Monitoring des binaires X et des novae avec le Burst Alert Telescope à board du satellite Swift

Monitoring des binaires X et des novae avec le Burst Alert Telescope à board du satellite Swift

The detector plane

The BAT detector plane is composed of 32768 pieces of 4 × 4 × 2 mm CdZnTe (CZT), forming a 1.2 × 0.6 m sensitive area. For electronic control, event data handling, and fabrications reasons, these pixels are grouped following a hierarchical structure. Groups of 128 detector elements are assembled into 8 × 16 arrays, each one connected to a different readout electronic circuit. Detector modules, each containing two such arrays, are further grouped by eights forming 16 blocks. For fabrications reasons, there is a gap between two adjacent pixels: it was restricted to be an integer multiple (2 or 3) of the basic pixel dimension, so that the image reconstruction process could easily handle it. Considering that a coded-mask telescope cannot derive the direction of any photon arriving at the detector, such a hierarchical structure is optimised to allow BAT to tolerate the loss of individual pixels, detector modules, and even whole blocks without losing the ability to detect bursts and determine locations. There is, of course, a loss in burst-detection and the survey sensitivities. The Fringe Shield A graded-Z Fringe Shield, composed of Pb, Ta, Sn, and Cu, is located on the side walls between the Mask and the Detector Plane and under the Detector Plane in order to reduce the event rate in the detector plane. Due to its performance, the isotropic cosmic diffuse flux and the anisotropic Earth albedo flux are reduced by 95%.

BAT operating modes

 BAT can work in two distinct operating modes: the so-called “Burst mode” and “Survey mode”. Most of the BAT time is spent in Survey mode, waiting for a GRB occurring in the FOV. Events are accumulated in the detector plane and searched by the on board algorithm for increases in the count rate over a range of time scales. If no special triggers are found, all these events are binned into eighty energy bins and integrated over ∼5 minutes (survey data) because there is not enough on board storage or down-link capacity to send them all to the ground. Instead, when the trigger algorithm is satisfied, the  Burst mode is enabled and photon-by-photon data are produced and sent to the ground. 

Burst mode 

The algorithm on board BAT, which search for a GRB, is composed of two processes. First, it looks for excesses in the detector count rate above the expected background and constant sources count rate. During a 96-minute low Earth orbit, the detected background rates can vary by more than a factor of two. Moreover, the duration of the gamma-ray emission from GRBs ranges from milliseconds to minutes. For such reasons the triggering algorithm was written in order to be able to extrapolate the background and compare it to the measured count rate over a variety of timescales and in several energy bands. If at least one trigger is found, the algorithm switches to the second process. Another independent trigger criterion is implemented to search for slow rising GRBs and transients: it consists in an image analysis of the detected count rate performed every 64 sec. Such images are scanned for point sources which are then cross-checked against an on board catalog. The detection of any new source produce a GRB alert, whereas any known source above a given level initiates the interesting-source response procedure. The second process begins whenever the trigger algorithm detects a count-rate excess in the detector. In this case the data are analysed to discover if this excess is due to a GRB. The system on board extracts source and background data based on the energy range and time intervals flagged by the trigger. Such data are converted to a sky map which is then searched for excesses. If a significant excess is detected in a position that not match any source of the on board catalog, then a GRB is declared. The imaging process is fundamental to confirm the reality of the previous count rate detection. Thus, in order to eliminate false triggers, the trigger threshold was set to an opportune level. Finally, when a GRB alert is produced, the system on board decides if it is worthy to begin a spacecraft slew to point the source, depending on the merit of the burst and the observing constraints (e.g. the Earth limb does not cover a large fraction of the FOV). When a GRB is declared, several products are constructed, the most significant of these being a dump of all event data included in the trigger time, with a total duration of about 10-15 minutes. Each event is tagged with an associated time of arrival, detector number and energy. This allows event data to be used for the extraction of light curves of all the sources in the FOV with different timescales and energy bands. These data can also be used to extract spectra of a given source. The BAT also produces several products all the time, regardless of whether there is a GRB or not. These are typically various array rates, spacecraft attitudes, housekeeping values, and trigger diagnostics. Survey mode Most of the time is spent by BAT waiting for a GRB occurring in the FOV (Survey mode). Photons interacting with the detector are processed (events) and then are tagged with an associated time of arrival, detector number and energy. Such information is stored on board in a memory buffer which may contain ∼10 minutes of data (depending on the actual count rate). If the burst trigger algorithm described above fails, the event data from the array are collected on board into Detector Plane Histograms (DPHs). DPHs are three dimensional histograms: for a given buffer set, every cell contains the number of events received in one of the 32768 pixels of the detector plane and in one of 80 energy channels. The energy bin widths are variable from one energy bin to the next, but have remained always the same since the Swift launch. Figure 2.5 shows a representation of a survey DPH. Such histograms are accumulated over a typical 5 minutes time interval and then stacked as independent rows in a “Survey” data file. In some cases the duration of a DPH row can be longer or shorter depending on operational reasons (e.g., telemetry reduction or to get diagnostic information about the instrument). Also, DPH row integration times are truncated whenever the spacecraft begins a slew or enters the South Atlantic Anomaly (SAA).