Dual-sided readout studies

1. Dual-sided readout studies Chiara Casella Matthieu Heller Oliver Holme

2. Dual-sided readout studies • Difference in Left and Right light yield is used to estimate gamma interaction position along z axis • Original AXPET concept was based on dual-sided readout • Initial studies showed insufficient z-axis resolution (~cm) • New results from A. Goertzen indicated better z resolution (~3.5mm) • Investigation started to find best achievable resolution • Following A. Goertzen method • Using digital SiPM for a compact axial module L R z-axis 0

3. Laboratory setup • Philips digital SiPM tiles – L, R and Tagger • L and R tiles are cooled to ~14˚C to reduce dark count • Only events with 3 tile coincidence are stored • Acts to collimate a beam of gamma photons • Automated data taking with Philips and LabVIEW software • Scans through several position along z-axis

4. Crystal scanning • 3x3x100mm LYSO crystals • Crystal aligned with one SiPM pixel • All other cells were inhibited • Typical scan properties • 11 data taking runs at different positions along z axis • Typical run properties • ~15 minutes (with 3.7 MBq source) • Up to 9000 events stored to file • Up to 1300 events after photopeak cuts

5. Data processing for each z position • Data from SiPMs is converted to ROOT files • Light yield is processed: • Cut on photopeaks (all three tiles) • Correct for SiPM cell saturation • Correct for SiPM intra-tile cross talk • Obtain photopeak central values Raw data Gaussian fit Corrected data

6. Data processing for complete scan • z position estimator calculated as for each event • Mean for each z position is plotted against z • Fit function is applied to plot • Fit gives a relationship between any value and z position (R-L)/(R+L) Error bars show σ of (R-L)/(R+L) (R-L)/(R+L) z position [mm]

7. Data processing for complete scan • z axis resolution • FWHM ofvalues for each z position mapped to FWHM of z • Beam width is subtracted to obtain final resolution • Beam size calculated as 1.7mm (R-L)/(R+L) Error bars show σ of (R-L)/(R+L) FWHM 7 Position [mm] Rz FWHM

8. Crystals and surface treatments • Coatings/wrappings: • Bare • White TiO2 paint • Teflon (2 layers) • Enhance Specular Reflector (ESR) • Surface treatments (one or multiple faces): • Polished (from manufacturer) • Depolished (with ~10 µm powder) • Machined strips (with CNC diamond tool)

9. Machined strips • Two patterns of machined strips • Every 5mm, aligned on four faces • Every 5mm, staggered by 1.25mm on four faces Machined strips are ~0.7mm wide

10. Results

11. Non-exponential LY versus z • Crystalswith less attenuation • LY versus z is not exponential • Crystals with greater attenuation • LY versus z is more similar to exponential Polished, teflon wrapped Two faces depolished, teflon wrapped

12. Inter-tile cross talk • SiPMs have cross talk between cells in a tile • Intra-cell cross talk • Caused by avalanches producing secondary photons • The secondary photons are emitted isotropically • They can also enter and propagate through the crystal • Inter-tile cross talk • Inter-tile cross talk reduces L & R difference • Degrades z resolution • This degradation can be measured by capturing L and R tiles independently, with the other tile off • Not possible in real applications • Physically filtering the cross talk photons might be possible

13. Photopeak shifting Machined strips, 5mm (aligned), teflon wrapped LEFT RIGHT SUM z-axis

14. Light yields and energy resolution Machined strips, 5mm (aligned), teflon wrapped

15. z resolution Machined strips, 5mm (aligned), teflon wrapped

16. Different surface treatments Energy resolution • Resolution down to ~3mm has been achieved • Treatments give different energy resolutions due to different LY sum Energy resolution [%] Polished + TiO2 paint Machined strips (four faces, aligned) + ESR Depolishing (two faces) + ESR

17. Machined strips • Fine scans reveal discrete structure • Gives stepped LY pattern • Seen in LY sum and energy resolution too Aligned strips, light yield - RIGHT Staggered strips, light yield - RIGHT Fitted photopeak position [pe] Fitted photopeak position [pe] Position [mm] Position [mm] Staggered strips give a pseudo-continuous behaviour

18. Depolishing One face depolished + ESR Two faces depolished + ESR

19. Wrappings Four faced machined strips (aligned) Light yield - LEFT Light yield - RIGHT z [mm] z [mm] Light yield - SUM ESR wrapped λeff ≈ 70 mm Teflon wrapped λeff ≈ 40 mm Unwrapped λeff ≈ 28mm ESR gives almost constant LY SUM z [mm]

20. Wrappings Four faced machined strips (aligned) Teflon and ESR give similar resolution ESR gives more linear z position estimator ESR gives more light to improve resolution further

21. Results summary A crystal with better Rz at z=50mm is not necessarily better at other z positions

22. Next steps and outlook • Improve results for 100mm crystals • Extend machined strips to full crystal length • Depolishmore faces • Shorter 60mm crystals • Could achieve resolution down to 2mm • Attenuation per mm would need to be increased • Photodetectors with higher PDE ? • Greater light yield would enable better resolution • Current SiPM PDE is about 40%

24. Toy Monte Carlo results • Measured results are consistent with Toy MC • Estimated ~3mm as best achievable resolution • Can indicate future possibilities Pure Poisson distribution Noise Not representative Increased LY improves resolution 60 mm crystals would enable improvement down to 2mm with current light yields 20mm crystal figures match other published studies

25. Machined strips width

26. Depolished (one face)