THE GAMMA-RAY BURST CATALOG OBTAINED WITH THE GAMMA-RAY BURST MONITOR ABOARD BeppoSAX

In the last 10 years a big step forward has been accomplished in gamma-ray burst (GRB) astronomy. A key role in this advancement has been played by the BeppoSAX satellite (Boella et al. 1997). Its capability of promptly and accurately localizing GRBs and the possibility of following them up with highly sensitive X-ray telescopes on board allowed the discovery of the X-ray afterglow emission and, with ground telescopes, of the optical/radio afterglow from GRB sources (Costa et al. 1997; van Paradijs et al. 1997; Metzger et al. 1997; Frail et al. 1997). In this way the distance scale of long (>1 s) GRBs could be eventually established and important information on the GRB sources, their emission mechanisms, and their environments could be derived.

The two BeppoSAX instruments that allowed the prompt and accurate GRB localization were the Gamma-Ray Burst Monitor (GRBM) and the two Wide Field Cameras (WFCs), the first with the role of recognizing the occurrence of a GRB and second with the role of localizing it in the case the event direction occurred in their field of view (FOV) (40° × 40° full width at zero response) and was above the instrument sensitivity in their operative energy range (2–28 keV, Jager et al. 1997). Only a handful of GRBs (50; see Frontera 2004) was detected with both instruments, while the GRBM alone detected 1082 GRBs, a number similar to that of GRBs included in the third BATSE (3B) burst catalog (Meegan et al. 1996). In this paper, we present the catalog of these GRBs with their main observational and statistical properties.

The BeppoSAX GRBM was part of the high energy experiment PDS (Phoswich Detection System; Frontera et al. 1997), being its anticoincidence (AC) shield. It was composed of four independent detection units, made of slabs of CsI(Na) scintillators, forming the four sides of a square box (see Figure 1). Each slab was 1 cm thick and had a geometric area of 1136 cm², with an open FOV. The light scintillation produced in each unit was viewed by two photomultipliers. A calibration source made of Am²⁴¹, embedded in a NaI(Tl) crystal, allowed to monitor the instrument gain. The gain could be varied by changing the high-voltage (HV) supply of each GRBM unit.

Zoom In Zoom Out Reset image size

The electronics associated with the GRBM was a simple spectroscopic chain in which signals from each of the four detection units between two analog thresholds, after being multiplexed, were analog-to-digital converted by an analog-to-digital converter (ADC). By telecommand the low-level threshold (LLT) could be varied in the nominal energy range from 20 to 90 keV (16 steps), while the upper level threshold (ULT) could be varied between 200 and 700 keV (8 steps). The signals detected by each unit between LLT and ULT (good events) and above the AC threshold (ACT) were continuously counted with a 1 s integration time and stored in the onboard memory. The ACT could be selected between 100 and 300 keV (8 steps). The GRBM electronic unit included an onboard GRB trigger (see Section 2.1). If the onboard trigger condition was satisfied for at least two detection units, then the following high time resolution profiles for each of the GRBM units were stored on board and then transmitted to ground:

• 1.  
  7.8125 ms time profiles from the trigger time back to 8 s before;
• 2.  
  0.48828125 ms time profiles from the trigger time to 10 s later;
• 3.  
  7.8125 ms time profiles from 10 s after the trigger time for 88 s.

In addition to these data, for each detection unit, the following data were continuously transmitted:

• 1.  
  1 s count rates between LLT and ULT;
• 2.  
  1 s count rates above ACT (in phase with the previous ones);
• 3.  
  240 channel spectra of the signals between LLT and ULT, integrated over 128 s (synchronous integration, in phase with the 1 s count rates).

Further details about the GRBM instrument can be found elsewhere (Amati et al. 1997; Feroci et al. 1997; Costa et al. 1998).

2.1. Onboard and On-ground GRB Trigger Logic

The determination of the GRBM response function, both in direction and energy, was performed with Monte Carlo techniques and tested with ground calibrations performed before the BeppoSAX launch and with various observations of the Crab Nebula. It was also cross-calibrated with the BATSE experiment (Fishman et al. 1994).

The GRBM response function derived via the Monte Carlo code (Rapisarda et al. 1997) was obtained using the Monte Carlo N-Particle Transport Code (MCNP), version 4.2, released by the Los Alamos National Laboratory (LANL; for the current MCNP version see Brown 2002). The code allows to transport photons, neutrons, and electrons from 1 keV to 100 MeV through matter. Every kind of three-dimensional geometry could be described by defining separate cells with composition and density chosen by the user. Photon interactions were treated very accurately and the entire satellite was modeled with a high degree of detail.

The model was tested using the GRBM ground calibrations performed before the flight when the instrument was integrated in the spacecraft (Amati et al. 1997). Preliminary results of these tests were reported by Calura et al. (2000).

The response function, obtained with the Monte Carlo code, was determined as a function of the energy and direction of the incident photons. Using as reference frame the satellite local frame with equatorial plane perpendicular to the axis of the Narrow Field Instruments (z-axis), and using the z-axis as polar axis, a grid of 576 directions was determined: 36 azimuthal angles ϕ_k in the range 0°–360°, in which the zero value corresponds to the azimuth of the axis of the detection unit No. 2, and 16 polar angles θ_j in the range −70° < θ < +80°. Ten different intervals of the photon energy E_n (n = 1, 11) in the 30–1000 keV were also simulated with logarithmic steps. For each grid point we determined the expected number of counts N_i(E_n, θ_j, ϕ_k) detected in the 40–700 keV and >100 keV channels by each of the four detection units (i = 1, 2, 3, 4 for detections by the units Nos. 1, 2, 3, 4 in the 40–700 keV channel, and i = 5, 6, 7, 8 for detections by the same units in the >100 keV channel), for an input photon of energy E_n incident from the direction (θ_j, ϕ_k).

A further refinement of the grid was obtained by interpolating with cubic splines N_i(E_n, θ_j, ϕ_k) in both energy and direction. The grid energy was refined up to 1 keV steps and the direction (θ_j, ϕ_k) was refined up to (1°, 1°) steps. Assuming a photon spectrum I(E) = Kf(E, α), where α are the model parameters for a photon beam coming from the direction (θ_j, ϕ_k) we expect to detect from each detection unit the following count rate in the above energy channels by each detection unit:

$$\begin{equation} C_i(\theta _j,\phi _k,{\bf \alpha }, K) = \sum _{n} \Delta E_n K f(E_n,{\bf \alpha }) N_i(E_n, \theta _j, \phi _k). \end{equation} \tag{ 1 }$$

The best estimate of the GRB arrival direction (θ, ϕ) and of the parameters α of the assumed photon spectrum is determined by minimizing the following χ² statistics (for a power-law (PL) photon spectrum, see Calura et al. 2000):

$$\begin{equation} \chi ^2 (\theta _j,\phi _k, K, {\bf \alpha }) = \sum _{i=1}^{8}\ \frac{1}{{\sigma ^2_i}}\ \left(n_i - \frac{n\,C_i(\theta _j,\phi _k, K, {\bf \alpha })}{C(\theta _j,\phi _k)} \right)^2, \qquad \end{equation} \tag{ 2 }$$

where n_i is the measured counting rate measured by the unit i as defined above, σ²_i is the variance of n_i, n = ∑⁸_{i = 1}n_i, and C(θ_j, ϕ_k) = ∑⁸_{i = 1}C_i(θ_j, ϕ_k, K, α).

Actually, given that we had only two energy channels, we were constrained to use as input model f(E, α) a simple power law (E^−Γ). The best estimate of the parameters (θ, ϕ, K, Γ) was that which minimized the χ² in Equation (2). In case more solutions were found for (θ, ϕ), no localization or spectrum was included in the catalog.

4.1. In-Flight Settings

4.2. Background Estimate and its Subtraction

After background subtraction, GRBM data were analyzed using the response matrix described above, which had also been converted to the Flexible Images Transport (FITS) format to be used with standard spectral analysis software packages like XSPEC.

The goodness of the response matrix was verified in flight with the Crab Nebula observed via source occultation from the Earth (Guidorzi et al. 1998) and with cross-checks with BATSE results obtained for a GRB-selected sample by Kaneko et al. (2006).

The Crab flux and spectrum were derived using both the 2 channel ratemeters and the 240 channel spectra. We found the spectral parameters to be consistent with the corresponding values found with other experiments in hard X-/soft gamma rays, i.e., a photon index of about 2.2 and a 100 keV flux density of about 60 × 10⁻⁵ photons cm⁻² s⁻¹ keV⁻¹, with reduced χ² values of about 1.3 for about 13 degrees of freedom (dof). By fixing the photon index at the commonly adopted X-ray value of 2.1, we obtain a normalization at 1 keV of 9.64 ± 0.49 photons cm⁻² s⁻¹ keV⁻¹, which is fully consistent with the classical value of 9.7 photons cm⁻² s⁻¹ keV⁻¹ (Toor & Seward 1974).

Cross-checks with BATSE results were performed with a sample of 46 strong GRBs detected by both experiments and arriving from various directions. The results show that the ratio of the 40–700 keV fluences derived from the GRBM response function with those derived using spectral law, parameters, and time integration given by Kaneko et al. (2006), is distributed according to a Gaussian centered at about 0.8 with a standard deviation of 0.3. The deviation of the ratio's mean value from 1 and the distribution spread are a likely consequence of the uncertainties in the response functions and GRB directions, and of the differences in integration time and spectral models adopted by us and by the above authors.

The uncertainty in the knowledge of the GRBM response function is on average of the order of 10%. This uncertainty is mainly due to the error in the knowledge of the flux of the calibration sources and the errors in the calibration data. This uncertainty was added in quadrature to that on the GRB count spectra before performing the fits.

The GRBM sky exposure, that is the fraction of time above the horizon in which a given sky direction was exposed to the GRBM, is shown in Figure 2 in celestial coordinates. As can be seen, the celestial poles were never blocked by the Earth and thus they were continuously visible by the GRBM, while the other sky directions, due to the BeppoSAX orbit, were monitored for a shorter time. The dependence of the sky exposure on the right ascension was very small, while that on the declination is shown in Figure 3, after averaging the sky exposure over the right ascension.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In Table 2, we show the GRBM catalog of GRBs. A preliminary version of the catalog can be found elsewhere (Guidorzi 2002), where the peak fluxes and fluences were given in counts (no correction for efficiency). The catalog includes 1082 events. The parameters associated with each event include the best GRB direction in Galactic and equatorial coordinates, two different estimates of GRB time duration, i.e., the classical T₉₀ (time during which the burst-integrated counts increase from 5% to 95% of the total counts; Kouveliotou et al. 1993) and the time duration T_det (see below), the integrated time T_a during which the burst-count rate is detected above a 2σ level, the 40–700 keV fluence, peak flux, the spectral hardness, the GRBM units used for the parameter determination, and the binning factor with respect to the default count accumulation time (1 s for long GRBs, 7.8125 ms in the case of short GRBs). For different reasons not all the parameters were determined for the entire GRB samples (see details in Table 3). Details on the data reported and the methods adopted for the derivation of the parameters reported in the catalog are given below.