Astron. Astrophys. 329, L37-L40 (1998)

2. Observations and data analysis

The BeppoSAX Wide Field Camera instrument (WFC, Jager et al. 1997) consists of two identically designed coded aperture cameras. The main characteristics of each camera are summarized in Table 1. The field of view (FOV) of this instrument is the largest of any flown X-ray imaging device. This implies an exceptional capability to find short duration and weak transient events. With regards to the sensitivity it is important to realize that an imaging device based on the coded aperture principle has one very basic difference with direct-imaging devices such as X-ray mirror telescopes: there is cross-talk between FOV positions located much further from each other than the angular resolution. In the case of WFC, there is a degradation in sensitivity to any sky position within 20 from a bright source. This has a relatively large impact on observations of the crowded galactic center field where the sensitivity is about two times less than at high galactic latitudes far from bright sources.

Table 1. Main characteristics per BeppoSAX-WFC camera

Detector data contain an image of the sky which is coded with the aperture pattern. The reconstruction of the sky image involves an algorithm whose basic component is a cross correlation of the detector data with the aperture pattern (Jager et al. 1997). This algorithm is optimum for point sources but not necessarily for diffuse sources. In the case of WFC the reconstruction can be performed with an arbitrary time and photon energy resolution within the limitations given in Table 1. The position and intensity of any point source is determined by modeling through a point spread function (PSF). The full-width at half maximum of the PSF is smallest on-axis at about 5 arcmin. The point source location accuracy is an order of magnitude better (e.g., In 't Zand et al. 1997).

NGC 6652 is located 11.5 degrees from the galactic center. Therefore, it was often observed as part of a monitoring campaign on the region around the galactic center. During the fall of 1996 a total coverage of about s was obtained on this part of the sky during about 24 days. During the spring of 1997 this was about s during about 15 days. A small level of emission was detected from NGC 6652 in a combination of all data of 3.0 0.4 mCrab in 2 to 8 keV (1 Crab unit is the intensity of the Crab X-ray source whose flux is between 2 and 8 keV).

Routinely, all data of each camera are systematically searched for burst phenomena by analyzing the time profile of the total detector in the full bandpass with a time resolution of 1 s. Enhancements beyond 5 above a background modeled to vary linearly with time are searched for on time scales of up to 48 s. For the observations discussed here this implies an on-axis sensitivity of about 0.6 Crab units (2-25 keV) in 1 s to 0.1 Crab in 48 s. If a burst is found, a reconstructed sky image is generated for the time interval when the burst occurs, another sky image is generated for a long time interval (usually ten times as long as the burst time interval) right before the burst time, the latter image is normalized and subtracted from the former and this image is searched for point sources with intensity increases equivalent to the increase found in the time profile of the detector. The image subtraction is not necessary but facilitates quick identification in case there are many point sources in the FOV (like is the case in the observations relevant to NGC 6652). So far, over 300 bursts were found and identified in all observations (Cocchi et al. 1997). Two of these were found at a position consistent with NGC 6652. Fig. 1 presents the error boxes of the bursts. The times, peak intensities and best-fit positions of the two bursts are given in Table 2.

Fig. 1. Celestial maps of 68% confidence level regions of burst 1 and 2 for equinox 2000.0. The crosses in both maps indicate the position of the X-ray source in NGC 6652 according to Johnston et al. (1996) and the asterixes the best fit positions for the BeppoSAX-WFC data

Table 2. Characteristics of two bursts

Both bursts are relatively weak, the signal-to-noise ratio of the image being around 6. Consequently, the extraction of meaningful time-dependent spectral decoded information is not possible. As an alternative, we analyze time profiles directly of the detector for the part illuminated from the sky position of NGC 6652 in two photon energy bands. By imposing less constraints on the time profile one, on the one hand, preserves the little available statistics but, on the other hand, ends up with a time profile that includes the combined flux from all sources that illuminate the same part of the detector. We regard the latter disadvantage as not important because the bursts are a coherent and clearly recognizable signal which can only be due to a single source of emission and the identification of the whole burst with the source is unambiguous. Figs. 2 and 3 present the time profiles for the bursts.

Fig. 2. Time profile of the first burst, per each of two bandpasses and for the complete WFC bandpass. The bin time is 2 s. The smooth curves are exponential models for the appropriate time profiles (see text)

Fig. 3. Time profiles of the second burst

The time profiles were tested for the evidence of spectral changes. They were modeled with an exponential decay function with 4 parameters: peak intensity, onset time of exponential, e-folding decay time and background level. The results for for both energy ranges and the peak intensity in 2-8 keV are given in Table 2. In both cases the ratio of the decay time between the upper and lower energy band is 0.4 0.2. We regard this as good evidence for the presence of spectral softening in bursts from NGC 6652.

The small statistical quality of the data do not permit an analysis of the time-dependent decoded (background-subtracted) spectrum. This is somewhat different for the average spectrum over both bursts. We fitted a number of simple spectral models to the spectrum. The models are described by 4 parameters and only the normalization was allowed to differ between both bursts. All of the models fit the data equally well with a reduced value of 0.9 for 58 independent PHA bins. The data does not allow to single out a best fit model. Nevertheless, if we assume that a black body model applies with absorption by cold interstellar matter according to the model of Morrison & McCammon (1983) the temperature is reasonably well constrained to  keV. If the black body radiation is isotropic and the distance is 14.3 kpc, we find a bolometric luminosity averaged over the first 16 s of each burst of 2 10³⁸  ergs s^-1 and a radius of the emitting region of  km.

© European Southern Observatory (ESO) 1998

Online publication: December 8, 1997
helpdesk.link@springer.de