$\pi$2 beam

The performance test of a prototype of the Belle ECL detectors was carried out using the $\pi$2 beam line at KEK in an energy range from 0.25 to 3.5 GeV [59]. An array of 6 $\times$ 5 CsI($Tl$) counters with the same mechanical assembly and readout electronics as those of Belle ECL were used to measure the energy and position resolution for electrons and the e/$\pi$ separation for two sets of matrix configurations: one corresponded to the center and the other to the edge of the barrel calorimeter. Fig. shows the schematic top view of the two matrix configurations. The whole matrix was placed on a movable table controlled by an online computer.

Figure: Top view of two matrix configurations: (a) normal array and (b) staggered array.

The energy resolution was measured by using electrons which impinged $\pm$2 cm from the center of the matrix by summing the energy deposit weighted by the calibration constant of each crystal. The summation was carried out for a 3 $\times$ 3 matrix (9 crystals) and a 5 $\times$ 5 matrix (25 crystals). The total energy deposit was then scaled event by event by the momentum measured by the spectrometer in order to compensate for the spread of beam momentum. The energy resolution measured for the two matrix arrangements (normal and staggered) is shown in Fig. as a function of beam momentum together with the measurements of CLEO II [65] and Crystal Barrel [66]. A Monte Carlo simulation using GEANT 3.15 reproduced the general behavior of the energy resolution but tended to predict a slightly better overall resolution. This may be explained by the nonuniformity of the light collection in the crystal, which was not included in the simulation.

Figure: Energy resolution as a function of beam energy. Results from a GEANT simulation are shown by the dashed curve (9 blocks) and the solid curve (25 blocks). Also plotted are the results of CLEO II (9 blocks) and Crystal Barrel. The dotted curve shows the contribution from electronics noise for 25 blocks.

The impact point of an electron on the matrix was calculated from the position of each crystal summed with the weight of its energy deposit. Position resolutions were determined from the correlation between the impact position and the position obtained by the drift chamber system to be 3.6 mm for 2.0 GeV/c electrons and 5.8 mm for 0.5 GeV/c electrons.
The difference of the energy deposit in the calorimeter can be used to distinguish electrons from charged pions. Fig. shows the energy deposit summed over 25 crystals for 1 GeV/c electrons and pions injected near to the center of the matrix. We see a difference in the spectra for $\pi^+$ and $\pi^-$ due to the difference in cross sections. We define the probability for a pion to be misidentified as an electron by $N_{mis}/N_{tot}$, where $N_{mis}$ is the number of pion events with an energy deposit in the electron energy region and $N_{tot}$ is the total number of pion events. The electron region is defined as $\pm 3 \sigma_e$ around the peak of the electron energy deposit, where $\sigma_e$ is the energy resolution of the electrons. The results of an e/$\pi$ separation for positive and negative pions are shown in Fig. for energies of 0.5$-$3 GeV and for the normal and staggered arrays. The misidentification probability is found to be less than 1 % above 2 GeV/c.

Figure: Distribution of the energy deposit by electrons (red histogram), by positive pions (green histogram) and by negative pions (blue histogram) at 1 GeV/c.

Figure: Probability to misidentify a pion as an electron (a) for the normal array and (b) for the staggered array.