2586 IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 56, NO. 5, OCTOBER 2009 Energy, Timing and Position Resolution Studies With 16-Pixel Silicon Photomultiplier Matrices for Small Animal PET Gabriela Llosá, Nicola Belcari, M. Giuseppina Bisogni, Gianmaria Collazuol, Sara Marcatili, Student Member, IEEE, Pierre Barrillon, Christophe de la Taille, Senior Member, IEEE, Sylvie Bondil-Blin, Nicoleta Dinu, Mirko Melchiorri, Alessandro Tarolli, Claudio Piemonte, Member, IEEE, and Alberto Del Guerra, Senior Member, IEEE Abstract—A high resolution small animal PET scanner that em- ploys Silicon Photomultiplier (SiPM) matrices as photodetectors is under development at the University of Pisa and INFN Pisa. The first SiPM matrices fabricated by the Center for Scientific and Technological Research, FBK-irst (Trento, Italy), are being eval- uated for this purpose. The devices are composed of 16 (4 4) pixel elements of 1 mm 1 mm in a common substrate. The first tests have been carried out employing the ASIC MAROC2 for the readout. Energy and timing resolution, and position determina- tion tests have been performed coupling both pixellated and con- tinuous LYSO scintillator crystals to the matrix, and the results have been compared with the ones obtained for single SiPMs. The first tests on position determination with continuous crystals and SiPM matrices have been performed. An intrinsic spatial resolu- tion of 0.61 mm FWHM has been obtained. Index Terms—G-APDs, silicon photomultiplier, small animal PET. I. INTRODUCTION A HIGH resolution small animal PET scanner with con- tinuous scintillator crystals and silicon photomultipliers (SiPMs) as photodetectors is under development at the Univer- sity of Pisa and INFN Pisa [1], [2]. The prototype will con- sist of two detector heads, each of them made of three detector layers. The detector layers are composed of a continuous LYSO slab and a SiPM matrix structure. For this tomograph, a spa- tial resolution better than 1 mm FWHM is expected for a F point source at the center of the field-of-view employing filtered backprojection algorithms for image reconstruction, according to GEANT4 simulations. The SiPM matrices fabricated at the Center for Scientific and Technological Research (FBK-irst) in Trento, Italy, are being tested for this purpose. The first matrices Manuscript received March 23, 2009; revised June 22, 2009. Current version published October 07, 2009. This work was supported in part by the European Commission’s 6th Framework Programme through a Marie Curie Intra-Euro- pean Fellowship. G. Llosá is with the University of Pisa, Department of Physics, I-56127 Pisa, Italy (e-mail: gabriela.llosa@pi.infn.it). N. Belcari, M. G. Bisogni, S. Marcatili, and A. Del Guerra are with the Uni- versity of Pisa, Department of Physics, and INFN Pisa, I-56127 Pisa, Italy. G. Collazuol is with the Scuola Normale Superiore and INFN Pisa, I-56127 Pisa, Italy. P. Barrillon, C. de la Taille, S. Bondil-Blin, and N. Dinu are with the Linear Accelerator Laboratory, IN2P3-CNRS, Université Paris Sud 11, Paris, France. M. Melchiorri, A. Tarolli, and C. Piemonte are with the FBK-irst, Center for Materials and Microsystems, I-38050 Povo di Trento, Italy. Digital Object Identifier 10.1109/TNS.2009.2030191 Fig. 1. First SiPM matrices produced at FBK-irst, composed of 16 (4 4) SiPM pixel elements of 1 mm 1 mm size in a common substrate. produced are composed of 16 (4 4) SiPM pixel elements in a common substrate [3] (Fig. 1). Measurements have been performed with the matrices cou- pled to both pixellated and continuous crystals. In addition, measurements with single SiPMs have been done in order to have a reference for the results obtained with the matrices. SiPM response, energy resolution and coincidence timing resolution have been evaluated. II. EXPERIMENTAL SETUP A. Detector Description The SiPM matrices are composed of 16 (4 4) SiPM pixel el- ements in a common substrate (Fig. 1). The pixels have an active area of 1 mm 1 mm with 625 microcells of 40 m 40 m size. The pitch is 1.06 mm, with a dead space of 60 m between the pixels to route the readout lines from each pixel to one of the two opposite sides of the device. A common bias voltage is applied to all pixel elements through the detector backplane. The full characterization of a matrix fabricated in the first run has been carried out at the Linear Accelerator Laboratory (LAL) in Orsay, France [4]. The device shows an excellent uniformity 0018-9499/$26.00 © 2009 IEEE