3. Observation and data analysis
Cyg X-3 was observed with the High Energy Transmission Grating Spectrometer (HETGS, see Markert et al. 1994for a description) on board of Chandra on October 20, 1999, with a total exposure time of 12.3 ksec, starting at 01:11:38 UT. The observation covered the binary phases from -0.3 to +0.4. Therefore the broad minimum of the lightcurve is found at the centre of our observation. The primary goal of the observation was an investigation of the emission spectrum of Cyg X-3 (Paerels et al. 2000). For the present study, we have used the non-dispersed (zeroth order) events only. The use of the grating reduced the countrate by a factor of four compared with a pure imaging observation. Nevertheless, Cyg X-3 is strong enough for producing pile-up effects within the innermost 1 arcsec. The dispersed spectra as well as the `out-of-time' events of the strong point source, i.e., those events recorded while reading out the CCD-detector, were cut out by applying a spatial filter which, in azimuth around the centroid of the source covers a total of 68o. This reduced the number of events used in this analysis by another 19%.
The extraction of a dust scattering halo has to rely on a careful subtraction of the instrumental point response function (PSF) which, at larger radii, is produced by scattering off the mirror surface. The amount of the mirror scattering increases towards higher energies, unlike the scattering on interstellar dust (see Eq. 2). Therefore one expects that the radial surface brightness profile around a source to be dominated in its inner part by the PSF alone, in the outer parts either by the dust scattering (at low energies) or by mirror scattering (at high energies). This behaviour is reflected in Figs. 2 and 3: at energies between 5 and 7 keV, intensity variations of the sources can be seen at all radial distances. In contrast, below 2 keV these variations disappear at radial distances greater than about 20 arcsec because the delay of the radiation scattered on interstellar dust becomes comparable with or larger than the timescales of the intrinsic intensity variations (4.8 h). Those are not only delayed but also smeared out because the delay is connected with the dust distribution along the line of sight (Eq. 4). A simple estimate using Eq. 4 with a mean value of equally distributed dust along the line of sight gives a distance of Cyg X-3 of about 10 kpc. Dickey (1983) has found a lower limit of 9.2 kpc 1 using 21 cm wavelength absorption data, Predehl & Schmitt 1994derived 8 kpc as distance through the galactic dust layer from their comparison of X-ray scattering and absorption.
In order to check this result independently and more precisely, we have constructed lightcurves within different energy bands and annuli of different radii around the source (Fig. 4). At energies between 5 and 7 keV, the total halo intensity is reduced by more than a factor of ten compared with the lowest possible energy range of 0.8 to 2 keV (Eq. 2). Here, the low energy limit is given by the interstellar absorption, the upper limit is chosen in order to get sufficient photon statistics. As expected, the lightcurve in the 5 to 7 keV band is independent of the annulus selected. Even at radial distances = 50 arcsec (not shown here) it remains almost unchanged. In contrast, in the low energy band, where the scattering halo dominates, the lightcurve is shifted by the time delays of the scattered radiation and is also damped.
In a second step, we have made a Monte Carlo simulation using the data itself: the high energy events (which reflect the intrinsic source variability) were `delayed' by a simulated scattering on interstellar dust which was assumed to be uniformly distributed between source and observer. Using Eq. 4, the distance was varied until we got a match between real and simulated lightcurves (dashed lines in Fig. 4). A problem arose from the fact that the observation was shorter than one orbital cycle. This leads to distortions at both ends of the lightcurve if the delays became comparable with the duration of the orbital period. As a result of this exercise, we have determined the distance d = kpc to Cyg X-3.
© European Southern Observatory (ESO) 2000
Online publication: May 3, 2000