Astron. Astrophys. 319, 525-534 (1997)

3. Data analysis

3.1. The sample

We selected from the ROSAT Public Data archive PSPC pointings in the direction of the Cyg OB7 and Cyg Rift molecular clouds. A description of the satellite and the detector can be found in Trümper (1983) and Pfeffermann et al. (1986). The 28 analyzed areas are shown in Fig. 1, overlaid onto the CO contour maps from Dame & Thaddeus (1985). The relevant data on the pointings are reported in Table 4. A few pointings lie just outside the contour of the clouds in Fig. 1, but have been nevertheless included in the sample. The total exposure time is 126603 seconds. Only the central region of the PSPC (radius ) has been considered for the analysis, to avoid problems due to the poor knowledge of the point spread functions outside that region. The areas of the two clouds covered by ROSAT pointings are 2.1 square degrees (Cyg OB7) and 5.6 square degrees (Cyg Rift), which correspond to 4.6% and 5.0% of the total areas of the clouds, respectively (adopting the values quoted in Dame & Thaddeus 1985). Because of the nature of the pointings, the fractional coverage of the cloud areas is limited and varies with threshold (see Tables 2 and 3). Although at the lowest sensitivity limits (1-2 c s^-1) only 1-2% has been covered, the theoretical expectations (see again Tables 2 and 3) indicate that this could be sufficient for detecting ONS candidates. Moreover, from the systematic analysis of these data a search methodology can be outlined.

Table 4. Observation log. The first two columns are the identifier of the pointing in this paper and the ROSAT observation number. Galactic coordinates are in decimal degrees. is the number of detected X-ray sources and is the number of GSC stars used for the boresight correction (noB indicates one star or less, see text).

3.2. The analysis procedure

Each pointing has been extracted from the ROSAT Public Archive at MPE Garching and analyzed separately by running a set of semi-automatic programs. The standard detection technique in the EXSAS package (Zimmermann et al. 1994) has been applied. First a background map is produced by removing all possible sources (identified by means of a sliding window technique) and running a two dimensional spline fit to the data. Then a Maximum Likelihood (ML) algorithm is applied (see Cruddace, Hasinger & Schmitt 1988) to detect significant deviations from the estimated background distribution. The threshold for detection has been set to the conventional value of ML=10, which corresponds to a chance detection probability of 4.5 for a single trial. This technique is repeated for the data in three different PHA channel ranges: total (T: channels 11-240, 0.1-2.4 keV), soft (S: channels 11-40, 0.1-0.4 keV) and hard (H: channels 41-240, 0.4-2.4 keV). In case of a detection in more than one band, only the band with the highest value of ML has been considered.

It is known that the ML technique becomes problematic in presence of large regions of extended emission, as it is likely to find in the pointing directions considered here, lying on the galactic plane. Therefore, all the detected sources have been visually inspected in the X-ray images and obvious spurious detections have been discarded.

In the cases where two or more pointings were coaligned, they were analyzed separately: the sources have been cross-correlated and in case of coincidences only the source with the highest detection ML has been considered. Eleven sources have been eliminated in this way, bringing the total number of sources to 109. The - distribution for the sources detected by ROSAT is shown in Fig. 2. There are a few strong sources, but most of them are below 0.01 c s^-1. Note that the targets of the pointings have not been removed.

3.3. Optical identification

The positions of the detected X-ray sources had to be corrected for an offset due to a residual systematic uncertainty in the position determination (the so-called boresight correction). In order to do this, preliminary identifications have been performed using the HST Guide Star Catalog (GSC; Lasker et al. 1990). The X-ray sources and GSC stars have been cross-correlated and the best shifts to the X-ray positions have been determined by means of a Maximum Likelihood technique. The shift so obtained has been considered only if at least two sources were identified in this way (with the exception of pointing S, where one of the two sources identified with GSC entries had multiple bright optical counterparts). For the remaining pointings, a (1 ) error has been quadratically added to the error radii of the detected sources (Kürster & Hasinger 1992). In Table 4 a summary of the X-ray detections and boresight corrections is reported. An X-ray source is considered to be identified if a GSC star falls into its 90% positional error circle. In case of more than one optical identification, the brightest counterpart has been considered.

67 X-ray sources were identified with entries in the GSC, which has a loosely defined magnitude limit of . For the remaining 42 sources, we used the Digitized Sky Survey (DSS; Postman et al. 1995), which consists of the POSS red plates. The magnitudes of the brightest stars in the error boxes have been estimated by using the empirical magnitude-diameter relation of King & Raff (1977) appropriate for red POSS plates. The values so obtained have a large uncertainty (up to one magnitude), but it is sufficient for our purpose. Only in three cases the possible optical counterpart was not bright enough to allow an estimate of the magnitude, the image being not saturated. In Fig. 3 we show the distribution of magnitudes of the optical counterparts for the two samples, GSC and DSS. As one can observe there is a considerable number of bright stars, since there are bright OB associations in the direction of most of the pointings; indeed a large number of the counterparts for which we were able to obtain a spectral type were of types O and B. The presence of strong optical candidates for many X-ray sources ensures the robustness of the boresight process.

Fig. 3. Distribution of optical magnitudes for the possible counterparts of the X-ray sources. GSC magnitudes are in the V band, POSS magnitudes in the red (see text). The lack of POSS stars at magnitude 19 is a probable artifact of the procedure for magnitude estimate.

After the optical screening, 7 sources were left for which no optical counterpart could be seen in the POSS red plates, to a limiting magnitude 20. Two of these were coincident with the original targets of the pointings and were therefore excluded. The brightest of the remaining sources was subsequently identified with Cyg X-3, so 4 final candidates remained.

3.4. ONS candidates

For the four sources with no optical counterpart, PSPC count rates in the different bands have been extracted manually from the data, with an accurate estimate of the background, in some case contaminated by extended emission. The relevant information about the candidates (two in each cloud) can be found in Table 5. As it can be seen from Table 4, all four sources come from pointings with long exposure times and with a large number of GSC identifications, which ensures an accurate boresight correction. Source OB7-2 lies just outside the outer rim of the cloud (see Fig. 1), but we decided not to discard it from the sample.

Table 5. ONS candidates. The columns are: source name, pointing identifier, celestial coordinates, 90% error radius, Maximum Likelihood of existence, PSPC total count rate (channels 11-240) and hardness ratio (defined as (H-S)/(H+S), where S and H are the counts in the channel intervals 11-40 and 41-240 respectively.

Source Rift-2 might have counterparts at different wavelengths. The POSS plate shows a very faint ring-like structure ( diameter) in the X-ray error box, reminiscent of a supernova remnant or a planetary nebula. The position of the source is compatible with that of the IRAS source 20179+4102, which has a 100 µm flux of 75 Jy (IRAS Catalogs and Atlases 1985). Putting the distance at 0.7 kpc, this translates roughly to a luminosity of erg s^-1. At the same distance the radius of the optical ring would be pc. If the ring and the IRAS sources were the same object, it would be too small and faint to be a supernova remnant, while the planetary nebula hypothesis could not be ruled out. In absence of a clear identification we do not exclude the source from our candidates list.

© European Southern Observatory (ESO) 1997

Online publication: July 3, 1998
helpdesk@link.springer.de