Endcap Disc DIRC for PANDA at FAIR Mustafa Schmidt, Klim Bigunenko, Michael D¨ uren, Erik Etzelm¨ uller, Klaus F¨ ohl, Avetik Hayrapetyan, Oliver Merle, Julian Rieke on behalf of the PANDA Cherenkov Group FAIR 2015 - Novosibirsk, Russia November, 2015 0/31 Mustafa Schmidt Disc DIRC 0 / 31
¯ PANDA Spectrometer Source: <http://www-panda.gsi.de> 1/31 Mustafa Schmidt Disc DIRC 1 / 31
Cherenkov Light Charged particles with speed higher than photon phase speed in medium emit Chrenkov light Equation of polar angle θ C for Cherenkov light cone: Cherenkov Angle θ C : 1 cos θ C = n ( λ ) β � � 1 − 1 /γ 2 = 1 − E 2 0 / E 2 with β = Number of photons per track length according to Frank-Tamm-Formula: � 1 � λ 2 1 dN � dx = 2 πα z 2 λ 2 − d λ n 2 ( λ ) β 2 λ 2 λ 1 α ≈ 1 / 137 z : charge number of particle Refractive index n normally a function of wavelength λ (dispersion) 2/31 Mustafa Schmidt Disc DIRC 2 / 31
Photon Prediction Theoretical prediction for 100 π + with momentum p = 4 GeV/c compared to simulated results with Geant4/PandaRoot Material thickness: ∆ x = 2 cm Cherenkov Angle Cherenkov Angle Number of photons Number of photons 0.85 ) [rad] Entries Simulated Results Simulated Results 600 0.845 λ ( C θ Theoretical Preditiction Theoretical Preditiction 0.84 500 0.835 0.83 400 0.825 300 0.82 0.815 200 0.81 100 0.805 0.8 0 300 400 500 600 700 800 300 350 400 450 500 550 600 650 700 750 800 λ λ [nm] [nm] Average photon amount per event: n = 1103 for λ = 300 . . . 800 nm 3/31 Mustafa Schmidt Disc DIRC 3 / 31
Disc DIRC Detector 4/31 Mustafa Schmidt Disc DIRC 4 / 31
Radiator Disk and Focusing Element Internal reflection of light inside radiator disk Cylindrical mirror on backside of focusing element for light focusing on readout plane Parallel photons focused on one spot Photons with different angles focused on different points 5/31 Mustafa Schmidt Disc DIRC 5 / 31
Microchannel Plate PMT Source: Merle, Oliver: Development, design and optimization of a novel Endcap DIRC for PANDA, Phd Thesis, JLU Giessen, 2015 6/31 Mustafa Schmidt Disc DIRC 6 / 31
Quantum Efficiency Collection efficiency approx. 30% (varying for different MCP-PMTs) MCP Efficiency 25 #Efficiency [%] 20 15 10 5 0 300 400 500 600 700 λ [nm] Product of quantum efficiency and collection efficiency equal to probability to detect photon (detection efficiency) 7/31 Mustafa Schmidt Disc DIRC 7 / 31
MCP Lifetime BINP #82 PHOTONIS XP85112 (9001223) BINP #1359 PHOTONIS XP85112 (9001332) BINP #3548 Ham. R10754X-01-M16 (JT0117) PHOTONIS XP85012 (9000296) Ham. R10754X-07-M16M (KT0001) PHOTONIS XP85112 (9000897) Ham. R10754X-07-M16M (KT0002) real PANDA time [a] 2 4 6 8 10 25 quantum efficiency [%] 20 15 10 5 0 2 3 4 10 10 10 integrated anode charge [mC/cm 2 ] Source: Lehmann, A. et al.: Improved lifetime of microchannel-plate PMTs. Nucl. Instr. and Meth. A, (0):-, 2014. 127, 128, 129 8/31 Mustafa Schmidt Disc DIRC 8 / 31
Detector Requirements Separation power ( π , K ): ≥ 4 σ Momentum coverage: 1 . 5 . . . 4 GeV/c θ x = 10 ◦ , θ y = 5 ◦ Polar acceptance min/max: θ x , y = 22 ◦ Detector lifetime: ≥ 10 years in duty cycles of 6 m/y Distance to intersection point: ≈ 194 cm in front of EM calorimeter Magnetic field: 0 . 5 . . . 1 . 3 T Energy deposit in radiator: ≈ 500 Gy for fused silica Energy deposit in optics: ≈ 10 Gy for fused silica ≈ 100 Hz/cm 2 ( E kin > 10 MeV) Charged hadron flux: 9/31 Mustafa Schmidt Disc DIRC 9 / 31
Required Resolution Calculation of required resolution for 4- σ -separation-power: � m π � m K � � � � �� � ��� σ θ C ≤ 1 1 1 4 · arccos 1 + − arccos 1 + 4 GeV 4 GeV n n 10 9 8 7 c resolution per track [mrad] 6 4 3 2 G G G e e e V V 5 V / / / c c c 4 BaBar 3 2 θ tracking error 1 1 2 3 4 5 6 7 8 9 10 π /K-separation [ σ ] 10/31 Mustafa Schmidt Disc DIRC 10 / 31
Effects on Photon Transport Photon trapping inside radiator due to internal reflection (approx. 70 % constant for θ > 10 ◦ ) Chromatic dispersion influencing photon resolution and time of propagation: t prop = s v = s n − λ dn � � c d λ Bulk losses of photons described by Beer-Lambert law: � x � I = I 0 exp − µ ( λ ) Fresnel and surface losses after N reflections due to surface roughness: I = I 0 · R N with R = 1 − (4 π cos θ i R q n /λ ) 2 Losses in filter due to spin vector rotation in strong magnetic field (Faraday effect) 11/31 Mustafa Schmidt Disc DIRC 11 / 31
Choice of Material Properties of chosen material: Large absorbtion length (less bulk losses) Small dispersion High radiation hardness Reasons for using fused silica: Already tested at BaBar DIRC High transmission for small wavelength Well understood technology Disadvantage: High production cost for polished radiator disk at large scale 12/31 Mustafa Schmidt Disc DIRC 12 / 31
Cherenkov Angle Distributions Cherenkov angle in fused silica: 50 48 46 44 c [ ◦ ] 42 θ 40 e- pion 38 kaon proton 36 1 2 3 4 5 6 p [GeV/c] Possible solutions for band width reduction: Higher photon statistics Reduction of wavelength acceptance (optical filter) Correction of dispersion by achromatic optics Correction by means of photons time of flight 13/31 Mustafa Schmidt Disc DIRC 13 / 31
Radiation Hardness Transmission losses of fused silica after radiation with γ -dose of 100 krad: Source: Hoek, M.: Tailoring the radiation hard ness of fused silica. Nucl. Instr. and Meth. A, 639(1):227 – 230, 2011. 107 14/31 Mustafa Schmidt Disc DIRC 14 / 31
CERN Testbeam 15/31 Mustafa Schmidt Disc DIRC 15 / 31
CERN Testbeam 16/31 Mustafa Schmidt Disc DIRC 16 / 31
Photos from Testbeam Setup Testbeam at CERN in May 2015 with 3 FELs and 2 MCPs: 17/31 Mustafa Schmidt Disc DIRC 17 / 31
Testbeam Results Pixel distribution with Monte-Carlo data (green) and testbeam measurements (black) for polar angles. . . θ = 6 ◦ θ = 7 ◦ Source: Etzelm¨ uller, Erik: DIRC 2015 18/31 Mustafa Schmidt Disc DIRC 18 / 31
Testbeam Results Source: Etzelm¨ uller, Erik: DIRC 2015 19/31 Mustafa Schmidt Disc DIRC 19 / 31
Disc DIRC Model Angle Definitions : FEL θ c ϕ Particle Radiator Disk tan ϕ tan ϕ ′ = α FEL particle tan α FEL φ rel ϕ ′ 20/31 Mustafa Schmidt Disc DIRC 20 / 31
Disc DIRC Model Calculation of the Cherenkov angle: θ c = arccos(sin θ p cos φ rel cos ϕ + cos θ p sin ϕ ) (1) θ p : θ angle of particle φ rel : angular difference between φ angle of particle and photon ϕ : Angle between total reflected photon and radiator disk surface Calculation of ϕ if θ c is known: � cos 2 θ p − cos 2 θ c cos ϕ = A cos θ c � A cos θ c � ± + (2) B B B with A = sin θ p cos φ rel and B = A 2 + cos 2 θ p 21/31 Mustafa Schmidt Disc DIRC 21 / 31
Calibration Correlation between pixel number and angle ϕ ′ : α Calibration _(FEL) Fit Parameters [rad] ± m = 0.00347196 1.68377e-05 1.2 ϕ tan ϕ = tan ϕ ′ · tan α FEL Angle ± b = 0.852934 0.000888066 1.15 1.1 1.05 1 0.95 0.9 10 20 30 40 50 60 70 80 90 100 Pixel # 22/31 Mustafa Schmidt Disc DIRC 22 / 31
Future Disc DIRC Prototype Test simulations with new Disc DIRC prototype in Geant4: 23/31 Mustafa Schmidt Disc DIRC 23 / 31
Reconstruction of Cherenkov Angles Testbeam simulations with 55% π + , 30% p , 5% K Beam momentum: p = 2 GeV/c (diameter: 2 cm uniform) Cherenkov Angle Distribution hist4 hist4 Entries Entries Entries 10000 10000 220 Mean Mean 44.26 44.26 200 RMS RMS 2.778 2.778 180 160 140 120 100 80 60 40 20 0 40 41 42 43 44 45 46 47 48 θ Reconstructed Cherenkov Angle [deg] c Reconstruction results without removing outliers 24/31 Mustafa Schmidt Disc DIRC 24 / 31
Separation Power Calculation of separation power for p = 3 GeV/c: ¯ θ c ,π − ¯ θ c , k n σ = θ c , k ) = 2 . 9 1 2 ( σ ¯ θ c ,π + σ ¯ Probability for misidentification: P misid ( n σ ) = 1 � � n σ �� √ 1 − erf = 7 . 1 % 2 2 · 2 n σ n σ 2 σ 1 2 σ 2 25/31 Mustafa Schmidt Disc DIRC 25 / 31
Reconstruction & PID Algorithm Input parameters: Particle momentum vector � p Particle angle and position ( θ p , φ p , x , y ) Hit pattern ( z i , t i , sensor id) Mass hypotheses ( m π , m K , m p ) Calculation of all possible photon paths Computation of theoretical hit pattern and time of propagation Removing unwanted bhits with | z − z pred | < z thresh Matching of arrival times and removing of outliers: | t − t pred | < t thres Assuming gaussian probability density function and calculating pseudo likelihood function for each hypothesis: N � ln L = [ln L ( z i | z pred , i ; σ z ) + ln L ( t i | t pred , i ; σ t )] i =0 26/31 Mustafa Schmidt Disc DIRC 26 / 31
Recommend
More recommend
