2. Observations and results
Let us start with the PSPC observation which was performed earlier than our HRI observations.
2.1. PSPC all sky survey observations
The source was first detected by ROSAT between September 5th and 6th, 1990 during the PSPC all sky survey. The position was retrieved from the 1RXS bright source catalog (Voges et al. 1996). KS 1731-260 is identified as 1RXS J173413.0-260527 at the position 17 34 13.0 and 05' 27.5" (equinox 2000) and has an associated error radius at the 90% confidence level of 17.2 arcsec. As shown in Fig. 1, this position is consistent with the HRI positions derived below.
As shown in Fig. 2 in the PSPC data, the source flux does not show any significant time variability. By selecting only photons from the inner region of the detector (i.e where vigneting corrections are unimportant), the source count rate was 18.2 cts s-1. Note that the value listed in the 1RXS catalog is only 14 cts s-1 ; in the 1RXS analysis the whole detector area is taken and an improper exposure time correction at the edge of the detector can cause systematic errors of the order of 20% (Voges, private communication). A refined spectral analysis has been performed in order to determine NH towards the source. Unfortunately, the limited energy range of the PSPC does not allow to distinguish between soft (blackbody) and hard models (Bremsstrahlung and power law). However, in all three models, the NH derived is consistent with a value of H atoms cm-2 with an error of about 0.3. This is about a factor of two lower than the value observed by TTM and a factor of five lower than the PCA value. The difference between the TTM and RXTE values, together with this result strongly suggest that KS 1731-260 is characterized by variable intrinsic absorption. In Fig. 3 we show the allowed grid of variations of the power law index and NH. As can be seen, KS 1731-260 is characterized by a very hard spectrum in the PSPC range (photon index around 1.). The PSPC count rate corresponds to a flux in the 0.1-2.4 keV range of ergs s-1 cm-2 for the power law model.
2.2. HRI observations
The ROSAT-HRI observation of KS 1731-260 was split in two parts: KS 1731-260 was first observed between March 16, 1995 (09:38:11) and March 16, 1995 (22:28:12) for a total exposure time of 1080 seconds, and later between September 3, 1996 (12:27:44.2) and September 14, 1996 (22:34:52) for about 4600 seconds. In both observations, the HRI field was centered on the TTM position. The observations were analyzed with the PROS software for the first one and with the EXSAS package for the second one. The source is clearly detected in both observations with a count rate of and cts s-1, respectively. The best fit positions derived are 17 34 13.5 and 05' 16.8" (equinox 2000), and for the second observation 17 34 13.1 and 05' 16.5" (equinox 2000). The error on the centroid position is arcsec (90% confidence level). Unfortunately, no other X-ray sources were detected during our short observation and therefore no boresight corrections could be made. In a conservative way, we set up the boresight uncertainty to 10 arcsec (David et al. 1996). This combines with the centroid error to give a 90% confidence error radius of 10.1 arcsec. The two positions are perfectly consistent with each other as shown in Fig. 1.
In both observations, the source does not show any variability nor X-ray bursts. Given the possibility that KS 1731-260 is characterized by variable intrinsic absorption as discussed above, conversion of the HRI count rate into unabsorbed X-ray flux may be misleading. However, taking the spectral parameters derived by Smith et al. (1997), the HRI count rate of the second observation corresponds to a 0.1-2.4 keV unabsorbed flux of ergs s-1 cm-2. At 8.3 kpc, this flux translates to an X-ray luminosity about a factor of 10 larger than the Eddington limit for a 1.4 neutron star. On the other hand, decreasing the NH down to H atoms cm-2, the 0.1-2.4 keV unabsorbed flux decreases by a factor of and becomes comparable to the value expected from the TTM/RXTE observations. This suggests that the change by a factor of in the source counting rate may be associated with a change in the NH between our two observations, and thus may not reflect a true change in the source intensity.
2.3. CFHT infrared observations inside the ROSAT HRI error box
Infrared J and H images were obtained at CFHT on 1996 June 1st using the OSIS instrument equipped with the Redeye and Nicmos 256 256 camera. In this configuration, the pixel size is 0.5" on the sky. The total exposure time was 180 s in J and 210 s in H. Raw images were corrected for dark and flat-field using standard MIDAS procedures. Observations of photometric standard stars allowed to derive absolute J and H photometry with a 1 systematic error of less than 0.04 mag in both bands. We show in Fig. 4 the H and J filter images with the HRI error circle overlayed. The infrared images reveal the presence of at least two tens of possible candidates in the small 10.1 arcsec radius HRI error circle derived from the first part of the observation. For indications, only objects C, D, G, H, K are included in the overlapping region of the error boxes (90% confidence level) of the PSPC and HRI positions. In Table 1 we list the magnitudes of the brightest objects located in and close to the HRI error circles.
Table 1. Infrared magnitudes, colors and positions of the brightest objects located in or close to the HRI error circle (equinox 2000). The error on the position is ".
With our observations, we also confirm the identification of IRAS 17311-2604 with variable star 2547 from Terzan & Gosset (1992). Our J and H images (not shown) reveal a bright infrared source at a position consistent with that of the optically variable star. The infrared source is unfortunately so bright that saturation prevents from deriving reliable J and H magnitudes.
© European Southern Observatory (ESO) 1998
Online publication: December 16, 1997