The large photomultipliers (PMT) are currently used in astroparticle and neutrino experiments. The light level and event rate are so low that the tube efficiency, noise rate and afterpulse rate have to be understood and improved.

The next generation of experiments will include large surfaces of photodetection. Even with large photomultiplier tubes, their total number may reach thousands of units. Therefore, cost reduction has become a critical issue.

The R&D Detection department is carrying out a research and development program with those requirements in mind, with Photonis, the French PMT producer. 9 inch diameter tubes with a new photocathode were characterized. The influence of the diameter was tested on critical parameters for the astroparticle experiments: timing, afterpulses, noise. Tubes with diameters of 5, 8, 9 and 10 inches were measured.

The light level is very low in astroparticle physics experiments: most of the detected signals correspond to one photoelectron. The surface of the photocathode was increased to collect more photons but this is not sufficient. Therefore, Photonis developed a new process to increase the photocathode quantum efficiency, which is currently around 25 %.

A dedicated test bench was set up at IPN Orsay R&D Detection department to measure the PMT detection efficiency (the product of the quantum efficiency by the photelectron collection efficiency of the multiplier). The light source comes from the interaction of alpha particles (²⁴¹Am source) in a small fast plastic scintillator (Bicron BC 422). This light source is very stable, homogeneous and noiseless. The trigger is given by a fast PMT (Photonis XP2020) directly coupled to the scintillator through a light guide in order to collect a large number of photoelectrons (around 80). This ensures a very good timing reference. The measured PMT is placed vertically at a large distance of 2.5 m from the light source in order to obtain a quasi parallel light and an injected light level of one photoelectron. The ratios of the number of pulses detected by the PMT under test and by the trigger PMT are measured in order to compare the detection efficiency between the two PMT families. This calculation is performed on the charge histograms obtained by triggering the data acquisition with the trigger PMT at a threshold of 50 photoelectrons. A mean improvement of around 16 % was measured.

This set-up is also used to measure the transit time spread. The time resolution as a function of the PMT voltage distribution was measured for the different diameters of tubes.

The results give guidelines to make a compromise between time accuracy, gain and collection efficiency.

The PMT noise consists in spontaneous pulse emission. It is measured at a threshold of 0.3 photoelectrons.

The noise rate decay time after an exposure to light (dazzling effect) was determined. The decay was measured with different types of photocathodes.

The noise rate as a function of the temperature was also measured. The light-tight box used for those measurements is placed in a climate test chamber (Vötsch VC 4034). This study was performed with different types of photocathodes and glass in order to characterize their contribution.

The afterpulse rate and time distribution characterize the influence of the production process on the vacuum quality inside the tube. The influence of the multiplier structure was studied by measuring the afterpulse distribution with several voltage distributions. The measured tube is placed vertically in a light tight box or in the climate test chamber (same set-up as for the noise measurement). A pulsed LED generates light. The signal is measured with a digitizing oscilloscope (Lecroy WaveRuner 6050A, 300 MHz bandwidth). The typical recording length is 20 µs. The sampling frequency is 500 MSPS. The curves are processed in real time. The position and amplitude of pulses are recorded to reduce the data size, and therefore the time access to the hard disk. This set-up allows to process up to 300 events per second.

Simulations of electron and ion trajectories inside the tube were carried out at the R&D Detection department with the SIMION software. The results were compared with the measurements: an hypothesis on the afterpulse generation was confirmed.