P ILC A. Calcaterra (Resp.), L. Daniello (Tecn.), R. de Sangro, G. Finocchiaro, P. Patteri, M. Piccolo, M. Rama Introduction and motivation for this study Silicon photomultipliers ), often called SiPM in literature, are semiconductor photon detectors built from a square matrix of avalanche photodiodes (APD s) on a common silicon substrate. The side of each single APD square microcell can vary in length from 2 to µm. Every microcell -or pixel- acts like a capacitor holding a given charge, and releases this entire charge when a photoelecton created in the cell reaches an active region inside the pixel, initiating a Geiger-like discharge. Pixel charges then build up again in a small time (2 2 ns). Although individual pixels are intrinsically digital, the SiPM behaves more like an analog device, because its signal is the sum of signals from all pixels discharged at the same time. The PILC-LNF group started working in in the context of the Proposal for ILC initiative, with the intent to develop complete detectors of a size suitable for ILC calorimetry, and to study their performance with cosmic rays and beam tests. 2 New prototypes built in and Results at the Beam Test Facility We exposed seven different assemblies of scintillator tiles coupled to different silicon photon detectors (PD) to the electron beam at the Frascati Beam Test Facility(BTF) 2) ( MeV electrons, ( )mm 2 RMS transverse size). We cut and polished a total of (3 3)cm 2 plastic scintillator tiles of 2 and mm thickness, and wrapped them in aluminized mylar and black tape; we used as scintillators used Saint Gobain BC and Eljen Technology EJ22 which have almost identical characteristics, and a similar scintillator made in Vladimir (Russia). We also studied, as a reference, a CALICE tile made with the Vladimir scintillator and a mm diameter Kuraray Y wavelength shifter fiber. We used photon detectors from Hamamatsu (see Tab. for details). The PD was attached with optical glue directly to the middle of one side of the tile in all configurations except for the (3 3mm 2 ), 3 pixel Hamamatsu MPPC which was instead glued to the center of a face, and the CALICE tile, which has the MEPhI/Pulsar SiPM mechanically coupled, without glue or optical grease, to one end of the WS fiber. Table : Configurations under test. See text for the definitions of V brk, V bias and G. Config. Scint. Type PD (tot. area) # of pxl pxl size V brk (V) V bias (V) G( ) mm BC Hamam. ( mm 2 ) µm. 2..3 2 mm BC Hamam. ( mm 2 ), 2µm. 2.. 3 mm Vladimir Hamam. ( mm 2 ) µm... 4 2mm EJ22 Hamam. ( mm 2 ) µm.. 2.2 2mm EJ22 Hamam. ( mm 2 ), 2µm. 3.3. mm BC Hamam. ( mm 2 ) 3, µm.3.. mm Vladimir MEPhI ( mm 2 ), 2µm.4 4..
The PD s were read out with a low noise amplifier based on the GALI- chip, an INFN- Pisa design built in Frascati. The seven tiles were then mounted in a test box, aligned to each other to mm in a fixed position. The box was equipped with a T-probe to monitor the operating temperature with a typical resolution of.2 C. Temperature in the experimental hall was regulated to 23.4±. C, and the air inside the box was kept constant at 2.2±.2 C using a Peltier cell to extract some of the heat produced by the amplifiers. To measure the impact point of the beam on the tile, we used an external tracker including glass RPC 3), 3 of which were placed in front and Figure : Test beam set up. 2 behind the test box on the beam line. The RPC were equipped with orthogonal planes of strips mm wide, digitally read out, providing X-Y measurement in each plane with a point resolution of 2.3mm. As the beam at the Frascati BTF can provide a tunable number of particles per pulse (-) 2), the test setup included a lead glass calorimeter module to measure the beam total energy on a pulse by pulse basis and allowing the selection of events containing any number of electrons (,,...n). A picture of the set up in the beam line is shown in Fig.. The response to MIP of the tiles under test, in terms of number of fired pixels, is given in Fig. 2. All data shown were taken with the values for V bias listed in Tab., close to the corresponding gains, expressed in number of electrons (Q pxl /e). 3 Results from cosmic ray tests We have collected several million events containing exactly MIP, and studied the performances of the various devices as a function of the beam particle impact point on the tiles. In Fig. 3 we show the amplitude of the PD signal (number of pixels), which is proportional to the amount of scintillation light collected, as a function of one coordinate, whereas in the four leftmost plots of Fig. 4 we show the efficiency in two dimensions. The efficiency is defined by the Config 2 4 Config 2 3 4 Config 2 2 4 2 4 2 4 2 3 4 2 3 4 Figure 2: Signal (number of pixels) collected by the configurations in Tab. number of times a signal above 2 pixels is observed in the PD over the number of times a particle
4 3 2 Config 4 2 Config 2 4 2 4 4 2 4 2 4 4 2 4 2 Config 4 4 3 2 4 4 Figure 3: Profile histograms of the signal (number of pixels) as a function of the MIP impact point of the tile, collected by the seven different test configurations of Tab.. has crossed the corresponding tile; the chosen threshold corresponds to about / to /4 of a MIP signal, depending on the configuration. From Fig. 3, we can see that all tiles read out without using the fiber show a somewhat higher non-uniformity in light collection, but without loss of efficiency. The signal amplitude is maximal when the impinging particle is closest to the PD, and decreases with distance. This effect is not seen for the CALICE tile, where the light is collected by the fiber and where, on the other hand, a degradation of the light collection near the edges is observed. We estimate a rather high nonunifomity of 3% when the PD is attached to the face of the tile (config. ), while config. 3 and 4 are more uniform ( % and 2% respectively). The latter two values are small compared to the intrinsic fluctuations of a MIP energy deposit, therefore their effect on an energy measurement should also be small. In the four rightmost plots of Fig. 4 we show the X,Y distribution of the pulse height for configurations number 3,4, and. As one can see, the largest response variation is also restricted to a small region near the position of the PD, being quite uniform elsewhere. This means that only a small fraction of particles crossing a tile are affected by this non-uniformity. The difference in efficiency between the mm and the 2mm tile is evident from the two upper leftmost plots of Fig. 4 (config. 3 vs 4); nevertheless, an efficiency greater than % over a large portion of the tile is observed even with the thinner scintillator. These preliminary results seem to suggest that direct read out of scintillator tiles with silicon PD for an ILC hadron calorimeter application is possible even using very thin tiles, and prompt for detailed Monte Carlo studies to estimate their performances in a detector.
4...4.2 4...4.2 4 4 3 2 4 3 2 4...4.2 4...4.2 4 4 2 4 3 2 Figure 4: The leftmost four plots show the efficiency as a function of the X,Y coordinates of the impact point on the tile for configurations number 3,4,, of Tab.. The rightmost four plots show the pulse height in number of pixels as a function of X,Y. 4 Results from cosmic ray data taking The small temperature variations at the BTF are useful for the studies above, but a definite disadvantage when finding the dependence from T of the PD gain. To obtain this measurement, we continuously collect cosmic ray data using the same setup in a separate laboratory. We report here preliminary numbers for detectors and 4 (see Fig. ): the fitted values of the relative gain variation per degree of T are greater (resp. %/ K and 4%/ K) than usually found in literature and moreover widely different for 2 supposedly identical PDs. Figure : The fitted T-coefficient (%/ K) for detectors and 4 in Tab..
Foreseen activity in This R&D program will continue, both at the BTF and using cosmic rays, with the study of more PD types and configurations. The tunability of the number of particles in the beam, peculiar to the Frascati BTF, will also allow studies of the dynamic range of each tile-pd read out configuration. Future studies will also include measurements of the timing performances of PDs. List of Conference Talks. R. de Sangro, Study of Scintillation Tile Detectors readout via Silicon Photomultiplier Devices, presented at the International Linear Collider Workshop, Chicago (Il.) Publications. A. Calcaterra et al, Nucl. Instr. & Meth. A, 3 (). References. A. Akindinov et al., Nucl. Instr. & Meth. A 3, 23 (); M. Danilov, arxiv:2.. 2. A. Ghigo et al., Nucl. Instr. & Meth. A, 24 (3). 3. A. Calcaterra et al., IEEE Trans. Nucl. Sci. 3, 34 ().