High Energy Astrophysics

A few 100 MeV gamma ray only produces a small cascade. This cascade steel produces some Cherenkovlight. But the Cherenkov signal is so weak that an individual shower could not be detected and identified. For example a 1GeV electromagnetic shower would contribute by less than 1 Cherenkov photon to the image recorded by the Whipple telescope which is 10m in diameter. Let us now consider a large number of few 100MeV gamma rays rather than an individual gamma ray. Let also assume that all the gamma rays arrive on the top of the atmosphere in a relatively short interval of time as in a burst. Each gamma ray will produce a small shower which will generate a small Cherenkov signal. The cumulated Cherenkov signal from the many electromagnetic showers can be detected and the recorded image presents some very specific characteristics both in its geometrical and time structure.

The image is centered on the burst direction and is center symmetric in time and intensity with a specific radial distribution extending over typically 1 degree. An infinitely short gamma ray burst would generate a Cherenkov signal on the ground with duration of the order of 40ns while individual showers produce signals with duration shorter the 20ns. This combined with the geometrical properties of the image allow an absolute identification of the potential events of that type. Only the atmospheric scintillation produced by the Ultra High Energy Cosmic Ray Showers is known to produce signal in the time range from 100ns to 10micro second. But the image is not center symmetric and the time structure presents a gradient. The maximum time scale for which this technique can be applied results from the random night sky light background and we have estimated that burst with duration up to more than 10 micro seconds could be interestingly studied.

Since their discovery in the 70's, gamma ray bursts have been intensively observed. On board of the CGRO satellite the BATSE detector was dedicated to the study of these still mysterious phenomena. The BATSE trigger integration time is 64ms and below ~10ms there is no sensibility from any detector. Among the many bursts recorded by BATSE, some were found to present very short time structure, down to 20 micro-second. This by itself justifies the interest for very short gamma ray bursts and their search is driven by an exploratory motivation. The SGARFACE experiment should be even more sensitive to those very short bursts than a space detector wich has to receive at least 10 gamma rays on it's 0.8 square metters collecting area to identify a burst. As for the volume of space being probed, the very good sensitivity of SGARFACE compensates for it's reduced field of view.

Possible sources of very short gamma ray burst are the primordial black holes arriving at the final phase of their evaporation today. Upper limits on the existence of primordial black holes are the strongest constraints on the smoothness of the universe in the very early times (fraction of pico seconds). Observation of a primordial black hole would not only bring important information on the early universe but would also be a first experimental proof of the evaporation mechanism and would carry some information on the very high energy physics which drives the final rate of evaporation. If no signal is detected in two years of observation with SGARFACE and GLAST, upper limits on the existence of Primordial Black Holes would be improved by two order of magnitude for all the time scales on which they are expected to explode according to various models. The figure bellow indicates such upper limits in the asumption of a 6.0E34 ergs gamma-ray burst independently of the time scale.

The SGARFACE experiment is described in this paper for the 2001 ICRC. It should be operated with out interfering with the standard TeV astronomy observation. This requires PMT signals to be sent to both the SGARFACE and the standard electronics. We have seen that the image is very extended and we can work with the PMT signals summed per bunch of 7 without any important information loss. Our first module will take the signals, send them to the standard electronics and sum them per bunch of seven. This module still is being developped. Here is our current schematic and here is a memo about it(postscript). The signal spliting is passive. Attenuation of TeV signals is less than 10%.

The time scale of the signals are not known a priori. The signal will be digitized and numericaly integrated over various time scales. A threshold depending on the time scale will be applied on the integrated signal. This is achieved by Xilinx reprogrammable FPGA chips in a 16 channel module described in this paper for the ICRC 2001. Here are the picture of the top and bottom of our prototype board before it got stuffed.

The image being quite extended it is possible to apply a high fold coincidence logic in order to reduce the triggering on noise fluctuations. We are planning on developping a coincidence unit with 64 inputs and pattern recgnition capability.