In the fall of 1989, the pion beta collaboration undertook a development run
during which a 
parasitic beam was stopped in an
active
plastic scintillator target in the 
area at the Paul Scherrer Institute.
The active target was viewed by two sets of detectors: on one side, an array
of seven large CsI modules and a NaI detector on the other side. There were
thin plastic veto counters in front of the two sets of detectors which
subtended matching solid angles at 
. The run took place under severe
time constraint due to the pending shutdown of the PSI Ring accelerator and the
late delivery of the CsI modules from Bicron Corporation. These constraints
limited the running time and consequently the counting statistics as can be
seen in figure 
which summarizes the results:
Figure: The results of the
development run in the fall of 1989. A parasitic beam
of 
was stopped in an active target viewed on the
sides by an array of CsI and a NaI detectors. The top figure is the relative
timing between a CsI block and the target. The central figure is the total
energy (CsI plus NaI) detected within a gate, 80 ns long and opened 10 ns
after the 
stop. The bottom figure is the central one cut on neutrals
in the plastic veto detectors in front of the CsI and NaI crystals. A
peak emerges but the statistics is low due to time constraints
during the data taking.
Figure (a) shows the relative timing between a CsI block and
the active target. The measured timing resolution was 
FWHM.
The CsI modules were readout with EMI 9821QB photomultiplier tubes with quartz
window and a nominal rise time of 
. In a previous test
conducted at the University of Virginia with a 
source, the
same photomultiplier tubes coupled to good pure CsI crystals of similar size
yielded a time resolution of 
between two detectors. This
meets the energy resolution requirement of the experiment.
Figure (b) is the spectrum of the total energy of the NaI and
CsI detectors within a delayed pion gate of 
, 
after a valid pion stop. This spectrum includes both charged and neutral
showers as no cuts were applied on the plastic veto detectors. The portion
of the spectrum below channel 310 was prescaled by a factor of about 100. The
pion stopping rate was 
. The spectrum is
dominated by accidental Michel events in the NaI and CsI detectors. The
conversion factor is about three channels per MeV.
Figure (c) is the same as figure (b) with the
selection of only neutral showers (the plastic veto counters do not fire in
either arm). With these requirements, a well defined pion beta peak emerges
at the correct energy and the number of counts is consistent with the correct
branching ratio of 
. However, the counting statistics
is poor --- it reflects only about four shifts of running time --- because of
the time constraints mentioned above.
The following conclusions can be drawn from the above results of the
development run:
. This enables the design of the calorimeter in a
``tower'' geometry with single-ended detector readout, which is more favorable
in high rate measurements.
ps per phototube and ranges
up to 600 ps for crystals with fifty or fewer photoelectrons per MeV.
sr operated very quietly in realistic
conditions with beam.
.
Table: Some properties of pure CsI.
Compared to some other inorganic crystals, for instance 
, pure CsI
possesses the following advantages which dictate its choice as the
calorimeter material: its has a shorter radiation length which means that the
volume of a calorimeter like that of the pion beta decay experiment is less than
the volume of the same calorimeter in 
. Its has a lower density
which suggests a lower weight for the calorimeter with pure CsI. Its higher
refractive index indicates that it would be easier to achieve better light
collection uniformity for tapered pure CsI crystals like the pion beta modules.
Finally, pure CsI has a long nuclear interaction length and consequently,
there would be fewer hadronic interactions in CsI, especially in the pion beta
modules.
Some other properties of undoped CsI include radiation hardness and temperature dependence of the light output. Wei and Zhu observed a continuous decrease in the light yield of undoped CsI after 1 kRads [Wei-92]. According to their findings, pure CsI can sustain high counting rates up to 10 kRads. Woody et al. reported an increase in the light yield and the decay time of the fast component of pure CsI, and a shift to longer wavelengths at low temperatures Woo-90. A tomography system has been designed and is in operation at the Paul Scherrer Institute with the objective to examine the optical properties of each of the CsI modules prior to the assembly of the calorimeter.