Light transport simulations of scintillator arrays fabricated using laser induced optical barriers

Objectives The goal of this work is to simulate and fabricate single-side readout PET detectors with DOI capability using laser induced optical barriers (LIOB).

Methods We have successfully used LIOB to fabricate 20 mm thick LYSO crystals with 1 mm pixel size, as well as 10 mm thick scintillator crystals with 0.77 mm pixels. We have further shown that the light spread and pixel separation in the flood map can be controlled by optimizing the laser pulse energy and optical barrier density. In this work we report on light transport simulations of laser processed LYSO:Ce scintillators using the DETECT2000 Monte Carlo code. We have simulated a 10x10x20 mm³ crystal with slab-like optical barriers of constant shape and size. The barriers are 50 um thick and arranged with 1 mm separation, resulting in a 10x10 pixel array. The crystal is surrounded by specular reflectors on five sides, and an isotropic light source is placed at different depths in one central pixel. We studied the scintillation light collection at the exit surface of each scintillator element as a function of barrier refractive index (RI) for two scenarios: constant barrier RI throughout the crystal thickness, and different RI in the top and bottom halves of the crystal respectively. We further analyzed the effect of barrier shape on the barrier reflectivity by simulating and comparing a slab-like - and a spherical barrier.

Results The simulation results show that when the barriers have constant RI of n=1 (corresponding to air), ~ 67% of the light is collected at the exit side of the pixel containing the light source, showing light channeling. At the exit surface of adjacent pixels around 0.5% of the light is collected. When the barrier refractive index is increased to n=1.2, 51% of the light is preserved within the pixel. For n=1.4 the corresponding number is 33%. We also see that light outside the pixel containing the source is channeled orthogonally along the barrier pattern, rather than diagonally. The results further show that no DOI information may be achieved for constant RI throughout the crystal, since similar results are acquired regardless of the position of the light source. When using different RI In the top and bottom halves of the crystal we observe a significant difference in the amount of light contained in the pixel, depending on the position of the light source. With RI of n=1.2 in the top half, and n=1.0 in the bottom half, 60% of the light is contained in the pixel when it is generated in the bottom half of the crystal, and ~51% when it is generated in the top half. The corresponding numbers when using n=1.4 in the top half and n=1 in the bottom half is 51% and 33% respectively. Our results also suggest that while the amount of reflected light from spherical and slab-like barriers are the same, the reflectivity signature of the barrier depends on its shape. CONCLUSIONS: The reported results are encouraging in that optical barriers in scintillators processed by the LIOB technique can be used to channel a large fraction (but not all) of the scintillation light, suggesting that the barriers are well suited for light sharing detector scheme. Further we showed that depth of interaction information in a single side readout detector may be achievable by simply using depth dependent refractive index for the optical barriers Research support: L. Bläckberg acknowledges financial support from the Swedish Research Council (VR).