found in the sky images of the ROSAT All-Sky Survey (RASS). Different attempts are underway to identify parts of this large data base with counterparts in other wavelength regimes to enable investigations of the broad band properties of different astronomical object classes.
This paper reports identification results obtained during the preparation of a study on the "X-ray properties of normal galaxies''. From observations with the EINSTEIN observatory it is known that the X-ray emission of normal galaxies is on the order of (1989). This should allow to detect them in the ROSAT All-Sky Survey data with typical exposures of a few hundred seconds up to distances of 100Mpc.
Galaxies cannot be identified in the RASS data solely by their X-ray signature. We therefore had to correlate RASS sources with an optical database comprising most of the galaxies in the relevant distance regime. After different attempts we finally concluded that the Catalogue of Principal Galaxies (Paturel et al. 1989), updated by the NASA/IPAC Extragalactic Database (NED), best fulfilled our purposes for a rather complete sample of the objects we are interested in the study.
For the study it is important to include only objects where the X-ray data promise high reliability for the existence of the source and allow some confidence in the derived positions and intensities. We found that the selection criteria, as applied for the compilation of the ROSAT All-Sky Survey Bright Source Catalogue (RASS-BSC) (Voges et al. 1999), are a reasonable basis for our work.
The RASS-BSC is a subset of the RASS database comprising 18811 X-ray sources selected from the whole sample by the following criteria:
The width of the effective survey point spread function (FWHM about 45arcsec) and the use of a maximum likelihood (ML) detection algorithm (Boese & Döbereiner 2001; Cruddace et al. 1987), delivering best estimates for point-like sources, limit the reliability of the derived source properties in cases where 2 or more X-ray sources are within the acceptance window used by the algorithm (typically a circle of 300arcsec diameter) and in cases where X-ray sources are clearly extended. The authors of the BSC paper demonstrated that 90% of point-like X-ray sources (stars) have a positional accuracy of 25arcsec.
The exposure times of the objects in the final X-ray/PGC correlation sample
range from 56 to 17165s, with 79% of the
sources having exposures between 200and600s
(see Fig. 1).
|Figure 1: Distribution of exposure times for the X-ray/PGC correlation sample (547 objects). Not shown are 8% of the observations with exposure times >800s.|
|Open with DEXTER|
Since the compilation of the PGC, for many individual galaxies updated information on some of the galaxy parameters became available. This is especially true for the positions of individual galaxies, where the accuracy in the original PGC often presents a major problem for unique correlations. We therefore decided to base the optical galaxy sample on the PGC objects but, for each individual object, rely on the parameters as given in the NASA Extragalactic Database (NED). The other possibility, to start directly from the NED data base, had also been considered. While the number of positional correlations, 4418 within a search radius of 100arcsec, is about a factor of 4 higher than with the PGC, the number of normal galaxies (not known to be active) is only marginally larger, as was found by checks. Due to the effort in screening visually 4 times more correlations while not detecting many more of these objects, we stayed with the PGC membership as the main galaxy selection criterium.
For 61 sources of our sample we performed optical observations (Pietsch et al. 1998; Bischoff et al. 1999) to better evaluate the position (accuracy <2arcsec) and classification of these objects. The optical observations will be continued in future projects.
|Figure 2: Distribution of distances between objects in the PGC and the BSC catalogs for search radii up to 300arcsec.|
|Open with DEXTER|
|Figure 3: Sky distribution of the 904 X-ray sources in our sample in galactic coordinates. Galactic longitude starts at the center. The lack of galaxies near the galactic plane is due to high absorption in our galaxy.|
|Open with DEXTER|
In such a representation the true correlations show up as a peak near small radii, while chance correlations between the two catalogs produce the right-hand part of the profile that is slowly rising with increasing radius. The width of the initial peak is broadened mainly due to positional errors in the two data bases. 90% of the BSC X-ray point sources have a mean position error of 25arcsec, while the PGC catalog does not provide explicit position errors. In many cases they were found to be appreciable larger. Additional broadening may arise when the maximum of the X-ray flux originates at positions off the optical center of the galaxy. Also the optical center of irregular nearby galaxies is not always easily determined. To include the vast majority of true correlations and at the same time accept only a small number of chance coincidences we chose a search radius of 100arcsec, at the minimum of the distribution in Fig. 2. From the figure we estimate that up to 7% of real correlations are missed by that choice while the number of chance coincidences (before screening) is about 12%.
The working sample is defined by 904 X-ray sources from the BSC surrounded by 1124 objects from the PGC lying within a radius of 100arcsec of the X-ray sources. The reduced spatial resolution in the X-ray data of this survey prevents in many cases to resolve multiple galaxy systems into their individual components.
Figure 3 shows the all sky distribution of the X-ray sources in the investigated sample.
Besides chance coincidences there may be other reasons that can put a correlation in question. In a further step we therefore tried to identify which of the correlation pairs are real, that is, which of the galaxies are also the sources of the observed X-ray emission.
For this purpose we overlaid X-ray contours smoothed by a 45arcsec Gaussian on an optical image (POSS plates, see acknowledgements) of about size centered on the position of the X-ray source (see examples in Figs. 4 to 7).
|Figure 4: Examples for the images used during the screening process. These 2 images represent examples with "very good'' reliability: on the left an optically resolved Seyfert-1 object with well centered X-ray contours, on the right a Seyfert 2 object at similar space distance, with the expected smaller X-ray intensity. The cross indicates the X-ray position, squares the PGC positions and circles the NED positions. The numeric labels at the different positions point to the entries in the name lists displayed in the left part of the map. X-ray contour levels (in sigma above background) are denoted at the lower left corner. Count rate and exposure of the X-ray source are given, together with the identification number of the X-ray source (xID, from Col. 2 in Table 1), in the upper right corner.|
|Open with DEXTER|
|Figure 5: The reliability of these 2 images is quoted as "good'': on the left a 1.8 Seyfert, optically resolved, with poor X-ray contours, on the right an example, where the optical position of the Seyfert 1.5 object is off from the X-ray object by almost 50 arcsec, but where the optical position error is stated as 150arcsec. The optical parameters (magnitude, extent) and the X-ray properties (contours, intensity) led us to accept the correlation as correct.|
|Open with DEXTER|
|Figure 6: 2 images that were quoted having questionable reliability: the left image shows X-ray contours that indicate extended hot gas, perhaps from the central cluster, while the right image is an example where we could not decide which of the optical galaxies is the probable counterpart and whether we see only hot gas in the galaxy group SCG 16.|
|Open with DEXTER|
|Figure 7: 2 examples where a correlation was rejected: in the image on the left the counterpart due to its optical appearance is a star with no obvious galaxy connection, while the X-ray emission in the right image, looking at the contours, is dominated by hot cluster gas.|
|Open with DEXTER|
In these images the original positions of the PGC sources and additionally all sources from the NED database up to a distance of 3arcmin around the X-ray counterpart are marked by white squares (PGC) and circles (NED) as well as the center of the X-ray emission (cross) as derived by the maximum likelihood process. Besides the appearance of the X-ray contours and the distance to the optical counterpart, also the position and object properties of the other sources in the field, like known activity or the membership in a group or cluster of galaxies, entered the qualification process.
The reliability of the correlation was estimated and expressed by the following 4 terms: no, questionable, good, and very good.
This classification scheme must be regarded as a first approach in determining whether the observed X-ray emission originates from the optical galaxy in question. A reliable answer is sometimes difficult or even impossible. Part of the problem arises from the X-ray data itself: often the X-ray contours originate from little more than a dozen photons, leading to poorly determined positions and contours. But also the optical positions of the NED report for 86 of the 1124 optical objects uncertainties between 30 and 300arcsec, making it sometimes difficult to identify uniquely the objects in the optical images.
Specific difficulties were encountered when the X-ray emission came from the direction of a known or suspected group or cluster of galaxies. From the X-ray data it is usually not possible to decide how much of the X-rays stem from the hot cluster gas and how much can be attributed to a single galaxy. Many clusters contain a dominant CD galaxy in their gravitational center, that forms also the X-ray emission center of the hot cluster gas.
On the other hand, if statistics are good enough, the X-ray contours from known clusters of galaxies show characteristics, that allow in some cases a recognition of these extended structures also in images where no explicit hints for the presence of a group or cluster of galaxies have been found in the literature. The source extent parameter, calculated by the ML source algorithm, in many cases gives first hints to these sources. In the main table we have marked all those X-ray sources, in the neighbourhood of which (3arcmin) no group or cluster of galaxies are reported (NED), but where the X-ray contours indicate relevant features. In addition to the contours also optical field properties entered the decision process (e.g. type of galaxy counterpart, no obvious counterpart, high density of galaxies, X-ray emission shifted, etc.).
In the case of narrow galaxy pairs or triplets, that were not resolved by ROSAT, we marked only one member, the most probable one, as the identified counterpart.
To give the reader a feeling of the considerations involved in the screening, Figs. 4 to 7 show some typical examples for the 4 reliability levels, including the principal lines of reasoning in the figure captions.
|SEQU||running number in table.||pos_err||optical position uncertainty in arcsec.|
|xid||identification number of the X-ray object.||type||type of source.|
|sid||identifies the optical candidates for a specific X-ray source.||type1||detailed type of optical candidate.|
|name_x||IAUC identifier for the X-ray source.||activity||type of activity.|
|RADEC_x||RA and DEC positions (2000) of X-ray source.||velocity||recession velocity in km/s.|
|exposure||exposure time of X-ray source in s.||or||origin of parameters: 1 from PGC only, 2 from NED,|
|pos_err_x||uncertainty of X-ray position in arcsec.||12 from NED + (Pietsch et al. 1998),|
|rate||count rate of X-ray source in counts/s (0.1-2.4 keV).||22 from NED + (Bischoff et al. 1999).|
|err_rate||uncertainty of rate in counts/s.||cl||whether in binary (1), group (2), or cluster of galaxies (3).|
|lh||likelihood of detection.||or line-of-sight projection near to cluster of galaxies (3).|
|counts||total counts from X-ray source.||galactic hydrogen column density in .|
|ext_lh||likelihood that X-ray source is extended (interpret with care!).||dist||distance between X-ray object and optical candidate (arcsec).|
|extent||extent in arcsec||logFlux_x.||log10 of X-ray flux in in the 0.1-2.4 keV band.|
|name_opt||name of optical candidate||log_Lx.||log10 of X-ray luminosity in erg/s ).|
|RADEC_opt||RA and DEC positions of optical candidate.||correlation||reliability of correlating the X-ray and optical source.|
|mag||B magnitude of optical candidate (NED).||comment||comments; cl? means source in suspected cluster sample.|
Table 1 shows the structure and content of the sample list. The example entries in the table refer to the sources shown in Figs. 4 to 7. The information on each source covers 2 lines. After 2 columns with entry identification follow 5 columns with X-ray parameters, specifying X-ray name, position, exposure, rate and counts in the 0.1-2.4keV energy band, detection likelihood, and information on possible extent (mostly taken from BSC). Then follow 5 columns with optical data, name, position, magnitude, galaxy type, activity, grouping info, and velocity (mostly taken from NED, in parts also from Pietsch et al. 1998; Bischoff et al. 1999). The last few columns contain an indicator of the data origin, the galactic absorbing column density towards the object (Dickey & Lockman 1990), the distance between the X-ray source and the optical candidate, the X-ray flux and luminosity, as well as the reliability of the correlation. X-ray fluxes and luminosities have been calculated assuming a power law with photon index 2.3 (Hasinger et al. 1991; Walter & Fink 1993), the relevant galactic absorbing column density and, if available, the distance to the source. For some sources additional remarks are available in column comment, which contains also the membership to the suspected cluster sample (marked as cl?) mentioned in Sect. 4 on Data Screening.
The complete table, with 1124 correlation entries for the 904 X-ray sources, as well as the X-ray contour overlays on optical images for all of the 904 X-ray sources, is available in electronic form at the CDS via anonymous ftp to cdsarc.u-strasbg.fr (22.214.171.124) or via http://cdsweb.u-strasbg.fr/cgi-bin/qcat?J/A+A/378/30
The reliability of the X-ray/PGC correlations for the 904 X-ray sources of the complete sample was denoted as "very good'' for 367 objects, as "good'' for 180, and as "questionable'' for 161 objects. The remaining correlations were classified as being not real, that is they are either chance coincidences or their X-ray flux originates predominantly from extended emission of hot gas in galaxy groups or clusters.
The following discussion refers only to the 547 objects of the reliable correlation sample, that is, the objects having correlation reliabilities good or very good. These are grouped in 2 categories:
In the reliable correlation sample of 547 galaxies, for 488 objects we have distance information and therefore can calculate their X-ray and optical luminosities from the relevant fluxes, adopting a Hubble constant , which is in the range of values determined from the HST distance scale project (see e.g. Mould et al. 2000). Figure 8 shows the X-ray luminosity versus the ratio of X-ray to optical luminosities (or fluxes) in these objects, demonstrating the well-known correlation between optical and X-ray luminosities (see e.g. Fabbiano et al. 1992). Optical fluxes and luminosities have been derived from the optical Bmagnitudines as provided by NED with no attempt to include bolometric corrections. The 6 objects at the high end of the X-ray luminosity scale are all quasars, 3C273 and the 5 objects in our sample with . All the other objects with known distances have z < 0.5.
|Figure 8: X-ray luminosity versus the ratio of X-ray to optical luminosity (or flux) for sample objects with distance information.|
|Open with DEXTER|
The optical activity as seen in Liners, Seyferts, QSOs, or Blazars originates from activity in the nucleus of the galaxy. While the high energy output of the nucleus, over many stages of absorption and reemission, finally deposits part of its initial energy in the observed optical output, the X-ray flux originates from regions closer to the central engine and therefore reflects generally more directly the activity in the central engine. In the optical regime therefore the nucleus usually contributes only indirectly to the overall output of the galaxy and absorption and galaxy geometry and aspect (with respect to the observer) play an important part. The X-ray flux from the nucleus, on the other hand, is the dominant part of the X-ray emission in the majority of X-ray active galaxies.
This is also reflected in the X-ray and optical luminosity distributions for the active galaxies and candidate galaxies in our sample. Figure 9 shows in the upper panel the X-ray luminosity distribution for both the 198 candidate galaxies (dashed line) and the 349 active galaxies (full line). As one would expect the distribution of active galaxies is shifted towards higher X-ray luminosities in comparison to the candidate galaxies. The optical luminosity distribution (lower panel), on the other hand, does not show the same dependency (lower panel).
|Figure 9: X-ray luminosity distribution (upper panel) and optical luminosity distribution (lower panel) of galaxies having known distances in sample candidate galaxies (dotted line) and in sample active galaxies (full line).|
|Open with DEXTER|
|Figure 10: Optical luminosity distribution of the spiral galaxies (full line) and the elliptical plus S0 galaxies (dashed line) independent of their activity status.|
|Open with DEXTER|
There is, however, as Fig. 10 shows, a difference in the width of the optical luminosity distribution between the spiral galaxies (full line) and the elliptical plus S0 galaxies (dashed line). The larger width of the distribution of spirals possibly derives from its broader scale in surface brightness due to the larger range of geometries compared to the elliptical plus S0 galaxy sample.
|Figure 11: distributions of the X-ray (circles) and optical (squares) flux S for 477 sample sources.|
|Open with DEXTER|
|Figure 12: distributions of the X-ray flux S for spirals (X) and elliptical plus S0 galaxies (circles), in the upper panel for the active galaxies sample, in the lower panel for the candidate galaxies. Optical activity is predominantly observed in spirals.|
|Open with DEXTER|
The distribution of the sample for both the X-ray and the optical flux is shown in Fig. 11. It demonstrates that the brightest X-ray galaxies have been detected rather well, with a gradient near to -1.5 as expected for an Euclidian source distribution, while the X-ray survey clearly is an extremely inefficient method to count optical galaxies, with a slope at the bright end near to 2/3, corresponding to about 3% detections. But it has to be stressed, that our sample in no way forms a complete sample.
Figure 12 shows the distribution for spiral galaxies and for elliptical plus S0 galaxies for both the active galaxies (upper panel) and for the candidate galaxies sample (lower panel). In the active galaxies sample spirals show up about a factor of 3 more abundant than in the elliptical plus S0 subsample, while this ratio has a value of about 0.8 for the candidate galaxies, showing that optical activity combined with X-ray activity is by far more frequent in spirals.
Taking the candidate galaxies sample, we may define a normal galaxy as one where the X-ray to optical luminosity ratio does not exceed 10-2. As can be seen from Fig. 8 this definition limits the maximum X-ray luminosity of our sample sources to about , in agreement with Fabbiano (1989). 97 of the candidate galaxies with known distances have X-ray luminosities >1042ergs-1, clearly indicating major X-ray activity also in these objects. In a search for normal galaxies remain 43 candidate galaxies with X-ray luminosities <1042ergs-1 and 58 galaxies with missing distance information as candidates for galaxies that possess no apparent active components in the optical or X-ray regimes.
At the X-ray high luminosity end of the candidate galaxies sample, there are 23 galaxies with reliability being "very good'', for which the X-ray luminosity exceeds our limit for a normal galaxy by more than a factor of 10, that is , and that are not inside or near the line of sight of a group or cluster of galaxies. This high X-ray luminosity sample forms an inhomogeneously studied subsample of galaxies, where a detailled classification will be required to reveal their true nature. As can be seen from Table 2, 11 of the 23 objects are reported also as infrared sources, 8 show radio emission, in 6 cases UV emission has been observed, and 2 are emission line galaxies (NED). For 4 galaxies no detections in other wave bands are reported. The study of the sample is part of a future project. Optical follow-up observations have been initiated.
|SEQU||entry number in main table.|
|name||name of optical candidate.|
|log_Lx||log10 of X-ray luminosity in ergs-1 ( ).|
|type||type of source (NED).|
|comment||detections in other bands: Rad means detected as radio source,|
|IrS as infrared source, UvE detected in the ultraviolet band,|
|EmL means emission line galaxy.|
In a search for X-ray emission of normal galaxies in the ROSAT All-Sky Survey, we correlated the 18811 RASS-BSC sources with the 73197 galaxies from the PGC, with their main parameters updated by the NED database. In 904 cases we detected within a search radius of 100arcsec around the X-ray source one or more candidates as optical counterparts (the complete sample). A visual screening of optical images overlaid with the observed X-ray contours left finally a sample of 547 optical/X-ray correlations that we regard as sufficiently reliable (reliable correlation sample). For the vast majority of these X-ray emitting galaxies pronounced activity in the optical band (Liners, Seyferts, QSOs, Blazars) has been reported in the literature and the typical X-ray luminosities are lying between 1042 and 1045ergs-1. For 61 objects data from optical follow up observations (Pietsch et al. 1998; Bischoff et al. 1999) have been used in addition to the NED data to identify and classify the galaxies. The spectral information of the RASS data for the individual galaxies was too scarce to gain additional screening information. There remain 101 sources in our sample as candidates for galaxies with no apparent active components, while a majority of 446 objects definitely belong to the class of objects with active components. This is not surprising, given the fact that sources with luminosities of 1040ergs-1 or less are visible in a typical RASS exposure of order 400s only up to distances of about 10Mpc.
Thus correlating X-ray surveys with optical galaxy catalogues is an efficient way to detect galaxies with X-ray active components. In our sample, nuclei both optically and X-ray active, are found predominantly in spirals. Luminosity histograms and distributions are used in the further characterization of the full sample and different subsamples.
Two special subsamples, interesting for further investigation, are defined: a reliable high X-ray luminosity sample of 23 objects, until now not reported as active nuclei, and a suspected cluster sample of 101 sources where X-ray emission is supposed to originate, at least in parts, from a group or cluster of galaxies.
In addition our work provides soft X-ray parameters, identification maps, and screening results for more than 500 galaxy identifications, useful as a basis for follow-up observations both in the optical and X-ray regimes.
The ROSAT project is supported by the Bundesministerium für Bildung und Forschung (BMBF/DLR) and the Max-Planck-Gesellschaft (MPG). This work made use of the NASA/IPAC Extragalactic Database (NED), which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with NASA. The optical images used for the X-ray overlays are taken from the Digitized Sky Survey, that was produced at the Space Telescope Science Institute under US Government grant NAG W-2166. It is based on photographic data of both the National Geographic Society - Palomar Observatory Sky Survey (NGS-POSS) obtained using the Oschin Telescope on Palomar Mountain, and the UK Schmidt Telescope. The NGS-POSS was funded by a grant from the National Geographic Society to the California Institute of Technology. The UK Schmidt Telescope was operated by the Royal Observatory Edinburgh, with funding from the UK Science and Engineering Research Council, until 1988 June, and thereafter by the Anglo-Australian Observatory. Original plate material is copyright (c) the Royal Observatory Edinburgh and the Anglo-Australian Observatory.