When excellent energy resolution of photons and electrons is required of a detector, scintillating crystals are usually the material of choice. Lead tungstate was chosen as the detector medium as it has a short radiation length and small Molière radius facilitating a compact detector design. The CMS ECAL will contain some 75000 crystals, making it one of the largest crystal detectors ever built.

Imperial College has been involved in the testing of lead tungstate crystals for the calorimeter endcaps and the Crystal Laboratory at Imperial has played an important role in the evaluation and optimisation of crystals from several manufacturers.

In order to fully exploit the resolution capabilities of crystal calorimeters, careful attention must be given to various factors. One of these is the uniformity of light yield along the length of the crystal.

When a high energy photon or electron strikes a crystal, the incident particle gives up its energy to the crystal by creating a shower of secondary photons and electrons. For a given incident particle energy, the position in the crystal at which the maximum energy deposit occurs fluctuates. If some property of the crystal causes the light yield to depend on the position of the maximum energy deposit then the energy resolution will be compromised.

The crystals to be used are tapered in shape such that the whole detector points more or less at the beam interaction point. This tapered shape focuses scintillation light produced at the front of the crystal onto the photodetector at the rear. This focusing effect, coupled with the material's high refractive index, means that light produced at the front of the crystal is more likely to be detected than light produced further along the crystal.

Simulations have shown that the change in light yield per cm can be no more than 0.4% if the target energy resolution is to be met. To measure such a change requires a precision measurement of the crystal light yield.

Lead tungstate has an intrinsically low light yield so in order to achieve a precision measurement of the number of photoelectrons produced by a given energy deposited in the crystal a hybrid photomultiplier tube (HPMT) was used. The HPMT has no dynode stages found in the familiar PMT, but rather a silicon diode provides gain. Excellent single photoelectron resolution can be obtained.

The plot on the right shows a typical photoelectron spectrum obtained at Imperial College using lead tungstate and a Co-60 source. The plot also shows a fit to the data. The fitting program was developed at Imperial College to allow a precise value of the photoelectron yield to be extracted.

The function used was derived from first principles and takes into account Compton scattering, pileup and backscattering of photoelectrons from the silicon diode (visible in the data as shoulders on the peaks). For high light yield crystals, a chi-squared as low as 1.1 can be achieved.

Distribution of non-uniformities for a sample of endcap crystals.