HKS Aerogel Cerenkov counter

An aerogel Cerenkov detector will be used to separate pions from kaons in the momentum range 1.08 $\sim$ 1.32 GeV/$c$. The expected pion and electron singles rates are 800 kHz and 2.6 MHz, respectively. With an electron-hadron coincidence time window of 50 ns, this results in a coincidence $e-\pi$ trigger rate of $\sim$100 kHz. The aerogel Cerenkov veto has to reduce this rate to a few hundred Hz. Allowing for a safety factor of 10, the aerogel Cerenkov should have inefficiency less than 0.1%. In order to keep the individual PMT rates low and also allow for more sophisticated trigger schemes that distinguish between multiple tracks, the detector will be segmented across the focal plane of the spectrometer.

The aerogel Cerenkov system consists of three identical detector planes with a radiator length of 5 cm for each. This will be achieved with 5 layers of 10 mm thick tiles. Each plane will further be segmented into 7 optically separated diffusion boxes. Five layers of 4 by 2 tiles will cover an area of $460\times 230$ mm$^2$. The aerogel tiles will rest on one of the side walls and be held in place by thin nylon wires. All internal walls, including the one behind the aerogel, will be covered with two layers of Millipore filter paper as a diffuse reflector. Separating walls between two segments will be two layers of Millipore paper. Two 5'' photomultipliers on each segment will collect the Cerenkov light. Detector entrance and exit windows will be made of thin light tight foil. Each detector plane of 7 segments will form one gas tight box with an internal volume of $1620\times 460\times300$ mm$^3$ (223.6 liter or 7.9 ft$^3$). During operation (maybe even storage) the boxes will be continuously flushed with UHP/Zero air at roughly 1-2 ft$^3$/h. A schematic layout of the three detector planes is shown in Figure 23.

As an aerogel radiator, SP-50 from Matsushita Electric Works with a refraction index of $n=1.05$ will be used. This material is hydrophobic and therefore is free from the problems that have been associated with absorption of water in older types.

As photomultipliers, Burle 8854, or Hamamatsu R1250, or Photonis XP4572B/D1 are considered. The Burle 8854 have been successfully employed for the Hall C aerogel and the Hall A A1 detector. The Hall A A2 detector, which also uses the Matsushita SP-50 aerogel, employed less expensive Photonis XP4572B/D1.

The expected performance of the proposed geometry has been simulated with the help of a Monte Carlo program provided by Doug Higinbotham (NIM 414, 332 1998). This program reproduces the performance of several existing aerogel detector systems reasonably well. With an assumption that the absorption length of the new aerogel material from Matsushita by far exceeds the scattering length, i.e. absorption is negligible, this Monte Carlo program predicts 26.6 photoelectrons for the Hall A Cerenkov detector A2 for pions with 1.741 GeV/$c$ momentum. This is in very good agreement with the experimental value of 27 to 28 photoelectrons (M. Coman, private communication). For the proposed geometry of the HKS detectors, an average number of 16.8 photoelectrons per segment is predicted at a momentum of 1.2 GeV/$c$. The response is fairly uniform over the active area of the detector (see Fig. 24a& b). As expected, the predicted signal from an individual tube varies with the vertical distance from the tube (see Fig. 24c& d). From the simulated distributions, the veto efficiency (or inefficiency) can be extracted. Fig. 25b shows the fraction of pions which failed to trigger the veto as a function of a threshold set on the sum of the signal from the top and bottom tube of an individual detector segment. A threshold of 8 p.e. still results in a veto inefficiency of less than 2%. Fig. 25c shows the inefficiency for a coincidence trigger between top and bottom tube as function of the individual thresholds. A threshold of $N_{p.e.}^{top}>2$.and. $N_{p.e.}^{bottom}>2$ results in a veto efficiency for one segment of 97.8%. Fig. 25a shows the effect of this cut on the photoelectron distribution.

As mentioned above, the entire detector system will consist of three layers. Three basic trigger schemes are possible for this configuration, 3/3 (three out of three, or all three in coincidence), 2/3, and 1/3. With an individual threshold set between 1 and 2 photoelectrons and an efficiency of 99% per plane (see Fig. 25c) this results in veto efficiencies for 3/3, 2/3, and 1/3 of 97%, 0.03%, and 0.0001%, respectively. The most desirable scheme would be a 2/3 coincidence. Even with an unvetoed $e-\pi$ coincidence rate of $10\times 100$ kHz this results in a trigger rate of roughly 300 Hz.