A&A 477, 735-746 (2008)
DOI: 10.1051/0004-6361:20078568

The XMM-Newton bright serendipitous survey[*],[*]

Identification and optical spectral properties

A. Caccianiga1 - P. Severgnini1 - R. Della Ceca1 - T. Maccacaro1 - F. Cocchia1,2 - X. Barcons3 - F. J. Carrera3 - I. Matute6 - R. G. McMahon4 - M. J. Page5 - W. Pietsch6 - B. Sbarufatti7 - A. Schwope8 - J. A. Tedds9 - M. G. Watson9


1 - INAF - Osservatorio Astronomico di Brera, via Brera 28, 20121 Milan, Italy
2 - INAF - Osservatorio Astronomico di Roma, via di Frascati 33, 00040 Monte Porzio Catone, Italy
3 - Instituto de Física de Cantabria (CSIC-UC), Avenida de los Castros, 39005 Santander, Spain
4 - Institute of Astronomy, Madingley Road, Cambridge CB3 0HA, UK
5 - Mullard Space Science Laboratory, University College London, Holmbury St. Mary, Dorking, Surrey RH5 6NT, UK
6 - Max-Planck-Institut für extraterrestrische Physik, Giessenbachstrasse, 85741 Garching, Germany
7 - INAF - IASFPA, via Ugo La Malfa 153, 90146 Palermo, Italy
8 - Astrophysikalisches Institut Potsdam (AIP), An der Sternwarte 16, 14482 Potsdam, Germany
9 - X-ray & Observational Astronomy Group, Department of Physics and Astronomy, Leicester University, Leicester LE1 7RH, UK

Received 29 August 2007 / Accepted 1 October 2007

Abstract
Aims. We present the optical classification and redshift of 348 X-ray selected sources from the XMM-Newton Bright Serendipitous Survey (XBS), which contains a total of 400 objects (identification level = 87%). About 240 are new identifications. In particular, we discuss in detail the classification criteria adopted for the active galactic nuclei (AGNs) population.
Methods. By means of systematic spectroscopic campaigns using various telescopes and through the literature search, we have collected an optical spectrum for the large majority of the sources in the XBS survey and applied a well-defined classification "flow chart''.
Results. We find that the AGNs represent the most numerous population at the flux limit of the XBS survey ($\sim $10-13 erg cm-2 s-1) constituting 80% of the XBS sources selected in the 0.5-4.5 keV energy band and 95% of the "hard'' (4.5-7.5 keV) selected objects. Galactic sources populate the 0.5-4.5 keV sample significantly (17%) and only marginally (3%) the 4.5-7.5 keV sample. The remaining sources in both samples are clusters/groups of galaxies and normal galaxies (i.e. probably not powered by an AGN). Furthermore, the percentage of type 2 AGNs (i.e. optically absorbed AGNs with $A_{\rm V}>2$ mag) dramatically increases going from the 0.5-4.5 keV sample ( $f=N_{\rm AGN 2}/N_{\rm AGN}=7$%) to the 4.5-7.5 keV sample (f=32%). We finally propose two simple diagnostic plots that can be easily used to obtain the spectral classification for relatively low-redshift AGNs even if the quality of the spectrum is not good.

Key words: galaxies: active - galaxies: nuclei - quasars: emission lines - X-ray: galaxies - Surveys

1 Introduction

In the past few years, XMM-Newton and Chandra telescopes have been an excellent tool for surveying the hard X-ray sky at all fluxes, from relatively bright (10-13 erg cm-2 s-1, e.g. Della Ceca et al. 2004, and references therein), to medium (10-13 erg cm-2 s-1-10-14 erg cm-2 s-1, e.g. Barcons et al. 2007, and references therein) and deep (10-14-10-16 erg cm-2 s-1, Brandt & Hasinger 2005; Worsley et al. 2005, and references therein) fluxes. At the energies ($\sim $0.5-10 keV) covered by the instruments on board these two telescopes, active galactic nuclei (AGNs) can be efficiently selected and studied even when affected by high levels of absorption (up to $N_{\rm H}\sim 10^{24}$ cm-2, corresponding to an optical absorption of $A_{\rm V}\sim 500$ mag). This important characteristic, combined with the good (or excellent) spatial and energy resolutions of the detectors, makes the ongoing surveys a fundamental tool for AGN studies. At the same time, these new surveys represent an observational challenge at different wavelengths from the X-ray ones: multiwavelength follow-ups of X-ray sources, particularly in the optical domain, are decisive in deriving the distance and understanding the properties of the selected objects, but they also require large amounts of dedicated observing time at different telescopes. Probably the most challenging and time-consuming effort is the optical spectroscopic follow-up of the selected sources.

One of the primary goals of all these hard X-ray surveys is to explore the population of absorbed AGNs and, to this end, an optical classification that can reliably separate between optically absorbed and non-absorbed objects is always required. Two important limits, however, affect the spectroscopic follow-ups of deep and, in part, medium surveys: first, the optical counterparts are often too faint to be spectroscopically observed even at the largest optical telescopes currently available. Second, even when a spectrum can be obtained, its quality is not always good enough to provide the critical pieces of information that are required to assess a reliable optical classification. These two problems often limit the final scientific results that are based on the optical classification of medium/deep surveys.

In contrast, bright surveys offer the important possibility of obtaining a reliable optical classification for virtually all the selected sources (with some exceptions, as discussed in the next sections). The disadvantage of dealing with shallow/wide-angle samples is that the techniques for efficiently observing many sources at once, like multi-objects or fibers based methods, cannot be applied for the optical follow-up, given the low space density of sources at bright X-ray fluxes. The only suitable method, the "standard'' long-slit technique, requires many independent observing nights to achieve the completion of the optical follow-up.

In this paper we present and discuss in detail the optical classification process of the XMM-Newton Bright Serendipitous Survey (XBS, Della Ceca et al. 2004), which currently represents the widest (in terms of sky coverage) among the existing XMM-Newton or Chandra surveys for which a spectroscopic follow-up has almost been completed. The aim of the paper, in particular, is to provide not only a generic classification of the sources and their redshift but also a quantification, within the limits of the available data, of the corresponding threshold in terms of level of optical absorption.

The paper is organized as follows. In Sect. 2 we describe the XBS survey, in Sect. 3 we describe the process of identification of the optical counterpart, in Sects. 4 and 5 we respectively summarize our own spectroscopic campaigns carried out to collect the data and the data obtained from the literature. In Sect. 6 we briefly discuss the data reduction and analysis of the optical spectra and in Sect. 7 give the details on the classification criteria adopted for the sources in the XBS survey. In Sect. 8 we propose two diagnostic plots that can be used to easily classify the sources into type 1 and type 2 AGNs. The resulting catalogue is presented in Sect. 9, while in Sect. 10 we briefly discuss the optical breakdown and the redshift distribution of the sources. The conclusions are finally summarized in Sect. 11. Throughout this paper H0=65 km s-1 Mpc-1, $\Omega_{\Lambda}=0.7$, and $\Omega_{\rm M} = 0.3$ are assumed.

2 The XMM-Newton Bright Serendipitous Survey

The XMM-Newton Bright Serendipitous Survey (XBS survey, Della Ceca et al. 2004) is a wide-angle ($\sim $28 sq. deg), high Galactic latitude (|b|>20 $\deg$) survey based on the XMM-Newton archival data. It is composed of two samples that are both flux-limited ($\sim $ $7\times 10^{-14}$ erg cm-2 s-1) in two separate energy bands: the "soft'' 0.5-4.5 keV band (the BSS sample) and the hard 4.5-7.5 keV band (the HBSS sample). A total of 237 (211 for the HBSS sample) independent fields have been used to select 400 sources, 389 belonging to the BSS sample and 67 to the HBSS sample (56 sources are in common). The details on the fields selection strategy, the source selection criteria, and the general properties of the 400 objects are discussed in Della Ceca et al. (2004).

One of the main goals of the survey is to provide a well-defined and statistically complete census of the AGN population with particular attention to the problem of obscuration. To this end, the possibility of comparing X-ray and optical spectra of good quality for all the sources present in the two complete samples offers a unique and fundamental tool for statistically studying the effect of absorption in the AGN population in an unbiased way. Indeed, most of the X-ray sources of the XBS survey have been detected with enough counts to allow a reliable X-ray spectral analysis. At the same time, most of the sources have a relatively bright (R< 22 mag, see next section) optical counterpart and they can be spectroscopically characterized using a 4-m class telescope.

To date, the spectroscopic identification level has reached 87% (87% and 97% considering the BSS and the HBSS samples separately). The results of the spectroscopic campaigns are discussed in the following sections.

3 Identification of the optical counterpart

The identification of the optical counterparts of the XBS sources is relatively easy given the combination of the good positions of the XMM-Newton sources (90% error $\sim $4 $^{\prime\prime}$, Della Ceca et al. 2004) and the brightness of the sources: X-ray sources with $F_{\rm X}>10^{-13}$ erg cm-2 s-1 are expected to have an optical counterpart brighter than 22 mag for X-ray-to-optical flux ratios below 20 (i.e. for the majority of type 1 AGNs, galaxies and stars). Only the rare (but interesting) sources with extreme X-ray-to-optical flux ratios, like the distant type 2 QSOs (e.g. Severgnini et al. 2006), are expected to have magnitudes as faint as $R\sim 25$. For this reason, for the large majority of the XBS sources we have been able to unambiguously pinpoint the optical counterpart using the existing optical surveys (i.e. the DSS I/II[*] and the SDSS[*]). In particular, we have found the optical counterpart for about 88% of the XBS sources on the DSS with a red magnitude (the APM[*] red magnitude) brighter than $\sim $20.5. All but 6 of the remaining sources have been optically identified either through dedicated photometry or using the SDSS catalogue. The red magnitudes of these sources are relatively bright (R between 20.5 and 22.5) with one exception: an R=24.5 object (XBSJ021642.3-043553), which turned out to be a distant (z=1.985) type 2 QSO (Severgnini et al. 2006). For 6 objects we have not yet found the optical counterpart, but only for two of these we have relatively deep images that have produced a faint lower limit on the R magnitude (R>22.8 and R>22.2). For the other 4 sources we only have the upper limit based on the DSS plates.


  \begin{figure} \par\includegraphics[width=8cm,clip]{8568fig1.eps} \end{figure} Figure 1: Magnitude distribution of the XBS optical counterparts. Most (94%) of the magnitudes are in a red filter. Shaded histogram represents the sources without a spectral classification.
Open with DEXTER


  \begin{figure} \par\includegraphics[width=8cm,clip]{8568fig2.eps} \end{figure} Figure 2: X-ray/optical positional offsets of the 348 XBS sources with a spectral classification. Open circles are stars and clusters of galaxies, while filled points are the remaining sources (AGNs and "elusive AGN candidates''). The two circles represent the regions including 68% and 90% of the points.
Open with DEXTER


  \begin{figure} \par\includegraphics[width=8cm,clip]{8568fig3.eps} \end{figure} Figure 3: X-ray/optical positional offsets of the XBS sources in common with the 2XMM catalogue (343 in total). In this case the X-ray positions are taken from the 2XMM catalogue that has been produced with recent versions of the Standard Analysis System. Symbols and regions as in Fig. 2.
Open with DEXTER

In conclusion, we found the most-likely optical counterparts for the large majority of the 400 XBS sources (all but 6 sources). The magnitude distribution of the counterparts is presented in Fig. 1. Since we did not carry out a systematic photometric follow-up of the XBS objects, we do not have a homogeneous set of magnitudes in a well-defined filter. In Fig. 1 we report the magnitudes either from existing catalogues (e.g. APM, SDSS, NED[*], Simbad[*]) or from our own observations. Most of them (94%) are in a red filter, while the remaining 6% (all bright stars with mag < 13) are in V or B filters.

In Fig. 2 we show the X-ray/optical positional offsets of the 348 XBS sources discussed in this paper (i.e. those with a spectral classification). All the identifications have offsets below $\sim $7 $\hbox{$^{\prime\prime}$ }$, with the majority ($\sim $90%) of sources having an offset below 3.8 $\hbox{$^{\prime\prime}$ }$. In Fig. 2 we have distinguished the objects spectroscopically classified as stars and clusters of galaxies from the rest of the sources since both stars and clusters may suffer from larger positional offsets due to the presence of proper motions (stars) or, in the case of clusters of galaxies, due to the intrinsic offset between the X-ray source (the intracluster gas) and the optical object (e.g. the cD galaxy). Indeed, the circle including 90% of stars and clusters is larger ($\sim $4.5 $\hbox{$^{\prime\prime}$ }$) than the circle computed using all the sources.

In the last few years, the XMM-Newton images have been reprocessed with improved versions of the SAS and the astrometry has been refined and corrected. We thus recomputed the X-ray-optical offsets using the improved X-ray positions included in the preliminary version of the second XMM-Newton Serendipitous EPIC Source Catalogue (2XMM, Watson et al. 2007, in preparation, see also http://xmm.vilspa.esa.es/xsa/). In Fig. 3 we plot these newly computed offsets for the objects that are in common with the 2XMM catalogue. The improvement is evident, with 90% of the sources (excluding stars and clusters) having an offset below 2.1 $\hbox{$^{\prime\prime}$ }$. The sources with relatively large offsets (4-5 $\hbox{$^{\prime\prime}$ }$) are mostly stars and clusters. All but 2 extragalactic "non-clusters'' objects have X-ray-to-optical offsets below 4 $^{\prime\prime}$. By inspecting the X-ray images of the two extragalactic "non-clusters'' objects with large offsets (XBSJ095054.5+393924, a type 1 QSO at z=1.299 and XBSJ225020.2-642900, a type 1 QSO at z=1.25), we found strong indications that both objects are the result of a source blending that has "moved'' the centroid of the X-ray position between two nearby objects. Interestingly, in one of these cases (XBSJ225020.2-642900), we also spectroscopically observed the second (and fainter) nearby object and found a very similar spectrum of type 1 QSO at the same redshift (1.25). This could either be a real QSO pair or, alternatively, the result of gravitational lensing caused by a (not visible) galaxy.

In conclusion, excluding these two objects for which the X-ray position is not accurate, all the XBS sources classified as extragalactic objects have an optical counterpart within 4 $^{\prime\prime}$ using the improved X-ray positions and 90% have offsets within 2.1 $^{\prime\prime}$.

3.1 Estimate of the number of spurious X-ray/optical associations

As discussed above, the optical counterparts found for the XBS sources have R magnitudes brighter than 22.5 (except for one object) with a large fraction (88%) of them having magnitudes brighter than 20.5 (i.e. they are visible on the DSS plates). Given the density of AGNs at the magnitude limit of R=22.5 (e.g. Wolf et al. 2003), the probabilty of finding an AGN by chance within 4 $\hbox{$^{\prime\prime}$ }$ from an unrelated X-ray source is $\sim $ $5\times10^{-4}$, which translates into an expected number of $\sim $0.2 spurious AGN identifications in the entire XBS survey. Therefore it is reasonable to consider all the objects optically classified as emission-line AGNs (or BL Lac objects) as the correct counterparts of the X-ray sources.

Table 1: Journal of observations.

Stars and galaxies, instead, may contaminate the identification process, given their higher sky density. In principle, a fraction of sources identified as stars or "normal'' galaxies (or elusive AGNs, see discussion in Sect. 7.5) could be spurious counterparts. Considering the density of stars and galaxies at the faintest magnitudes observed in the two classes of sources ( $R_{\rm stars}\leq 18$ and $R_{\rm galaxies}\leq 21$), we expected about 12 stars and 4 galaxies falling by chance within 4 $\hbox{$^{\prime\prime}$ }$ from the 400 X-ray positions. This is clearly an upper limit given the adopted identification process: we usually observed all the bright (i.e. visible on the DSS) objects falling within the circle of 4 $\hbox{$^{\prime\prime}$ }$ radius and, whenever an AGN is found, we considered it as the right counterpart and discarted the others (either stars or galaxies. As described above the probability of finding an AGN by chance is very low in our survey). This strategy excludes the large majority of possible spurious galaxy or star identifications: only those stars or galaxies falling by chance close to an X-ray source whose real counterpart is weak (e.g. weaker than the DSS limit) have the possibility of being considered the counterpart by mistake. Since the majority ($\sim $90%) of the real counterparts are expected to be brighter than the DSS limit, we conclude that only $\sim $1/10 of the 12 stars and 4 galaxies falling by chance in the error circle have the possibility of being considered as the counterpart. Therefore, the actual number of spurious stars and galaxies in the sample should be $\sim $1.2 and $\sim $0.4, respectively. In conclusion, we do not expect more than 1-2 misidentifications in the entire XBS survey.

4 Optical spectroscopy

About 2/3 of the spectroscopic identifications (i.e. $\sim $240 objects) of the XBS survey come from dedicated spectroscopy carried out during 5 years (from 2001 to 2006) at several optical telescopes. Most of the identifications were obtained at the Italian Telescopio Nazionale Galileo (TNG, 51% of the identifications) and at the ESO 3.6m and NTT telescopes (37%). The remaining 12% was collected from other telescopes like the 88 $^{\prime\prime}$ telescope of the University of Hawaii (UH) in Mauna Kea and the Calar Alto 2.2 m telescope.

The instrumental configurations are summarized in Table 1. We always adopted a long-slit configuration with low/medium dispersion (from 1.4 Å/pixel to 3.7 Å/pixel) and low/medium resolution (from $\sim $250 to 450) grisms to maximize the wavelength coverage. For the data reduction we have used the IRAF longslit package. The spectra were wavelength-calibrated using a reference spectrum and flux-calibrated using photometric standard stars observed the same night. Most of the observations were carried out in non-photometric conditions. Since the main goal of the observations was to secure a redshift and a spectral classification of the source we did not attempt to obtain an absolute flux calibration of the spectra.

In general, we have two exposures for each object, except for a few cases in which we have only one spectrum or three exposures. Cosmic rays were subtracted manually from the extracted spectrum or automatically if three exposures of equal length are available.

On average, the seeing during the observing runs ranged from 1 $^{\prime\prime}$ to 2 $^{\prime\prime}$ with a few exceptional cases of seeing below 1 $^{\prime\prime}$ (0.5 $^{\prime\prime}$-0.8 $^{\prime\prime}$, typically during the runs at the ESO NTT). Usually, during very bad seeing conditions ($\geq$2.5 $^{\prime\prime}$), no observations were carried out. We have used a slit width of 1.2 $^{\prime\prime}$-1.5 $^{\prime\prime}$ except for the periods of sub-arcsec seeing conditions, where a slit width of 1 $^{\prime\prime}$ was used to maximize the signal.

5 Data from the literature

The remaining 1/3 of the spectroscopic identifications of the XBS survey were taken from the literature (NED and SIMBAD[*]) or from other XMM-Newton identification programs like AXIS (Barcons et al. 2007). Whenever possible we obtained the optical spectrum of the extragalactic sources, either in FITS format or a printed spectrum, and then analyzed it using the same criteria adopted for the spectra collected during our own observing runs. In a few cases we did not find a spectrum but instead tables presenting the relevant pieces of information on the emission lines. Therefore the spectral analysis (for classification purpose) was possible for nearly all the extragalactic identifications coming from the literature or from the AXIS program.

If a classification is present in the literature but no further information is found we have kept the classification only if it can be considered unambiguous (e.g. a type 1 QSO, see discussion in Sect. 7).

6 Spectral analysis

For more than 80% of the extragalactic identifications (either from our own spectroscopy or from the literature), we have an optical spectrum in electronic format. We used the task "splot'' within the irafpackage to analyze these spectra and get the basic pieces of information, like the line positions, equivalent widths (EW), and FWHM. During the fit we used a Gaussian or a Lorentzian profile. When two components are clearly present in the line profile (e.g. a narrow core plus a broad wing), we attempted a de-blending.

Given the moderate resolution of the spectroscopic observations ( ${\it FWHM}\sim 650$-1200 km s-1), we applied a correction to the line widths to account for the instrumental broadening, i.e.:

$\Delta\lambda = \sqrt{\Delta\lambda_{\rm o}^2 - \Delta\lambda_{\rm inst}^2}$
where $\Delta\lambda$, $\Delta\lambda_{\rm o}$, and $\Delta\lambda_{\rm inst}$ are the intrinsic, the observed and the instrumental line width respectively.

The errors on EW and FWHM were estimated with the task "splot''. This task adopts a model for the pixel sigmas based on a Poisson statistics model of the data. The model parameters are a constant Gaussian sigma and an "inverse gain''. We set this last parameter to "0''; i.e., we assume that the noise due to instrumental effects (RON) is negligible. This is reasonable for our spectra. The de-blending and profile-fit error estimates are computed by Monte-Carlo simulation (see iraf help for details). We found that the errors computed in this way are sometimes underestimated, in particular when the background around the emission/absorption line is not well-determined and/or when the adopted model profile (Gaussian or Lorentz profile) does not correctly describe the line. In these cases we have adopted a larger error that includes the values obtained with different background/line profile models.

For all the identifications for which only a printed spectrum is available, we performed a similar (but rougher) analysis and included the larger uncertainties in the error bars.

7 Spectroscopic classification and redshift

On the basis of the data collected from the literature and the spectra obtained from our own spectroscopy, we have determined a spectroscopic classification and a redshift for 87% (348) of the XBS objects. The sources can be broadly grouped into stars, clusters of galaxies, and AGNs/galaxies. Stars and AGNs/galaxies represent the most numerous populations in the sample, being 17% and 80%, respectively, of the total number of the identified XBS sources. An extended analysis of the X-ray and optical properties of the 58 stars found in the sample has been already presented in López-Santiago et al. (2007) and will not be discussed in this paper anymore.

The classification of an XBS source as a cluster of galaxies is essentially based on the visual detection of an overdensity of sources in the proximity of the X-ray position on the optical image and on the spectroscopic confirmation that some of these objects have the same redshift. In all these cases, the object closer to the X-ray position is an optically "dull'' elliptical galaxy. The cluster nature of the XBS sources is usually confirmed by a visual inspection of the X-ray image, which shows that the X-ray source is extended. In the XBS survey, we currently have only 8 objects classified as clusters of galaxies. However, this type of objects is certainly under-represented because the source detection algorithm is optimized for point-like sources (see Della Ceca et al. 2004). This is also true for normal galaxies whose X-ray emission (due to diffuse gas and/or discrete sources) is extended.

In this paper we do not discuss stars and clusters of galaxies, and they will be excluded from the following analysis. In this section we present the criteria adopted to classify the remaining extragalactic sourcesin detail, i.e. AGNs (including BL Lac objects) and galaxies.


  \begin{figure} \par\includegraphics[width=15cm,height=15cm]{8568fig4.eps} \end{figure} Figure 4: Classification flow chart of the XBS extragalactic sources (excluding clusters of galaxies). Numbers near the arrows indicate the number of XBS sources that have followed the corresponding path. The sources within the "optically elusive AGN candidates'' group have been classified on the basis of the X-ray spectrum and re-distributed into the other classes accordingly (see Caccianiga et al. 2007).
Open with DEXTER

7.1 The classification scheme

The large majority ($\sim $90%) of the extragalactic sources in the XBS survey show strong ( ${\it EW} >10$ Å) emission lines in the optical spectrum. In most of these objects the analysis of the emission lines gives a clear indication of the presence of an AGN.

One of the primary goals of the XBS survey is to explore the population of absorbed AGNs. For this reason, we want to adopt an optical classification that can reliably separate optically absorbed from non-absorbed objects. The criterion typically used to separate optically absorbed and non-absorbed AGNs is based on the width of the permitted/semi-forbidden emission lines, when present. However, different thresholds have been used in the literature to distinguish type 1 (i.e. AGNs with broad permitted or semi-forbidden emission lines) and type 2 AGNs (i.e. with narrow permitted/semi-forbidden emission lines) ranging from 1000 km s-1 (e.g. Stocke et al. 1991 for the Extended Medium Sensitivity Survey, EMSS) up to 2000 km s-1 (e.g. Fiore et al. 2003, for the Hellas2XMM survey). Both thresholds present some limits.

On the one hand, the 2000 km s-1 threshold may misclassify the Narrow Line Seyfert 1 (NLSy1s) and their high-zcounterparts, the "Narrow Line QSO'' (NLQSO, see for instance Baldwin et al. 1988), as type 2 AGNs. These sources typically show permitted/semi-forbidden lines of width between 1000 and 2000 km s-1 (or even lower, see for instance Véron-Cetty et al. 2001), but it is generally accepted that the relatively narrow permitted/semi-forbidden lines are not due to the presence of strong optical absorption but, rather, they are connected to the physical conditions of the nucleus (e.g. Ryan et al. 2007 and references therein).

On the other hand, the adoption of a lower threshold (e.g. 1000 km s-1) to distinguish type 1 and type 2 AGNs can systematically misclassify high-z QSO 2, where the observed permitted lines are typically between 1000 and 2000 km s-1 (e.g. Stern et al. 2002; Norman et al. 2002; Severgnini et al. 2006). It is thus clear that a simple classification based exclusively on the widths of the permitted lines cannot be realistically adopted. Additional diagnostics are necessary for a reliable optical classification.

In Fig. 4 we present the flow chart that summarizes the classification criteria used for the XBS extragalactic sources (excluding the clusters of galaxies). The complexity of the presented flow chart is mainly due dealing with sources distributed in a wide range of redshift (from local objects up to $z\sim 2$): the emission lines that can be used for the spectral classification are thus different depending on the redshift of the source. Another source of complexity is the problem of optical "dilution'' due to the host-galaxy light (see below).

The final classes (represented by 4 boxes) are type 1 AGNs, type 2 AGNs, BL Lac objects and the "normal'' (i.e. not powered by an AGN) galaxies. In 35 cases the optical spectrum is dominated by the starlight from the host galaxy and establishing the presence of an AGN and its type (e.g. type 1 or type 2) through the optical spectrum is not possible. For this group of objects, named "optically elusive AGN'' candidates, we used the X-ray data to assess the presence of an AGN and to characterize its nature (i.e. absorbed or unabsorbed, see Caccianiga et al. 2007, and Sect. 7.5 for details).

We considered the intermediate types 1.2, and 1.5 in the type 1 AGN class, while the type 2 AGN class includes the 1.8 and 1.9 types. This distinction is expected to correspond to a separation into a level of absorption lower/higher than $A_{\rm V}\sim 2$ mag (see discussion below), i.e. a column density ($N_{\rm H}$) higher/lower than $\sim $ $4\times 10^{21}$ cm-2 assuming a Galactic standard $N_{\rm H}$/$A_{\rm V}$ conversion.

We applied these steps to the 275 objects for which the required information is available (either from our own spectroscopy or from the literature). Besides these 275, we have 7 additional objects whose classification has been taken from the literature, but it is not possible to directly apply the classification criteria discussed here since a spectrum or a table reporting the lines properties is not available. These 7 objects are all classified as type 1 AGNs with redshift between 0.64 and 1.4 and X-ray luminosities between 1044 and 1046 erg s-1 (i.e. they are type 1 QSO). We adopted the published classification for these objects even if they had not passed through the classification steps presented in Fig. 4, which we discuss briefly here.

7.2 AGNs with broad (FWHM > 2000 km s-1) permitted emission lines

The first main "arrow'' of Fig. 4 considers the detection of one (or more) very broad ( ${\it FWHM}> 2000$ km s-1) permitted emission line. In this step we did not consider the H$\alpha $ line. The reason is that, whenever only a strong and broad H$\alpha $ emission line is detected in the optical spectrum, it is not possible to correctly classify the object. Indeed, sources where only a broad H$\alpha $ line is clearly detected can be both unabsorbed AGNs or intermediate AGNs, like Sy1.8 or Sy1.9. Since, as discussed above, we consider Sy1.8 and Sy1.9 as type 2 AGNs, we are not able to correctly classify these sources as type 1 or type 2 just on the basis of the H$\alpha $ line.

This first step allows us to directly classify as type 1 AGNs all the sources with very broad ( ${\it FWHM}> 2000$ km s-1) permitted/semi-forbidden emission lines. These sources are "classical'' type 1 AGNs (Sy1 and QSO).

7.3 Objects with permitted emission lines with FWHM < 2000 km s-1

The second main "arrow'' regards sources for which "narrow'' ( ${\it FWHM} <2000$ km s-1) permitted emission lines (H$\alpha $ excluded) are detected. As discussed above, in this group many different types of sources can be found, including absorbed AGNs, AGNs with intrinsically narrow permitted/semi-forbidden emission lines (NLSy1 and NLQSO), and emission-line galaxies like starburst/HII-region galaxies. As already noted, a proper classification of these objects requires the application of diagnostic criteria. For the sources at relatively low z (below $\sim $0.65) the detection of two critical emission lines, i.e. the H$\beta $ and the [OIII]$\lambda $5007 Å, can help the classification significantly. We thus discuss the sources separately according to whether the H$\beta $/[OIII]$\lambda $5007 Å spectral region is covered (i.e. sources with z below $\sim $0.65) or not (i.e. sources with z larger than $\sim $0.65).

H$\beta $ and [OIII]$\lambda $5007 Å covered

In all but 3 objects with strong and relatively narrow ( ${\it FWHM} <2000$ km s-1) permitted/semi-forbidden emission lines the H$\beta $ and [OIII]$\lambda $5007 Å spectral region is covered. As discussed by several authors (e.g. Véron-Cetty & Véron 2003; Winkler 1992; Whittle 1992), a clear distinction between different types of AGNs can be based on the ratio between [OIII]$\lambda $5007 Å and H$\beta $ line intensity. Optically-absorbed Seyferts, like Seyfert 2 or Seyfert1.8/1.9, present high values of the [OIII]$\lambda $5007 Å/H$\beta $ flux ratios (>3), while moderately absorbed or non-absorbed Seyferts (Seyfert 1.5, Seyfert 1.2, and Seyfert 1 and NLSy1) show a [OIII]$\lambda $5007 Å/H$\beta $ flux ratio between 0.2 and 3. In Fig. 5 we show the [OIII]$\lambda $5007 Å/H$\beta $ flux ratio versus the H$\beta $ width for all the XBS sources where these lines are observed (including sources with ${\it FWHM}> 2000$ km s-1 emission lines). The two quantities are strongly coupled, and the objects with broad (>1000 km s-1) H$\beta $ have all (but one) [OIII]$\lambda $5007 Å/H$\beta $ flux ratio below 3. We classify all these objects as type 1 AGNs, including the source where the [OIII]$\lambda $5007 Å/H$\beta $ flux ratio is marginally greater than 3, since the value is consistent, within the errors, with those observed in type 1 AGNs.


  \begin{figure} \par\includegraphics[width=9cm]{8568fig5.eps} \end{figure} Figure 5: [OIII]$\lambda $5007 Å/H$\beta $ flux ratio versus the H$\beta $ emission line width for the sources in the XBS survey for which these lines have been detected (z below $\sim $0.65). Filled points are type 1 AGNs, open circles are type 2 AGNs, stars are objects spectroscopically classified as emission line galaxies (HII-region/starburst galaxies). The two dashed lines indicate the reference values used for the spectroscopic classification (see text and Fig. 4 for details).
Open with DEXTER

In contrast, the objects with a narrow (<1000 km s-1) H$\beta $ present a wide range of [OIII]$\lambda $5007 Å/H$\beta $ flux ratios, from 0.3 to 15. This class of sources includes both type 2 AGNs, "normal'' galaxies (e.g. HII-region galaxies or starburst galaxies) and some NLSy1s. To distinguish all these cases it is necessary to apply either the diagnostic criteria discussed, e.g. in Veilleux & Osterbrock (1987), to separate type 2 AGNs from HII-region/starburst galaxies, and/or the diagnostics based, for instance, on the FeII$\lambda $4570 Å/H$\beta $ flux ratio to recognize the NLSy1 (Véron-Cetty et al. 2001). The adopted criteria are indicated near the corresponding arrows of Fig. 4.

H$\beta $ and/or [OIII]$\lambda $5007 Å not covered

Only for 3 sources with strong and relatively narrow ( ${\it FWHM} <2000$ km s-1) permitted emission lines, the H$\beta $/[OIII]$\lambda $5007 Å spectral range is not covered. As discussed above, these sources can be optically-absorbed AGNs (i.e. type 2 AGNs) or NLQSOs. The distinction between these two classes at high redshift is more critical than at lower z and other diagnostics must be used, like the intensity of the FeII$\lambda $4570 Å hump or the relative strength of the HeII emission line when compared to the CIV$\lambda $1549 Å (e.g. Heckman et al. 1995). One of these objects (XBSJ021642.3-043553, z=1.985) has been extensively discussed in Severgnini et al. (2006) and it is classified as type 2 QSO on the basis of the relative strength of the HeII emission line when compared to the CIV$\lambda $1549 Å.

The second source (XBSJ120359.1+443715, z=0.541) has a blue spectrum and quite a strong Fe II4570 Å hump that is usually considered as the signature of a NLSy1. Unfortunately we cannot quantify the strength of this hump further with respect to the H$\beta $ line since this line falls outside the observed spectrum. We classify this object as an NLQSO candidate.

Finally, in the third object (XBSJ124214.1-112512, z=0.82), we detected the MgII$\lambda $2798 Å emission line with a relatively narrow (FWHM $\sim $ 1900 km s-1) core plus a broad wing. Both the FeII$\lambda $4570 Å and the HeII lines fall outside the observed range, so we cannot apply the diagnostic criteria discussed above. Using the spectral model described in Sect. 8 we have successfully fitted the observed continuum emission using a value of $A_{\rm V}\sim 0.5$ mag i.e. below the 2 mag limit that corresponds to our classification criteria (see Sect. 8). We thus classify this object as a type 1 AGN.

7.4 Sources with weak (or absent) permitted emission lines

The last main "arrow'' of Fig. 4 corresponds to sources with no (or weak) permitted emission lines (excluding the H$\alpha $ line, as discussed above). This group of sources includes both "featureless'' AGNs (the BL Lac objects) and sources whose optical spectrum is dominated by the host-galaxy, so no evidence (or little evidence) for the presence of an AGN can be inferred from the optical spectrum. As already discussed, these objects are considered as "elusive'' AGN candidates and analyzed separately using the X-ray information (see next section).

BL Lac objects are classified on the basis of the lack of any (including the H$\alpha $) emission line and the shape of the continuum around the 4000 Å break ($\Delta $[*]). In fact detection of a significant reduction in the 4000 Å break when compared with elliptical galaxies is considered as indicating the presence of nuclear emission. We adopt the limit commonly used in the literature of $\Delta<40$% to classify the source (with no-emission lines) as a BL Lac object (e.g. see the discussion in Landt et al. 2002).

There are 5 BL Lacs in total, all of which have been detected as radio sources in the NVSS (Condon et al. 1998) radio survey, something that is considered as further confirmation of the correct classification. The properties of the XBS BL Lacs are presented in Galbiati et al. (2005). As discussed in Caccianiga et al. (2007), we cannot exclude the possibilty that some of the "elusive'' AGNs are actually hiding a BL Lac nucleus. The best way to find them is through a deep radio follow-up. On the basis of the current best estimate of the BL Lac sky density, however, we do not expect more than 1-2 BL Lacs hidden among the XBS-elusive AGNs.

7.5 The optically "elusive'' AGN candidates

As summarized in Fig. 4, different classification paths lead to the group of optically "elusive'' AGN candidates. All these sources (35 in total) are characterized by the presence, in the optical spectrum, of a significant/dominant contamination of star-light from the host galaxy. In some cases, i.e. for the so-called X-ray bright optically normal galaxies (XBONG) and the HII-region/starburst galaxies, we do not have clear (optical) evidence of an AGN. We also consider as optically "elusive'' AGN candidates those sources where a broad (>1000-2000 km s-1) H$\alpha $ line is probably present but where most of the remaining emission lines (in particular the H$\beta $ emission line) are not detected. Even if the presence of an AGN in these sources is somehow suggested by the detection of a broad H$\alpha $ emission line, the "dilution'' due to the host galaxy is critical also in these cases, because it does not permit quantification of the optical absorption (i.e. type 1 or type 2 AGNs). Similarly, some other sources in this group show quite strong [OIII]$\lambda $5007 Å, which can be suggestive of an AGN, but no H$\beta $ is detected, something that prevents us from a firm classification of the source.

Given the objective difficulty of using the optical spectra to assess the actual presence of an AGN and to give a correct classification of it (i.e. type 1, type 2 or BL Lac object), we analyzed the X-ray data. In particular, we have shown that the X-ray spectral shape combined with the X-ray luminosity of the sources allows us to assess the presence of an AGN and to quantify its properties. While the detailed discussion of this analysis is found in Caccianiga et al. (2007), we summarize here the main conclusions. In the large majority of cases (33 out of 35 objects), the X-ray analysis revealed an AGN, while the X-ray emission is probably due to the galaxy only in 2 cases (either due to hot gas or to discrete sources) given the low X-ray luminosities (1039-1040 erg s-1). In 20 sources where an AGN has been detected the column densities observed are below $N_{\rm H}=4\times 10^{21}$ cm-2, while the values are higher in 12. The data do not allow an estimate of the column density only for one object. According to the Galactic relationship between optical ($A_{\rm V}$) and X-ray absorption ($N_{\rm H}$), the value of $N_{\rm H}=4\times 10^{21}$ cm-2 corresponds to $A_{\rm V}\sim 2$ mag, which is the expected dividing line between type 1 and type 2 sources as defined in this paper, i.e. following the scheme of Fig. 4 (see the discussion in Sect. 8). We thus classified these 32 "elusive'' AGNs into type 1 and type 2 according to the value of $N_{\rm H}$ measured from the X-ray analysis. In Table 3, these classifications are flagged to indicate that they are not based on the optical spectra.

8 Diagnostic plots

Using a simple spectral model, discussed in Severgnini et al. (2003), we produced some diagnostic plots that may help in classifying X-ray selected sources. This model uses an AGN template composed of two parts: a) the continuum with the broad emission lines and b) the narrow emission lines. According to the basic version of the AGN-unified model, the first part can be absorbed, while the second one is not affected by the presence of an obscuring medium. The AGN template is based on the data taken from Francis et al. (1991) and Elvis et al. (1994), while the extinction curve is taken from Cardelli et al. (1989). Besides the AGN template, the spectral model also includes a galaxy template, produced on the basis of the Bruzual & Charlot (2003) models. We then applied different levels of $A_{\rm V}$ and measured the expected values of some critical quantities such as the 4000 Å break, the [OIII]$\lambda $5007 Å, and the H$\alpha $ line equivalent width.

8.1 Non-elusive AGNs

In Fig. 6 we have plotted the 4000 Å break versus the [OIII]$\lambda $5007 Å equivalent width for all the XBS sources classified as type 1 o type 2 AGNs, for which these quantities have been computed, excluding the elusive AGNs. On this plot, type 2 and type 1 AGNs occupy separated regions, with type 2 AGNs showing the largest [OIII]$\lambda $5007 Å equivalent widths and largest 4000 Å breaks. This separation is expected since the presence of a high level of absorption in these sources significantly suppresses the AGN continuum, on the one hand, and increases the narrow lines equivalent widths, on the other. In the same figure we then plotted the curves based on the spectral model described above for three different values of absorption, from $A_{\rm V}=1$ mag to 3 mag, assuming a 10 Gyr old early-type host galaxy. The $A_{\rm V}=2$ mag curve is clearly the one that separates the two classes of AGNs better. This result does not depend significantly on the host-galaxy type as shown in Fig. 7, where a much younger host-galaxy is assumed (1 Gyr). Also in this plot, the line that separates type 1 and type 2 AGNs better is the one corresponding to $A_{\rm V}=2$ mag. This weak dependence on the host-galaxy type is no longer true if we consider the elusive AGNs i.e. those sources whose optical spectrum is dominated by the host-galaxy and which occupy the upper-left region of the diagram. Therefore, this plot cannot be used as a diagnostic for the elusive AGNs.

The clear separation between type 1 and type 2 AGNs observed in an [OIII]$\lambda $5007 Å/4000 Å plot can be used as a simple diagnostic, at least for objects not dominated by the host-galaxy light. In Fig. 8 we report the typical regions occupied by type 1 and type 2 AGNs and (most of) the elusive AGNs. This diagnostic diagram is simple to apply, requiring just the measure of the fluxes across the 4000 Å break and the [OIII]$\lambda $5007 Å equivalent width, and can be used up to $z\sim 0.8$ (or higher if infrared spectra are available).


  \begin{figure} \par\includegraphics[width=8cm,clip]{8568fig6.eps} \end{figure} Figure 6: 4000 Å break ($\Delta $) versus the [OIII]$\lambda $5007 Å equivalent widths for the BSS type 1 (filled points) and type 2 (open circles) AGNs, excluding the elusive ones. The three lines show the expected regions corresponding to different optical absorptions, from $A_{\rm V}=1$ mag to $A_{\rm V}=3$ mag, assuming a 10 Gyr early-type host galaxy.
Open with DEXTER


  \begin{figure} \par\includegraphics[width=8cm,clip]{8568fig7.eps} \end{figure} Figure 7: Same plot as in Fig. 6 but assuming a younger (1 Gyr) host-galaxy. Symbols as in Fig. 6.
Open with DEXTER


  \begin{figure} \par\includegraphics[width=8cm,clip]{8568fig8.eps} \end{figure} Figure 8: Typical regions occupied by the XBS AGNs on the 4000 Å break/[OIII]$\lambda $5007 Å EW plot.
Open with DEXTER

8.2 Elusive AGNs with a broad H$\alpha $ emission line

By definition, elusive AGNs have an optical spectrum that is dominated by the host galaxy light, making it difficult/impossible to obtain a clear classification directly from the optical data. However, as already mentioned, a possibly broad H$\alpha $ line in emission is found in a number of elusive AGNs. In itself, this piece of information cannot give a clear indication of the type of AGN present in the source. With the support of the spectral model previously discussed, we now want to find a method of estimating the level of optical absorption in these sources. We want to use only the few AGN emission lines that usually can be detected even in the presence of a high level of dilution, i.e. the [OIII]$\lambda $5007 Å and the H$\alpha $ emission lines.


  \begin{figure} \par\includegraphics[width=8cm,clip]{8568fig9.eps} \end{figure} Figure 9: [OIII]$\lambda $5007 Å versus the H$\alpha $+[NII] line blend equivalent widths of the XBS AGNs classified as type 1 and type 2 on the basis of the optical spectrum (panel a)) and of the elusive AGNs for which a broad H$\alpha $ emission line has been detected (panel b)). In this case the classification is based on the X-ray spectrum. Open circles are type 2 AGNs, while filled points are type 1 AGNs. Solid lines show the theoretical separation between objects with large ( $A_{\rm V}>2$ mag) and small ( $A_{\rm V}>2$ mag) optical absorption corresponding to a threshold of $N_{\rm H}=4\times 10^{21}$ cm-2 assuming a Galactic standard relation. The two lines correspond to different ages of the host-galaxy (t=1 Gyr and 10 Gyr).
Open with DEXTER

Interestingly, the combination of the H$\alpha $ line intensity with the [OIII]$\lambda $5007 Å emission line can help to classify the source. In Fig. 9 we show the [OIII]$\lambda $5007 Å versus the H$\alpha $ + [NII] blend[*] equivalent widths of the XBS AGNs classified as type 1 and type 2 on the basis of the optical spectrum (panel a). In panel (b) we report the 9 elusive AGNs with a broad (FWHM > 1000 km s-1) H$\alpha $ emission line. In this case, the symbols represent a classification based on the X-ray spectral analysis. In the two panels we also report the theoretical lines that separate between AGNs with large ( $A_{\rm V}>2$ mag) and small ( $A_{\rm V}<2$ mag) optical absorption (corresponding to $N_{\rm H}$ larger or lower than $4\times 10^{21}$ cm-2 assuming a Galactic standard relation). Each point on these lines corresponds to a different AGN-to-galaxy luminosity ratio (that increases from left to right).

In Fig. 9b we do not include the sources classified as starburst or HII-region galaxies on the basis of the diagnostic diagrams because the H$\alpha $ line is likely to be produced within the host galaxy rather than by the AGN. We also exclude the sources with a narrow H$\alpha $ emission line to avoid sources whose H$\alpha $ line is contaminated by the emission from the host galaxy. The solid line nicely separates the elusive objects affected by large absorption (> $4\times 10^{21}$ cm-2) from those with low absorption (< $4\times 10^{21}$ cm-2). More important, this separating line is fairly independent of the host-galaxy type even when the host-galaxy light dominates the total spectrum (unlike the $\Delta $/[OIII]$\lambda $5007 Å plot). Therefore, Fig. 9 can be used as diagnostic tool to separate type 1 and type 2 AGNs, as defined in the XBS sample, when dilution from the host galaxy does not allow applying the usual diagnostic criteria and when X-ray data are not available.

9 The catalog

The result of the spectral classification of the XBS sources is summarized in Table 2 while in Table 3 we report the relevant optical information for each object. We note that the classification of the XBS sources was presented in part in Della Ceca et al. (2004). The classification presented in that paper has been revised and refined to take the complexity of some spectra into account (like the presence of a significant star-light contribution) and, therefore, some of the published classifications (20 in total) have now changed. Most (14 out of 20) of the sources with a different classification from that presented in Della Ceca et al. (2004) are optically-elusive AGNs or "normal galaxies'', so the new classification is based on the X-ray spectrum. In Table 3 we have flagged the sources for which the classification presented here differs from the one published in Della Ceca et al. (2004).

Table 2: Breakdown of the optical classification.

In Table 3 we also list an optical magnitude for each optical counterpart. As already discussed, we did not carry out a systematic photometric follow-up of the XBS sources, so the magnitudes are not homogeneous since taken from different catalogues or observations. For about half of the objects (172 objects), we collected a red (R or r) magnitude either from our own observations or from existing catalogues (mostly the SDSS catalogue). Some of the R magnitudes derived from our own observations were computed from the optical spectrum. Another substantial fraction of magnitudes (150) are taken from the APM facility (we used the red APM filter). For bright (and extended) objects, the APM magnitude is known to suffer from a large systematic error. In these cases we applied the correction described in Marchã et al. (2001) to compensate for this systematic error. Finally, for 26 objects classified as stars, we give the magnitude V or B present in Simbad.

10 The classification breakdown

Table 2 reports the current classification breakdown of the sources in the BSS and HBSS samples. Given the high identification level (87% and 97% for the BSS and the HBSS samples, respectively), the numbers in Table 2 should reflect the true relative compositions of the two samples. The first obvious consideration is that the percentage of stars decreases dramatically from the BSS sample (17%) to the HBSS sample (3%). Similarly, the relative fraction of type 2/type 1 AGNs is significantly different in the 2 samples, of a factor 6 higher in the HBSS (0.48) than in the BSS (0.08). As expected, the 4.5-7.5 keV energy band is much more efficient in selecting type 2 AGNs (efficiency $\sim $29%) when compared to the softer 0.5-4.5 keV band (efficiency $\sim $ 6%). It must be noted, however, that the optical recognition of the AGNs in the hard energy band is more difficult when compared to the 0.5-4.5 keV band, since about 21% of the AGNs are elusive (while only 10% of the AGNs in the BSS are elusive). The different impact of the problem of dilution on type 1 and type 2 AGNs and on different selection bands should be kept in mind when deriving statistical considerations on the populations of AGNs present in X-ray surveys.

As far as the BL Lac objects are concerned, the selection efficiency in the 0.5-4.5 keV band is about 1-2%. If this efficiency was the same in the 4.5-7.5 keV band, we would expect $\sim $1 BL Lac, something that is statistically consistent with the fact that no BL Lacs are observed in the HBSS sample.

The redshift distribution of type 1 and type 2 AGNs in the two samples is shown in Fig. 10. In the BSS sample, the mean redshift of type 1 AGNs ( $\langle z\rangle _{\rm Ty1}=0.69 \pm 0.03$) is significantly different from the mean redshift of type 2 AGNs ( $\langle z\rangle_{\rm Ty2}=0.47\pm 0.10$), while they are closer in the HBSS sample ( $\langle z\rangle _{\rm Ty1}=0.47\pm 0.06$ $\langle z\rangle_{\rm Ty2}=0.33\pm 0.05$). A K-S test confirms that the z-distribution of the two classes of AGNs is consistent with being derived from the same parent distribution when considering the HBSS sample (K-S probability = 33%), while they are significantly different (at 95% confidence level) when considering the BSS sample (K-S probability = 1.6%). This result probably reflects the fact that the hard-energy (4.5-7.5 keV) selection is less biased by the obscuration (at least in the Compton-thin regime) when compared to a softer (0.5-4.5 keV) energy selection.

Finally, in Fig. 11 we plot the extragalactic XBS sources and the unidentified objects on the magnitude/X-ray flux diagram. The identified extragalactic sources, with the exception of three objects, have an X-ray-to-optical flux ratio (X/O) between 0.005 and 20. At the two "extreme'' ends of the distribution, we find the two "normal'' galaxies that have the lowest values of X/O ($\sim $10-4), similar to those observed in some stars, and, at the other end of the distribution, the high z type 2 QSO, discussed in Severgnini et al. (2006), which has the highest value of X/O ($\sim $200). Among the unidentified sources, we have at least one object whose lower limit on the magnitude (R> 22.8) implies an X/O greater than 60, making it an excellent candidate of high-z type 2 QSO.

  \begin{figure} \par\includegraphics[width=8cm,clip]{8568fig10.eps} \end{figure} Figure 10: Redshift distribution of the type 1 (solid line) and type 2 (dashed line) AGNs in the two samples (BSS and HBSS).
Open with DEXTER


  \begin{figure} \par\includegraphics[width=8cm,clip]{8568fig11.eps} \end{figure} Figure 11: Magnitude vs. 2-7.5 keV flux (as derived from the count-rates) plot of the XBS extragalactic objects plus the unidentified sources: filled points = type 1 AGNs, open circles = type 2 AGNs, open triangles = BL Lac objects, open squares = clusters of galaxies, stars = normal galaxies, crosses = unidentified sources. The continuous lines indicate the region of constant X-ray-to-optical flux ratio, from 0.01 to 100.
Open with DEXTER

Interestingly enough, 3 type 1 AGNs are found among the high (>10) X/O sources. These objects represent a non negligible fraction considering that about half of the high X/O sources are still unidentified, and more cases like these may show up after the completion of the spectroscopic follow-up. A significant presence of type 1 AGNs among high X/O sources has also been found at lower X-ray fluxes ($\sim $10-14 erg s-1 cm-2) in the XMM-Newton Medium sensitivity Survey (XMS, Barcons et al. 2007).

11 Summary and conclusions

We have presented the details of the identification work of the sources in the XBS survey, which is composed of two complete flux limited samples, the BSS and the HBSS sample, selected in the 0.5-4.5 keV and 4.5-7.5 keV bands respectively. We secured a redshift and a spectroscopic classification for 348 (including data from the literature) out of 400 sources, corresponding to 87% of the total list of sources and to 87% and 97%, considering the BSS and HBSS samples separately.

The results of the identification work can be summarized as follows:

Acknowledgements
We thank the referee for useful suggestions. This paper is based on observations made with: ESO Telescopes at the La Silla and Paranal Observatories under program IDs: 069.B-0035, 070.A-0216, 074.A-0024, 075.B-0229, 076.A-0267; the Italian Telescopio Nazionale Galileo (TNG) operated on the island of La Palma by the Fundación Galileo Galilei of the INAF (Istituto Nazionale di Astrofisica) at the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias; the German-Spanish Astronomical Center, Calar Alto (operated jointly by Max-Planck Institut für Astronomie and Instututo de Astrofisica de Andalucia, CSIC). A.C., R.D.C., T.M., and P.S. acknowledge financial support from the MIUR, grant PRIN-MUR 2006-02-5203 and from the Italian Space Agency (ASI), grants n. I/088/06/0 and n. I/023/05/0. This research made use of the Simbad database and 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. The research described in this paper was conducted within the XMM-Newton Survey Science Center (SSC, see http://xmmssc-www.star.le.ac.uk.) collaboration, involving a consortium of 10 institutions, appointed by ESA to help the SOC in developing the software analysis system, to pipeline process all the XMM-Newton data, and to exploit the XMM-Newton serendipitous detections.

References

 

  
Online Material

Table 3: Optical properties of the identified XBSS sources.


Copyright ESO 2008