6. Optical counterparts of the X-ray sources
We generated finding charts in the B-band COSMOS blue plates and in the I-band (NTT EMMI , cf. Fig. 6) to search for optical counterparts of the ROSAT sources. The B-band plates were used to classify the source using the flux ratio while the I-band image was used to determine the source morphologies from an analysis of their point spread function (see below).
6.1. Matches in the B-band
We searched for the optical counterparts in the 90% confidence circle of the ROSAT source. We list the parameters of the matches in Table 3. For the two sources 28 and 92 we find bright (B=11.7 mag and B=12.1 mag) optical counterparts in the 90% confidence circle of these ROSAT sources. This indicates a galactic foreground nature of these sources. For the sources 36, 41, 52, 68 and 73 we find weak (B=20.4 mag to B=22.3 mag) unresolved optical counterparts. For source 73 there is also a galaxy (B=22.7 mag) in the error circle. For source 53 we cannot find an optical counterpart as it is in the crowded field of NGC 3109 and for 63 no optical counterpart exists within the ROSAT 90% confidence circle.
Table 3. Optical (COSMOS blue plates) matches for ROSAT sources from Table 2. Given is the B-mag, and some rough source classification (stellar, galaxy, unresolved or faint).
In addition we calculated the flux ratio with the equation given in Haberl & Pietsch (1999). We make use of the B-magnitude given in Table 3 and the PSPC countrate given in Table 6. We use the abbreviation and obtain the flux ratio from the equation:
6.2. Matches in the I-band
For the first run of optical matching of X-ray sources in the central part of NGC 3109 we obtained three optical I-band images (60.E-0818) from the ESO archive. 1 All images were taken on 1998 February 2 with the red arm of the ESO Multi-Mode Instrument (EMMI ) at the New Technology Telescope (NTT ) with a total exposure time of 2700 s. The images were taken with the Tektronix 20482048 pix2 chip with a scale of 0.268" /pix. The resulting field of view covered by the stacked images is 8.8´ 8.8´ and the final image is centered on the galaxy nucleus. The seeing was measured to be " throughout all three exposures.
The data are used to identify optical couterparts within the region of highest HI column density, which resides at the center of NGC 3109 (Jobin & Carignan 1990) and is best covered by the field of view of this particular dataset, and to support or reject the classification provided by the hardness-ratio estimate (see Sect. 5). Further photometric studies in optical passbands of ROSAT sources for most objects in the catalog in Table 6 with accurate positions will be given in a subsequent paper.
In order to obtain bona-fide optical counterparts we transformed the pixel coordinates of the I-band to new equatorial coordinates. We used positions of 43 stars out of the USNO Astrometric Catalog (Monet et al. 1998) which was obtained from Centre de Données de Astronomiques de Strasbourg (CDS). Using the task ccmap within the IRAF 2 environment (Tody et al. 1993) we find the mean rms uncertainty for all given optical coordinates being RA" and Dec". Applying the plate solution to coordinates provided by the ROSAT PSPC catalog (see Table 6) we find 7 optical matches within the field of view. The finding charts for all these regions are given in Fig. 6 with the appropriate scaling of the 90%-confidence circle as given in Table 6. The five brightest objects are encircled and labeled according to their luminosity (label 1 marks the brightest object).
Subsequently, we perform an analysis of the shape parameters of the optical point spread function (PSF) for the 5 brightest objects within each of the 7 object's finding circles. We used the source-extraction software SExtractor v2.1.4 (Bertin & Arnouts 1996) for determination of ellipticity , full width at half maximum (FWHM), and the star-galaxy-classification parameter CL (using a neural-network algorithm which was extensively trained, see Bertin & Arnouts for details). We compare our findings to all remaining optical objects most of which are assumed to be point-sources. The code yields 7500 detections for which we plot the three PSF-shape parameters (, FWHM, and CL) as a function of instrumental I-band magnitude. Fig. 7 shows the parameters of all detections together with the parameters for the brightest source within the finding circle (see Table 4). Note that the brightest object is not necessarily the optical counterpart of the ROSAT source.
Table 4. Shape parameters of the 5 brightest sources within 90% confidence finding circle of 7 optically matched ROSAT source coordinates. Col. 1 gives the number which agrees with the labels in Fig. 6. Col. 2 and 3 are the instrumental magnitude and the corresponding error. The FWHM, ellipticity, and the star-galaxy-parameter (from 1=star to 0=galaxy) as calculated by SExtractor are given in col. 4, 5, and 6.
The brightest sources 52.1, 62.1 and 65.1 (the first number being the source number and the decimal the source label) in Fig. 6 are at least one magnitude brighter than the next fainter object within the finding circle (see Table 4). Object 63.1 has about the same magnitude as object 63.2. Therefore no clear preference can be given just from considering the I-band flux. The PSF-shape parameters of these 4 objects agree with those found for stars (see Fig. 7). The CL classification of SExtractor puts all 4 sources in the point-source regime. Yet, the optical findings are prone to mis-classification since not every brightest object is most central within the finding circle.
Although they are not the brightest, the sources 62.5 and 63.2 are the most likely candidate optical counterparts (infering from their most central position only). These were classified as slightly extended and slightly elliptical, respectively.
As can be seen from Fig. 6 there are 2 finding charts (i.e. charts 62 and 65) from the edge of the CCD image. Thus, we cannot exclude brighter objects within the area not covered by the exposure. Chart 59 is too crowded to give a reliable identification only on the basis of any photometric PSF-shape parameter. However, we note that the center of the ROSAT error circle coincides with the galaxy's center. Thus, it is very likely that the ROSAT source is connected with a X-ray source which is located near the center of the galaxy.
© European Southern Observatory (ESO) 2000
Online publication: October 2, 2000