Ultra Cherenkov based SAC M. Raggi Sapienza Università di Roma Meeting INFN referee July 2017 Sapienza 10/07/2017
PADME Small angle calorimeter Need to measure photons from ~100 MeV No need for high light yield material 0.5-2p.e./MeV more than enough Need to cope with very high rate several ~10 e+ per 100ns Avoid scintillation mechanism if possible (t too long) Need a good time resolution ~200 ps Need very fast photosensors with low transit time spread Need to be radiation tolerant (oder 1Gy per 1013 e+ on target) Mauro Raggi - Sapienza Università di Roma
PADME SF57 based SAC Inefficiency for electrons at BTF Very fast Cherenkov signal Few100MHz rate capability Low light yield expected Need ~0.5 p.e./MeV Very good time resolution ~200ps Need very fast photosensors with low transit time spread Inefficiency for electrons at BTF NA62 (from OPAL) lead glass (Schott SF57) Mauro Raggi - Sapienza Università di Roma
The PMT Hamamatsu R9880U-110 Compact ultra fast High Gain PMT Diameter 16 mm only 8 mm sensitive area Only 0.57 ns rise time and 0.2 ns transit time spread Typical gain 2x106 Mauro Raggi - Sapienza Università di Roma
Read out board CAEN V1742 CAEN V1742 Sampling frequency 5Gs/s 12bits for 1V dynamic 1024 samples (200ns) Mauro Raggi - Sapienza Università di Roma
First Small Angle Calorimeter test R9880U-110 First tests during November calorimeter beam-time Just one lead-glass bar, 20×20×200 mm3, Wrapped in Teflon no optical coupling Hamamatsu R9880U-110, operated at 950V (G~1.5x106) Readout with CAEN V1742 digitizer set to 5 GS/s Obtained 0.15 p.e/MeV with 20x20x200mm3 SF57 crystals Mauro Raggi - Sapienza Università di Roma
Improving crystal LY SF57 Can we found a glass with better transparency in the UV region? SF57 seems to have a bad fall down at ~400 nm Important gain is expected in the light yield due to better matching with Cherenkov spectrum Mauro Raggi - Sapienza Università di Roma
Lead-fluoride PbF2 vs SF57 Density 7.77 5.51 X0 0.93 1.54 Moliere radius 2.12 2.61 Interaction Length (l) 22.1 20.6 l/X0 23.65 13.3 n 1.8 Higher density, more compact showers, better l/X0 ratio. Better transparency down to ~250 nm 10x more radiation hard wrt SF57 SF57 arXiv:1412.5525v2 Mauro Raggi - Sapienza Università di Roma
Looking for the best PMT Hamamatsu PbF2
Data for the PMT choice Transparency and Cherenkov do not match the Y scale just for comparison of wavelength regions involved Cherenkov PbF2 transparency R13478Q R13478UV 9880U-110 Compute the convolution of PBf2*R13478UV-10*Cherenkov; PBf2*R13478Q-10*Cherenkov; PBf2*R9880U110*Cherenkov; To select the best solution Possible PMTs for the SAC 400U= R13478UV-10/-11 = 640 Euro 400S = R13478Q-10/-11 = 1.024 Euro 9880U-110 ~ 600 Euro
Convoluted spectra PBf2*R13478UV-10*Cherenkov; PBf2*R13478Q-10*Cherenkov; PBf2*R9880U110*Cherenkov; Due to the PbF2 transparency there is no difference in between R13478UV R13478Q Integral R13478UV-10 = 289356 A.U. Integral R13478Q-10 = 282977 A.U. Integral 9880U110 = 408280 A.U. Diff=2.2% Diff=30% 9880U-110 R13478UV R13478Q We don’t need the more expensive UV PMT we can buy R13478Q-10
R13478 Tapered or untapered 9880U110 comparing performance
Expected signals with different solutions Comparing signals obtained from R13478UV tapered and untapered dividers with R9880U Assuming the following R13478UV: Npe = 1 p.e./MeV and Singal wdt = 2 ns GTap=3.2E5 and GUnt=5.3E5 R9880U: Npe = 0.1 p.e./MeV and Singal wdt = 1 ns Using the following formula: Signal(V) = Qtot/SWDT*Rload = (Npe*Eg*G*Qe)/SWdt*50W Signal(Q) = Qtot = (Gaus(Npe*Eg,sqrt(Npe*Eg)) *G*Qe)
Expected signal R13478UV untapered R9880U R13478UV tapered Thanks to better gain x8 wrt R13478UV tapered even R9880U has higher signal at nominal gain! 1.43/21 = 6.8% 1.82/10.4 = 17.5% 0.88/13.2 = 6.7%
Transparency spectrum comparisons PbF2 sample 30x30x140 mm3 measured by Atomiki Lab in Debrecem SF57 sample 20x20x200 mm3 measured by Atomiki Lab in Debrecem Results in agreement with expectation from literature SF57 Atomiki measurement PbF2 Atomiki measurement PbF2 Literature
Comparing different materials SF57 Atomiki PbF2 Atomiki PbF2 Literature Integral 20786.5 a.u. Integral 28967.4 a.u. Integral 32616 a.u. Diff % 0 Diff + 39.3% wrt to SF57 Diff + 56.9% wrt to SF57 Convolution=TranspSF57(nm)*QE_UV(nm)*Cherenkov(nm) Convolution=TranspPbF2(nm)*QE_UV(nm)*Cherenkov(nm) Convolution=TranspPbF2Lite(nm)*QE_UV(nm)*Cherenkov(nm)
The PBf2 Crystals test beam 1 crystals 30x30x140 mm3 obtained on loan from Fermilab G-2 used at LNF test beam. 2x9800U PMT used during the test Coverage is only 11% of the surface We will reach 42% with the new PMT All the crystal is black tapered Quotation asked to SICCAS for brand new crystals: 30x30x150 mm3 ~ 2000$ 3-6 month delivery time Some issues on transmittance for longer 180mm samples. 30 mm 30 mm Mauro Raggi - Sapienza Università di Roma
Surface coverage Surface coverage of one PMT 9880-U110 30 mm Surface coverage of one PMT 9880-U110 0.4x0.4 x p = 0.5 cm2 on 3x3=9cm2 just 5.5% coverage Surface coverage of 2 PMT 9880-U110 0.4x0.4 x p = 0.5 cm2 on 3x3=9cm2 just 11.% coverage Experiment setup 1inch R13478 hamamatsu PMT Using 30x30mm2 front face for the PbF2 crystals 2.2x2.2xp = 0.5 cm2 on 3x3 = 9cm2 just 42.% coverage Roughly a factor 4 improvement in LY expected 30 mm 30 mm 30 mm Mauro Raggi - Sapienza Università di Roma
First look at June test beam data Structures observed in laser pulses as well maybe a characteristic of this fast PMT Naïve reconstruction: QCh[i] = sum samples in the interval [-1.5ns,1.5ns] wrt maximum sample TCh[i] = position of the signal maximum Mauro Raggi - Sapienza Università di Roma
Different PMT charges Ch[0] Ch[1] Mauro Raggi - Sapienza Università di Roma
Comparing signals Ch[0] Ch[1] Ch[0] seem to have really smaller signals Mauro Raggi - Sapienza Università di Roma
Charge for >1e- 450 MeV Ch[1] Ch[0] Difference in charge seems to be related to a gain difference Tubes where having the same HV setting with no equalization. Difference in resolution is not that important: Ch(1)= 3.2/9.9=32% Ch(0)= 1.39/4.35=32% Mauro Raggi - Sapienza Università di Roma
Charge correlation Mauro Raggi - Sapienza Università di Roma
Preliminary efficiency Mauro Raggi - Sapienza Università di Roma
Preliminary random veto (noise) Mauro Raggi - Sapienza Università di Roma
TCh[0] TCh[1] correlation Very good time correlation in between the two PMT! Evidence for out of bunch particles leakage Mauro Raggi - Sapienza Università di Roma
Preliminary time resolution Just the peak position is used Expected time resolution per single block <100ps with 5Gs sampling rate! T0 of V1742 digitizer ~ 50ps in between adjacent channels. Mauro Raggi - Sapienza Università di Roma
Conclusions Cherenkov radiator coupled with a R9880-U100 PMT can provide extremely fast signal RMS~600ps have been measured with electrons at BTF Reasonably high light yield ~0.15 p.e. MeV can be reached Even higher performance ~1p.e. MeV can be achieved using PbF2 Interesting solution to be explored PbF2 10x higher radiation hardness wrt SF57 4-5x higher light yield More compact calorimeter (X0 only 0.93 cm) Two samples of PbF2 received from mu2e are ready to be tested Cherenkov radiators maybe a suitable technology for PAMDE SAC detector Mauro Raggi - Sapienza Università di Roma
SPARE SLIDES Mauro Raggi - Sapienza Università di Roma
SF57+R9880-U110 Very short signals! 3 peaks in ≈5 ns Black tape at end 1.5 ns pulse Black tape 20 ns Light reflections? n = 1.8467, speed of light≈16 cm/ns, 40 cm=2.5 ns Number of events with multiple peaks reduced by rotating the crystal by ≈ 60° wrt the beam direction Much better results by placing black tape absorber on the crystal front face. Technology choice SF57+R9880U-110 seem ok! Mauro Raggi - Sapienza Università di Roma
Long beam pulses 150 ns 150 ns 200 ps 200 ps Mauro Raggi - Sapienza Università di Roma
Automatic peak fitting Signal wdt = 3.5 samples Means 3.5*0.2 = 700ps Very short signals!! 200 ps Root macro able to identify multiple peaks and measure the position (time) Can be used to measure the double pulse resolution of the detector. Integral of fit function to compute charge under development. Mauro Raggi - Sapienza Università di Roma
Multi peak event Two real peak identified and position measured 200 ps Mauro Raggi - Sapienza Università di Roma
Signal amplitude and time Mauro Raggi - Sapienza Università di Roma
Npeaks and TDiff Minimum Tdiff 2.5ns means that we can distinguish peaks 2.5 ns apart! Mauro Raggi - Sapienza Università di Roma
Run 490 high multiplicity long bunch 200 ps Run 490 was a long pulse high multiplicity run Mauro Raggi - Sapienza Università di Roma
Amplitude and time Several overlapping particles Signal up to 140 ns but distribution not very flat. SAC can be used to monitor beam bunch structure Mauro Raggi - Sapienza Università di Roma
200 MeV run 494 (one peak charge) No Black tape Spectrum fitted with a Landau distribution MPV = 5.2 pC Q=5.2 pC= eNpeG ⇒ Npe=Q/(eG)= 5.2E-12/(1.6E-19*1.5E6)~21.6 p.e. Order 0.1 p.e./MeV of incident energy (200MeV) to be corrected with MC for deposited energy Mauro Raggi - Sapienza Università di Roma
200 MeV run 495 (one peak charge) Black tape Spectrum fitted with a Landau distribution MPV = 4 pC Q=4pC= eNpeG ⇒ Npe=Q/(eG)= 4E-12/(1.6E-19*1.5E6)~17.2 p.e. Order 0.083 pe/MeV of incident energy (200MeV) to be corrected with MC for deposited energy Mauro Raggi - Sapienza Università di Roma
Energy deposit simulation 200 MeV Simulated single crystal of SF57 20x20x200 mm3 Incident electrons energy 200MeV Radius of the beam spot 3mm. Energy deposit ~ 130 MeV Fraction of deposit ~65% Renormalizing LY of the SF57 Run 494 = 0.1/0.65 = 0.154 Run 495 = 0.083/0.65 = 0.127 Mauro Raggi - Sapienza Università di Roma
Radiation damage Radiation damage 60Co: 1) PbF2 after 200Gy of 60Co 2) Lucite after 200Gy of 60Co 3) SF5 after 200Gy of 60Co PbF2 damage after high dose No effect up to 100Gy Serious damage at 1KGy ~1Gy is the expected dose at PADME arXiv:1412.5525v2 Mauro Raggi - Sapienza Università di Roma
Results obtained with PbF2 G-2 arXiv:1412.5525v2 Mauro Raggi - Sapienza Università di Roma