3 The Emission Line objects

In this paper we present newly identified Emission Line objects. We define Emission Line objects those sources for which at least one strong (EW $\gg$ 5 Å in the source rest frame) emission line is present in the spectrum. This criterium excludes BL Lac objects and radiogalaxies without emission lines in their spectrum. There is evidence (Vermeulen et al. 1995; Corbett et al. 1996; Marchã et al. 1996) that some "true'' BL Lacs might show emission lines with an equivalent width slightly larger than the "canonical" limit of 5 Å, proposed by Stocke et al. (1991). However, the majority of the objects presented here show very strong emission lines well above the limit of 5 Å.

We further classify objects as Broad Emission Line (BEL) AGNs if at least one line has a FWHM >1000 km s^-1 (in the source rest frame), or Narrow Emission Line (NEL) object if all the observed lines have FWHM< 1000 km s^-1.

This classification reflects the common division between "type 1'' (i.e. Broad Line) and "type 2'' (Narrow Line) AGNs often discussed in literature. The majority of the EL objects presented here is "type 1'' AGNs. In one case (1REXJ213248-0219.8), the FWHM of H$\beta$ and H$\alpha$ is $\sim 1300-1400$ km s^-1, thus slightly higher than the limit used to classify the object as NEL AGN. This value of FWHM and the ratio between [OIII] and H$\beta$ (=1.4) suggests a classification as narrow-line Seyfert 1 galaxy (NLS1, [OIII]/H$\beta<$3 and FWHM<2000 km s^-1, e.g. Osterbrock & Pogge 1985). In the few cases of Narrow Line objects, we have applied, when possible, the diagnostic criteria described in Veilleux & Osterbrock (1987) to separate a starburst galaxy from an AGN or LINER. However it is worth noting that, as discussed in Veilleux & Osterbrock (1987), the separation between starburst galaxies and AGNs is not very sharp and a mixture of both phenomena could be present in the same object (Hill et al. 1999; Goncalves et al. 1998). Given the quality of our spectra, we didn't attempt to estimate a correction for reddening. However, for the classification of these objects we have used, as long as possible, the ratios between couples of emission lines like [OIII]$\lambda$5007/H$\beta$, [NII]$\lambda$6583/H$\alpha$, [SII]$\lambda$6716+$\lambda$6731/H$\alpha$ that are not very reddening sensitive.

We briefly discuss here the 4 NEL objects for which we can apply the Veilleux & Osterbrock (1987) diagnostic criteria. 1REXJ005924+2703.5: the values of the ratios between [SII]$\lambda$6716+$\lambda$6731 and H$\alpha$ (Log([SII]/H$\alpha$) = -0.29) and between [OIII]$\lambda$5007 and H$\beta$ (Log([OIII]/H$\beta$) = 0.02) put this object in the zone intermediate between AGNs and HII regions.

1REXJ022840-0935.2: the intensity ratio between [OIII]$\lambda$5007 and H$\beta$ (Log([OIII]/H$\beta$) = 0.26) and between [NII]$\lambda$6583 and H$\alpha$ (Log([NII]/H$\alpha$) = -0.9) suggests a classification as H II region or starburst galaxy rather than AGN for this object.

1REXJ121815+0744.4: the H$\alpha$ is blended with [NII]$\lambda$6548 and it is difficult to measure its width. The FWHM is about 800-1000 km s^-1. The strong intensity of the [OIII]$\lambda$5007 line if compared with the H$\beta$ (Log([OIII]/H$\beta$) = 0.88) and the ratio between [SII]$\lambda$6716+$\lambda$6731 and H$\alpha$ (Log([SII]/H$\alpha$) = -0.37) strongly suggest the presence of an AGN.

1REXJ230311-0859.3: the H$\alpha$ is blended with [NII]$\lambda$6583. The FWHM of the H$\alpha$ is about 1000 km s^-1although a proper de-blending is necessary for a more accurate measurement. The measured value of Log([N II]$\lambda$6583/H$\alpha$) $\sim -0.15$ and the lack of [SII]$\lambda$6716+$\lambda$6731 lines lead to a tentative classification of this source as Narrow Emission Line galaxy (NELG). This object belongs to a galaxy pair.

For the remaining 4 NEL objects (1REXJ041734-1154.5, 1REXJ220451-1815.5, 1REXJ235139-2605.0, J014716-0008.2) we cannot apply the diagnostic criteria described above. These objects typically show [OII]$\lambda$3727 and [OIII]$\lambda$5007 lines. All these 4 NEL objects show an X-ray luminosity exceptionally high for this class of source ( $L_{\rm X}>10^{44}$ erg s^-1). One of these sources, 1REXJ235139-2605.0, is the cD galaxy of the cluster A 2667 which is very luminous in the X-ray band (2.1$\$10⁴⁵ erg s^-1in the 0.1-2.4 keV band, Rizza et al. 1998). In other two cases (1REXJ041734-1154.5, 1REXJ220451-1815.5) the CCD image taken at the UH 88'' telescope reveals the presence of an overdensity of galaxies around (< 1 $\hbox{$^\prime$ }$) the radio source thus suggesting that these objects belong to a cluster/group of galaxies. Finally, J014716-0008.2 is probably in a cluster since at least two companions at the same redshift of the radio source, have been found. In conclusion, all the NEL objects presented in this paper with an X-ray luminosity greater than 10⁴⁴ erg s^-1 probably belong to a cluster/group of galaxies. In these cases, the very high X-ray luminosity is probably originated from the intracluster gas and not from the galaxy itself. Further X-ray observations with higher spatial resolution are required in order to measure the true luminosity of the object. On the contrary, all the isolated NEL objects have a low luminosity in the radio and in the X-ray band ( $L_{\rm X}<2\ 10^{43}$ erg s^-1, $P_{1.4}\leq10^{31}$ erg s^-1 Hz^-1).

The classification of the EL objects in Broad Line (B) and Narrow Line (N) objects is reported in Table 5 (Col. 2) together with the measured redshift and the relative confidence (Cols. 3 and 4), X-ray and radio luminosities (Cols. 5 and 6), the observed features in emission (Col. 7) and in absorption (Col. 8). The luminosities have been computed and K-corrected under the assumption of power-law spectra ( $f_{\nu}\propto\nu^{-\alpha}$). In the radio band we have used the spectral index ( $\alpha_{\rm R}$) as computed between 1.4 GHz (NVSS) and 5 GHz if the source is included in the GB6 (Gregory & Condon 1991) or in the PMN (Griffith & Wright 1993) catalog. For three sources (1REXJ061757+7816.1, 1REXJ082656+6542.5, 1REXJ085211+7627.3) not present in the GB6 and PMN catalogs, we have used the flux at 365 MHz derived from the WENSS survey (Rengelink et al. 1997) to estimate the radio spectral index. Finally, if no fluxes at frequencies different from 1.4 GHz are available, we have used the lower limit on $\alpha$ computed taking into account the proper flux density limit at 5 GHz for the GB6 ($\sim$ 18 mJy) and PMN catalogs (from 40 to 72 mJy depending on the declination). These monochromatic luminosities (indicated by an asterisk in Table 5) should be considered as lower limits. The actual value of the luminosity is close to these values for low-redshift objects (z<0.5) while it can change up to a factor of 10 for those sources at z>1.

Figure 3: Distribution of the radio spectral index between 1.4 GHz and 5 GHz ( $f\propto \nu ^{-\alpha }$) for the AGN identified so far in the REX sample and for which a radio flux density at 5 GHz was available. The shaded histogram represents the sources with a flux density at 1.4 GHz larger than 200 mJy and falling in the area of sky covered by the GB6 catalog

In the X-ray band, the assumption of $\alpha_{\rm X}=1$ is adequate to convert the count-rate into an X-ray flux (error less than 20%) but it is not good enough for the K-correction, in particular for high redshift objects. Thus, we have used the relationships found in Brinkmann et al. (1997) that give the $\alpha_{\rm X}$ as a function of the redshift for a sample of X-ray selected radio-loud AGNs found in the ROSAT All Sky Survey (RASS). In particular, we have used the appropriate relationship for steep spectrum (SS, $\alpha_{\rm R}>0.5$) and flat spectrum (FS, $\alpha_{\rm R}<0.5$) radio sources given in Brinkmann et al. (1997) on the basis of the radio slope between 1.4 GHz and 5 GHz. If a radio slope is not available, we have used the relationship between $\alpha_{\rm X}$ and z proper for the SS, if the lower limit on $\alpha_{\rm R}$ (based on the GB6 or PMN flux limits) is larger than 0.5; otherwise, we have used the formula for the FS. The used values of $\alpha_{\rm X}$ range from 0.8 to 1.15.

Figure 4: Redshift distribution of the 226 AGNs found in the REX survey so far

A separate table (Table 6) contains the EL objects that are not included in the final version of the REX catalog. These sources have been excluded because they do not satisfy anymore the final selection criteria of the REX survey described in Caccianiga et al. (1999).

Figure 5: Monochromatic radio luminosity versus X-ray luminosity for the AGNs belonging to the REX survey and for which we have a measured radio spectral index. The circles represent the objects presented in this paper while the triangles represent the objects previously identified. The solid line represents the best least squares fit to the data

We note that the spectra presented here have been collected with the primary aim of identifying the sources and measuring the redshifts. The nominal error on the redshift is $\sim$0.001 (at 5000 Å) but in many cases, given the signal-to-noise ratio of the spectrum, the error is larger (about 0.005 - 0.01). When only one strong emission line is present in the spectrum, we have assumed, as usual, that this feature is the MgII$\lambda$2798. We have flagged as tentative the values of redshift computed in these cases (a letter "T'' after the value of redshift in Tables 5 and 6). We have also flagged as tentative the redshift of 1REXJ220451-1815.5 for which we have assumed that the two observed features are [OII]$\lambda$3737 and [OIII]$\lambda$5007 respectively.

All the sources presented here were unidentified when we observed them. During the preparation of this paper, 12 sources have been observed and identified independently by other authors, in particular during the identification of other surveys that combine X-ray and radio data (i.e. the DXRBS survey, Perlman et al. 1998 and the RGB survey, Laurent-Muehleisen et al. 1998). Moreover, in two additional cases we have intentionally re-observed the object because the redshift was based on poor quality data (i.e. low dispersion prism observations, 1REXJ005229-3743.8) or we wanted to assess the presence of the H$_\alpha$ in emission (1REXJ230311-0859.3). Except for 1REXJ005229-3743.8, for which we have found a redshift significantly different from the one published (1.60 instead of 2.25), in all the other cases we confirm the published values of redshift, with discrepancies below 3% and only in one case of 6%. A footnote in the source name in Tables 5 and 6 indicates that an optical identification and a redshift for that object have been already published in literature.

In Figs. 1 and  2 we present the spectra of all the EL objects included (Fig. 1) and not included (Fig. 2) in the REX catalog. The y axis (the flux) has arbitrary units.