Astron. Astrophys. 331, 251-261 (1998)

2. XMPC instrument details and observations

Hard X-ray observations of Cyg X-1 were carried out in a balloon flight of a telescope consisting of two xenon filled multi-anode proportional counters (XMPCs) each with an area of 1230 cm², carried out on 1992 April 5/6 from Hyderabad, India. These detectors have an average X-ray detection efficiency of about 50% between 20 and 100 keV. Active volume of the detectors is divided into three layers, with four anode cells of cross-sectional area 4.8 cm 4.8 cm in each layer. Alternate anode cells in each layer are joined together. The anode cell assembly is surrounded by veto cells on three sides. In order to reduce background induced by charged particles, anode cells are operated in mutual anti-coincidence to reject simultaneous events from different anodes. To avoid rejection of genuine X-ray events above 34.5 keV, escape gating technique is used, which accepts two simultaneous events, provided one of them is in 25 to 35 keV band, corresponding to the xenon K-shell fluorescent event. In case of the escape gated event, only non-K event is analyzed for the pulse height and the energy of the K-shell fluorescent event is assumed to be 29.7 keV. The veto layer is operated in anti-coincidence with the main detection cells to reject background induced by charged particles. A mechanical graded slat collimator of tin and copper restricts the field of view to FWHM. For details of the X-ray telescope refer to Rao et al. (1987, 1991).

The balloon flight was carried out on 1992 April 5 at 19:01 UT, and the balloon reached a ceiling altitude corresponding to a residual atmospheric column density of 4 gm cm^-2 at 21:15 UT. The payload is oriented using an alt-azimuth orientation system with a pointing accuracy of . Source tracking is done according to azimuth and elevation angles stored in an on-board programmer, which are updated every minute. Cyg X-1 was observed continuously for a duration of one hour starting at 1:50 UT on 1992 April 6, followed by 4 cycles of source and background observations, for 20 and 10 minutes, respectively. Count rate profile of Cyg X-1 obtained from one of the XMPCs during this balloon flight is shown in Fig. 1. The effect of change in air mass shows as a decrease in the source count rate away from the meridian transit time. As can be seen from the figure, the background count rate remained steady throughout the observations. The gap in the observation before the meridian transit is the duration for which the source NGC 4151 was being tracked. Aspect calibration by the triangulation method is attempted several times during the beginning of the observations. From a detailed fitting of the variation of count rates during aspect calibration and also with the zenith angle of the source, we estimate that the error in the orientation of the telescope can contribute to an extra vignetting correction of about 0.2 (corresponding to an angular offset of ). Increase in the count rate for a duration of about 10 minutes at the end of the one hour tracking is due to a gamma-ray burst. Average Cygnus X-1 count rate near the meridian transit was 25.5 0.4 counts s^-1, after subtracting the background count rate.

Fig. 1. The count rate profile of Cyg X-1 obtained from a balloon flight carried out on 1992 April 5/6 using the XMPC detector. The source and background observations are marked in the figure. The decrease in the source count rate after the meridian transit is due to increase in the air mass. During the initial part of the observation the source was scanned across for aspect calibration. The increase in the count rate around 02:50 is due to a gamma-ray burst.

2.1. Detector response matrix

In order to characterize the energy spectra of X-ray sources it is necessary to have a detailed knowledge of the response of the detector. Fig. 2 shows the response of a single anode of a detector for fluorescent and X-rays from Tb. Various photo-peaks and escape peaks can be seen. The energy resolution of the detector is 9.5% at 44 keV and it shows a weak dependence on energy. As can be seen from the figure, xenon gas has a very large fluorescence yield and to take care of such effects, we have generated response matrix of the XMPC using a Monte Carlo routine. Inputs to this program include detector characteristics such as partial pressure of xenon gas in the detector, layer-wise conversion gain (relation between output pulse height and input energy), energy resolution of the detector, geometry of the detector, and characteristics of event selection logic such as thresholds of K-band, thresholds of upper and lower level discriminators which defines energy range of acceptable events. In the simulation program, each photon is tracked assuming random incidence at the detector surface. Position and layer number for the interaction is then calculated. Possibility of emission of K-X-ray is considered and layer number corresponding to interaction of the K-X-rays is calculated. Taking into account gain and energy resolution of the detector, the output pulse height is evaluated. Escape gating technique is taken into consideration by calculating pulse height only for non-K event in case of simultaneous events. Depending on the type of interaction and the layer in which the interaction took place, a layer identification number is generated. Simulation is done typically 10000 times for each energy bin of width 1 keV, between 10 and 130 keV. The corrections due to absorption in the air and the window material are calculated numerically.

Fig. 2. Response of the detector for characteristic X-rays from Tb radioactive source. Various peaks including photo peaks and escape peaks are marked.

Various inputs to this routine, e.g. layer-wise gain and energy resolution of the detector, various event selection logic thresholds etc., are computed by calibrating the detectors with various radioactive X-ray sources of known energies. The observed response of the detector to mono-energetic X-rays is compared with the simulated one. It was found that one of the detectors had better uniformity of gain and overall energy resolution (10% FWHM at 60 keV) and the observed spectra from calibration sources were found to show good agreement with the predicted spectra correct to about 2% in each spectral bin.

Fig. 3 shows the response of the second layer of the detector A for Am²⁴¹ radioactive source. Data points with error bars correspond to the observed pulse height distribution and the histogram corresponds to the predicted pulse height distribution obtained by convolving a Gaussian line at 60 keV with the Monte Carlo simulated response of the detector. Am²⁴¹ photo-peak at 60 keV and two escape peaks at 26 and 29.9 keV corresponding to the escape of xenon and X-rays can be seen in the figure.

Fig. 3. Response of layer 2 of detector A for Am241 source. Data points with error bars correspond to the observed pulse height distribution and histogram corresponds to the predicted pulse height distribution for a Gaussian line at 60 keV, convolved through the Monte Carlo simulated response of the detector.

2.2. Hard X-ray spectrum of Cyg X-1

We have selected data from the continuous tracking of Cyg X-1 near meridian transit, from one of the detectors which has better spectral response, for spectral fitting. The spectral files are generated for separate layers and data for each layer are re-binned in 42 channels. We have used the XSPEC package (Arnaud 1996) for the spectral fitting. Simultaneous spectral fits for the three layers were carried out. A power-law with a photon index () of 1.62 0.07 (90% confidence limits) gives an acceptable value of of 129 for 125 degrees of freedom (dof). The observed 20 - 100 keV flux is 1.3 10^-8 ergs cm^-2 s^-1, which corresponds to a luminosity of 0.9 10³⁷ ergs s^-1 (for a distance of 2.5 kpc). The observed flux density at 100 keV is 6.4 10^-4 photons cm^-2 s^-1 keV^-1. The observed count spectra obtained from layer 1 and layer 2 and 3 summed together, are shown in fig. 4. The best fit power-law spectrum, convolved with the detector response is shown as histograms. The residuals to the fit are shown in lower panel, as contribution to .

Fig. 4. Observed count rate spectra from Cyg X-1 obtained from XMPC, shown separately for layer 1 and bottom 2 layers. The best fit power-law model with photon index () of 1.62, convolved through the detector response is shown as histograms. The residuals to the model fit are shown in the lower panel of the figure as contribution to the .

We have also made an attempt to fit other spectral models to the XMPC data. Thermal bremsstrahlung model gives a of 126 for 125 dof, with temperature 126 keV. The CompST model gives of 123 for 124 dof, with electron temperature of keV and the optical depth of (90% confidence errors for 2 free parameters). Due to limited bandwidth of the data we cannot distinguish between any of these models.

In order to know the exact nature of the continuum, which is not possible due to the limited dynamic range of the XMPC data, we have extended the dynamic range by combining XMPC data with EXOSAT ME Argon and GSPC data spanning 2 - 20 keV band and OSSE data over 50 - 500 keV. The source normally remains in the state, occasionally going to the super-low () or the flare () state (Ling et al. 1987). The flux density at 100 keV derived from power-law spectral model for XMPC data agrees well with the average value derived from 10 observations from the SIGMA detector (6.37 10^-4 photons cm^-2 s^-1 keV^-1) made between 1990 March and 1992 March, when the source was in state (Laurent et al. 1993), and hence we can conclude that the source was in the state during our observations. We have selected EXOSAT and OSSE data pertaining to state of Cyg X-1 and carried out simultaneous fit to EXOSAT, XMPC and OSSE data spanning an energy range of 2 - 500 keV.

© European Southern Observatory (ESO) 1998

Online publication: February 4, 1998
helpdesk.link@springer.de