3. Identification criteria

The basic criterion for the classification of optical counterpart candidates is the appearance of the high-resolution density spectra. A small collection of typical spectra are shown in Fig. 2 (click here). In some of the brighter density spectra emission lines (cf. classification CV, QSO in Fig. 2 (click here)) and absorption features (stars of spectral type FG, K and M, some WDs) can be distinguished, and the shape of the continuum can be derived (cf. blue continua of AGN, WDs). Extended direct images help to identify galaxies which often show featureless spectra (cf. galaxy, blue galaxy). At lower densities, near the plate limit, only basic colour information is available and we classify the spectra in "extreme blue and weak '', "blue and weak '' and "red and weak '' spectral categories. We associate the blue spectra with X-ray emitting AGN, while the red and weak spectra are considered as unidentified sources. To identify the most plausible counterpart, we required the candidate to have in addition to an acceptable positional distance to the X-ray source to have a log(/f_B) value consistent with its object class (Stocke et al. 1991).

 
Figure 2:   Examples of high resolution objective prism spectra for several important object classes. The inset in the upper right of each panel contains the low resolution spectrum. The object classes GALAXY and BLUE GAL are only given if the direct image is extended

For the final evaluation and the determination of the most plausible optical counterpart all obtainable information is put together: sky positions from the direct plates, classifications and optical magnitudes (or upper limits) from the spectral plates, hardness ratios and X-ray to optical flux ratios (log(/f_B)). For white dwarfs (WDs) and some kinds of cataclysmic variables (CVs) information about the softness of the X-ray spectrum (from the hardness ratios) is useful for their classification. Information about the extent of X-ray sources is also helpful for identifying some object classes (e.g. clusters of galaxies). Based on this information and applying various classification criteria (Bade et al. 1992b) a final classification code is determined, and an entry in a "master catalogue of identifications'' is constructed.

The classification is coded by a two or three-digit number. The first digit identifies the class of objects, with "1'' to "3'' identifying extragalactic objects, and "5'' to "7'' stellar objects (Table 1 (click here)). A third digit is present for classes that could be subdivided, e.g. the stars. The second digit describes the reliability of the classification:

"0''

"highly probable'', the proposed counterpart fulfills all requirements of its class and no other plausible counterpart is found in the X-ray error circle.

"1''

"probable'', the proposed counterpart fulfills the requirements of its class, but there are some limitations. Either the objective prism spectrum is not typical, or small conflicts with the X-ray information (spectral or spatial information, distance to X-ray position) are present or there is another (considerably less) plausible counterpart in the error circle.

"2''

"possible'', the proposed counterpart fulfills some requirements of its class, but there are doubts arising from insufficient objective prism data, conflicting X-ray data (e.g. extended emission for an AGN) or another neighbouring plausible counterpart.

In Table 2 (click here) the portion of the reliability classes is tabulated for the relevant object classes.

 

2 Galaxies 138 extended optical image, contain some AGN

3 Galaxy clusters 113

5 M-dwarfs 155

6 White Dwarfs 31

7_1 K Stars 136 can also contain some stars of early M type

7_2 F or G Stars 4 Classification needs further support from high resolution spectra

7_3 CVs 16

7_4 Bright Stars 956

8 Unidentified 619 Majority consists probably of optically weak AGN and clusters

0 Empty 105 empty on objective prism and direct HQS IIIa-J plates

Table 1:   Classification codes with contents

 

Table 2:   Percentage of the reliability classes in the object classes

3.1. Object classes

The following classes of objects were defined (Table 1 (click here)):

• QSO/AGN (Class 1). This is the most frequent classification. Their objective prism spectra show a blue continuum and occasionally emission lines (Fig. 2 (click here)). Other objects with similar spectra, which can be confused with QSO/AGN, are faint (B>17) hot subdwarfs or white dwarfs. They are generally ruled out because of an unlikely high log(/f_B) -ratio. White dwarfs also have very soft X-ray spectra, often without hard photons above 0.4keV, which is untypical (but not impossible) for AGN. Another class with objective prism spectra potentially similar to QSO/AGN are cataclysmic variables. Follow-up spectroscopy has shown that the fraction of CVs among the QSO/AGN candidates is below .

   
  Figure 3:   Example for a finding chart taken from the ftp-Server with an optical counterpart classified as GALAXY (Code 20)
  

• Galaxies (Class 2). This classification is given, if the counterpart is clearly extended on the direct plates (Fig. 3 (click here)) and the objective prism spectrum shows no AGN characteristics. Objective prism spectra are taken without slit. Therefore features in objective prism spectra of extended objects are smeared out, which hinders a classification of such spectra. Follow-up spectroscopy of members of this class shows that they are X-ray luminous early-type galaxies and partly Seyfert 1 galaxies with z < 0.1. Some galaxies might be members of unrecognized distant galaxy clusters.
• Galaxy clusters (Class 3). This class of X-ray sources is identified mainly by their appearance on the direct plates. The prism spectra of galaxies in clusters are inconspicuous and of little help. Often hard X-ray spectra, as inferred from the hardness ratios, and information that the X-ray source is extended, helps to increase the plausibility of the identification. For redshifts z > 0.35 the 4000Å break of the cluster galaxies is shifted out of the sensitive range of the KODAK III-aJ emulsion. Therefore the detection probability of such galaxy clusters is very low on HQS Schmidt plates.
• M-dwarfs (Class 5). These are frequent counterparts of RASS sources and are easy to identify, because of the presence of strong TiO absorption in the spectra (Fig. 2 (click here)). Comparison of their positions on the Palomar Sky Survey I and on our direct plates reveal frequently proper motions. As the HQS Schmidt plates were taken within the last ten years, identification problems due to large proper motions are negligible compared to the use of POSS I.
• White dwarfs (Class 6). The current version of the HRC contains 31 WDs. They are relatively bright (B<17), have very blue density spectra, often with weak and broad Balmer absorption lines (DA) (Fig. 2 (click here)) and have in general very soft X-ray spectra.
• Stars (Class 7). This class contains several subgroups indicated by the third digit. "1'' as third digit stands for stars of spectral class K. They have rather red continua and a CaII absorption line (Fig. 2 (click here)). Due to their typical log(/f_B) ratios main sequence stars are usually saturated on the objective prism plates, if they were detected by the RASS. In particular for partly saturated stars the inclusion of some M or K dwarfs to this subclass cannot be excluded. "2'' as third digit stands for stars of spectral type F and G. Only few X-ray sources were identified with this subclass. These identifications need further support by more detailed optical spectra since most of the known X-ray emitting F or G stars are optically brighter than the saturation limit of the HQS plates. The next subgroup are the cataclysmic variables (CVs) with the third digit "3''. They are rare on the high galactic latitude sky, but nevertheless CVs with strong emission lines are easily detected in the objective prism spectra (Fig. 2 (click here)). Optically weak CVs near the plate limit show similar spectra as AGN and can therefore been mixed up with this class. X-ray sources identified with saturated spectra have the third digit "4''. The objective prism plates saturate in the range 12 < B < 14, so that for many X-ray sources the only classification criterion is the positional coincidence with the X-ray source and the lack of other plausible candidates. Among them unrecognized M-dwarfs (Class 5) and white dwarfs (Class 6) can hide, although the WDs would need an untypical flat X-ray spectrum to get lost in the identification process.
• Unidentified (Class 8). This class comprises all X-ray sources for which no plausible counterpart was assigned, although objects were found on the Schmidt plates inside the error circle. But the corresponding objective prism spectra were unclassifiable or we did not find a plausible counterpart or, in a very few cases, more than one plausible counterpart was found. Several subclasses were defined, but we kept only one. The third digit "3'' designates X-ray sources for which we found a single candidate on the direct plate too faint to show a spectrum. In general, they turn out to be members of the "QSO/AGN'' class.
• Empty fields (Class 0). These X-ray sources do not have an optical counterpart neither on the spectral plate nor on the direct plate. Often the HQS direct plate are just not deep enough and optical counterparts are found on the Palomar Sky Survey. Nevertheless, these sources often have a high log(/f_B) -ratio and were successfully used for an efficient search for BL Lac objects (Nass et al. 1996).