1. S. Niccolai, IPN OrsayCLAS12 Workshop, Genova, 2/27/08 Conseil Scientifique IPN Orsay, 12/11/09 INFN Frascati, INFN Genova, IPN Orsay, LPSC Grenoble SPhN Saclay University of Glasgow Deeply Virtual Compton Scattering on the neutron at JLab with CLAS12 GPDs and nDVCS JLab-Hall A measurement Neutron kinematics for nDVCS Central Neutron Detector (CND) for CLAS12 Simulations: expected performances of CND Technical challenges Ongoing R&D in Orsay Cost estimate M. Guidal, S. Niccolai, S. Pisano (groupe PHEN/JLab) B. Genolini, T. Nguyen Trung, J. Pouthas (R&D)

2. Deeply Virtual Compton Scattering and GPDs e’ t (Q 2 ) e L*L* x+ξ x-ξ H, H, E, E (x,ξ,t) ~ ~  p p’ « Handbag » factorization valid in the Bjorken regime: high Q 2,  (fixed x B ), t<<Q 2 Q 2 = - (e-e’) 2 x B = Q 2 /2M  =E e -E e’ x+ξ, x-ξ longitudinal momentum fractions t = (p-p’) 2  x B /(2-x B )   0,x  ),(Ex q  2 1 Hxdx q  J G =  2 1 J q  1 1  )0,,( Quark angular momentum (Ji’s sum rule) X. Ji, Phy.Rev.Lett.78,610(1997) Vector: H (x,ξ,t) Tensor: E (x,ξ,t) Axial-Vector: H (x,ξ,t) Pseudoscalar: E (x,ξ,t) ~ ~ conserve nucleon helicity flip nucleon helicity «3D» quark/gluon image of the nucleon H(x,0,0) = q(x) H(x,0,0) = Δq(x) ~

3. Extracting GPDs from DVCS spin observables  LU ~ sin  Im{F 1 H +  (F 1 +F 2 )H +kF 2 E}d  ~ Polarized beam, unpolarized proton target: Unpolarized beam, longitudinal proton target:  UL ~ sin  Im{F 1 H+  (F 1 +F 2 )(H + … }d  ~  = x B /(2-x B ) k=-t/4M 2 H n, H n, E n Kinematically suppressed H p, H p ~ A =           = ~   leptonic plane hadronic plane p’ e’ e  LU ~ sin  Im{F 1 H +  (F 1 +F 2 )H - kF 2 E}d  ~ Polarized beam, unpolarized neutron target: Suppressed because F 1 (t) is small Suppressed because of cancellation between PPD’s of u and d quarks H p, H p, E p ~ nDVCS gives access to E, the least known and least constrained GPD that appears in Ji’s sum rule H p (ξ, ξ, t) = 4/9 H u (ξ, ξ, t) + 1/9 H d (ξ, ξ, t) H n (ξ, ξ, t) = 1/9 H u (ξ, ξ, t) + 4/9 H d (ξ, ξ, t) Unpolarized beam, transverse proton target:  UT ~ sin  Im{k(F 2 H – F 1 E) + ….. }d  H p, E p Measured or planned at JLab in Hall B

4. J u =.3, J d =.1 J u =.8, J d =.1 J u =.5, J d =.1  = 60° x B = 0.2 Q 2 = 2 GeV 2 t = -0.2 GeV 2 Beam-spin asymmetry for DVCS: sensitivity to J u,d VGG Model (calculations by M. Guidal) DVCS on the proton J u =.3, J d =.8 J u =.3, J d =-.5 E e = 11 GeV

5.  = 60° x B = 0.17 Q 2 = 2 GeV 2 t = -0.4 GeV 2 Beam-spin asymmetry for DVCS: sensitivity to J u,d The asymmetry for nDVCS is: very sensitive to J u, J d can be as big as for the proton depending on the kinematics and on J u, J d → wide coverage needed VGG Model (calculations by M. Guidal) DVCS on the neutron J u =.3, J d =.1 J u =.8, J d =.1 J u =.5, J d =.1 J u =.3, J d =.8 J u =.3, J d =-.5 E e = 11 GeV

6. First measurement of nDVCS: Hall A E e = 5.75 GeV/c P e = 75 % L = 4 · 10 37 cm -2 · s -1 /nucleon Q 2 = 1.9 GeV 2 x B = 0.36 0.1 GeV 2 < -t < 0.5 GeV 2 HRS Electromagnetic Calorimeter (PbF 2 ) LH 2 / LD 2 target  e’e’ e Subtraction of quasi-elastic proton contribution deduced from H 2 data convoluted with initial motion of the nucleon Analysis done in the impulse approximation: Active nucleon identified via missing mass Twist-2 M. Mazouz et al., PRL 99 (2007) 242501

7. nDVCS in Hall A: results S. Ahmad et al., PR D75 (2007) 094003 VGG, PR D60 (1999) 094017 M. Mazouz et al., PRL 99 (2007) 242501 Q 2 = 1.9 GeV 2 - x B = 0.36 Im(C I n ) compatible with zero (→ too high x B ?) Strong correlation between Im[C I d ] and Im[C I n ] Big statistical and systematic uncertainties (mostly coming from H 2 and  0 subtraction) Model dependent extraction of J u and J d F. Cano, B. Pire, Eur. Phys. J. A19 (2004) 423

8. CEBAF @ 12 GeV Add new hall CEBAF@12 GeV: large Q 2, x B ~2012

9. Low threshold Cerenkov Counter Forward Time-of-Flight Hall B @12 GeV: CLAS12 High threshold Cerenkov Counter Drift Chambers Forward Electromagnetic Calorimeter Preshower Calorimeter Inner Electromagnetic Calorimeter Central detector Design luminosity ~ 10 35 cm -2 s -1 Concurrent measurement of deeply virtual exclusive, semi-inclusive, and inclusive processes Q 2 > 2.5 GeV 2 Central Detector (CD): 40°<  <135° Forward Detector: 5°<  <40°

10. nDVCS with CLAS12: kinematics More than 80% of the neutrons have  >40° → Neutron detector in the CD is needed! DVCS/Bethe-Heitler event generator with Fermi motion, E e = 11 GeV (Grenoble) Physics and CLAS12 acceptance cuts applied: W > 2 GeV 2, Q 2 >1 GeV 2, –t < 1.2 GeV 2 5° <  e < 40°, 5° <   < 40° ~ 0.4 GeV/c ed→e’n  (p) Detected in forward CLAS Detected in FEC, IC Not detected PID (n or  ?), p, angles to identify the final state CD In the hypothesis of absence of FSI: p μ p = p μ p’ → kinematics are complete detecting e’, n (p, ,  ),  p μ e + p μ n + p μ p = p μ e′ + p μ n′ + p μ p′ + p μ  FSI effects can be estimated measuring en , ep , ed  on deuteron in CLAS12 (same experiment)

11. limited space available (~10 cm thickness) → limited neutron detection efficiency → no space for light guides → compact readout needed strong magnetic field (~5 T) → magnetic field insensitive photodetectors (SiPMs or Micro-channel plate PMTs)  CTOF can also be used for neutron detection  Central Tracker can work as a veto for charged particles CND CTOF Central Tracker CND: constraints & design Detector design under study: scintillator barrel MC simulations done for:  efficiency  PID  angular resolutions  reconstruction algorithms  background studies

12. Simulation of the CND Geometry: Simulation done with Gemc (GEANT4) Includes the full CD 4 radial layers (or 3, if MCP-PMTs are used) 30 azimuthal layers (can still be optimized) each bar is a trapezoid (matches CTOF) inner r = 28.5 cm, outer R = 38.1 cm Reconstruction:  Good hit: first with E dep > threshold  TOF = (t 1 +t 2 )/2, with t 2(1) = tof GEANT + t smear + (l/2 ± z)/v eff t smear = Gaussian with  =  0 /√Edep (MeV)  0 = 200 ps·MeV ½ → σ ~ 130 ps for MIPs  β = L/T·c, L = √h 2 +z 2, h = distance between vertex and hit position, assuming it at mid-layer  θ = acos (z/L), z = ½ v eff (t 1 -t 2 )  Birks effect not included (will be added in Gemc)  Cut on TOF>5ns to remove events produced in the magnet and rescattering back in the CND z y x

13. CND: efficiency, PID, resolution p n = 0.1 - 1.0 GeV/c  = 50°-90°,  = 0 ° Efficiency: N rec /N gen N rec = # events with E dep >E thr. Efficiency ~ 10-16% for a threshold of 5 MeV and p n = 0.2 - 1 GeV/c Layer 1 Layer 2 Layer 3 Layer 4  distributions (for each layer) for: neutrons with p n = 0.4 GeV/c neutrons with p n = 0.6 GeV/c neutrons with p n = 1 GeV/c photons with E = 1 GeV/c (assuming equal yields for n and  ) n/  misidentification for p n ≥ 1 GeV/c “Spectator” cut  p/p ~ 5%  ~ 1.5°

14. nDVCS with CLAS12 + CND: expected count rates σ(nb GeV 4 )N 160.017945354 420.006271873 740.00276824 1040.00174520 1340.00137410 1650.00127379 1950.00126377 2250.00140417 2560.00172513 2860.00279835 3170.006161838 3470.01825432  t = 0.2 GeV 2  Q 2 =0.55 GeV 2  x B = 0.05  = 30° L = 10 35 cm -2 s -1 Time = 80 days R acc = bin-by-bin acceptance E eff = 15% neutron detector efficiency (CND+CTOF+FD) N = ∆t ∆Q 2 ∆x ∆  L Time R acc E eff Count rates computed with nDVCS+BH event generator + CLAS12 acceptance (LPSC Grenoble) ≈ -0.4 GeV 2 ≈ 2GeV 2 ≈ 0.17 →  N = 1%- 5%

15. Electromagnetic background Electromagnetic background rates and spectra for the CND have been studied with Gemc (R. De Vita, INFNGE): The background on the CND produced by the beam through electromagnetic interaction in the target consists of neutrals (most likely photons) Total rate ~2 GHz at luminosity of 10 35 cm -2 ·s -1 Maximum rate on a single paddle ~ 22 MHz (1.5 MHz for Edep>100KeV) This background can be reconstructed as a neutron: with a 5 MeV energy threshold the rate is ~ 3 KHz For these “fake” neutrons  <0.1-0.2 → p n < 0.2 GeV/c The actual contamination will depend on the hadronic rate in the forward part of CLAS12 (at 1 KHz, the rate of fake events is 0.4 Hz) , for Edep>5 MeV

16. Technical challenge: TOF resolution & B=5T SiPM - PROS: Insensitive to magnetic field High gain (10 6 ) Good intrinsic timing resolution (30 ps/pixel) Good single photoelectron resolution SiPM - CONS: Very small active surface (1-3 mm 2 ) → small amount of light collected (  TOF ~1/√N phel ) Noise SiPM APD – PROS: insensitive to magnetic field bigger surface than SiPM → more light collected APD – CONS: low gain at room temperature timing resolution? MCP-PMT – PROS: resistant to magnetic field ~1T big surface timing resolution ~ordinary PMT MCP-PMT – CONS: behavior at 5T not yet studied high cost (2K euros/PMT) lifetime? MCP-PMT

17. Measure TOF resolution with 2 standard PMTs Substitute PMT at one end with one SiPM, one APD Test of MCP-PMTs Redo the same measurements with extruded scintillator (FNAL) + WLS fiber (Kuraray) + SiPM (used in IC hodoscope @CLAS, ~ x5 more γ’s/mm 2 ) Test of MCP-PMTs in magnetic field (Saclay, mid November) Tests on photodetectors with cosmic rays at Orsay “Trigger” PMTs (Photonis XP2020) Scintillator bar (BC408) 80cm x 4 cm x 3 cm “Trigger” scintillators (BC408) 1cm thick “Reference PMT” Photonis XP20D0

18. Results from Orsay’s test bench σ 2 test =1/2 (σ 2 test,trig + σ 2 test,ref − σ 2 ref,trig ) Test Ref Trig Test = 1 SiPM Hamamatsu (MPPC 3x3mm 2 ) rise time ~5 ns (> capacitance) more noise than 1x1 mm 2 Test = 1 APD Hamamatsu (10x10 mm 2 )  TOF ~ 1.4 ns high noise, high rise time Test = 1 MCP-PMT Photonis/DEP (two MCPs)  TOF ~ 130 ps will be tested in B field at Saclay (end of November) Thi Nguyen Trung Bernard Genolini S. Pisano J. Pouthas Test = PMT  TOF < 90 ps nphe ~1600 Single pe Test = 1 SiPM Hamamatsu (MPPC 1x1 mm 2 )  TOF ~ 1.8 ns rise time ~ 1 ns nphe ~1 Test = 1 MPPC 1x1mm 2 Extruded scint. + WLS fiber  TOF ~ 1.4 ns WLS -> Width ~ 15 ns

19. Prototype (0 layer) 2 layers Scint. 1 Scint. 2 3 layers Scint. 2Scint. 1 Scint. 3 Simulations with Litrani Pulse shapes Relative light yields Simulations on time resolutions with MCP-PMT

20. Time resolution along the scintillator length Relative light yield taken on the first part of the signals (4ns) Comparison of time resolutions for a same amount of light created 130 ps Adjusted on the Prototype measurements Ongoing:Introduction of the energy deposited by neutrons Conversion in light (including Birks’ law) Simulations on time resolutions with MCP-PMT

21. Plan for the next months R & D: Measurements of time resolution with MCP-PMTs in magnetic field (Saclay, end of November) Construction of prototype (one  bin, 3 radial layers + MCP-PMTs) Measurements with prototype in neutron beam? Simulation: Optimization of number of  bins (signal/background studies with nDVCS/nep  0 generator, Litrani simulations for time resolution vs. paddle size) Inclusion of Birks effect and new estimation of efficiency and resolutions Studies on electromagnetic background to estimate radiation damage on MCP-PMTs, possible adoption of metal shield between CTOF and CND

22. Using scintillator as detector material, detection of nDVCS recoil neutrons with ~10-15% of efficiency and n/  separation for p < 1 GeV/c seems possible from simulations, provided to have ~130 ps of TOF resolution for MIPs The strong magnetic field of the CD and the limited space available call for magnetic-field insensitive and compact photodetectors: MCP-PMTs seem the best candidate, but their timing performances in a high magnetic field still need to be tested CTOF and neutron detector could coexist in one detector, whose first layer can be used as TOF for charged particles when there’s a track in the central tracker, while the full system can be used as neutron detector when there are no tracks in the tracker. Conclusions and outlook nDVCS is a key reaction for the GPD experimental program: measuring its beam-spin asymmetry can give access to E and therefore to the quark orbital angular momentum (via the Ji’s sum rule) A large kinematical coverage is necessary to sample the phase-space, as the BSA is expected to vary strongly → CLAS12 at upgraded JLab is the ideal facility The detection of the recoil neutron is very important to ensure exclusivity, reduce background and keep systematic uncertainties under control The nDVCS recoil neutrons are mostly going at large angles (  n >40°), therefore a neutron detector should be added to the CLAS12 Central Detector, using the available space LoI submitted to JLab’s PAC34 (January 2009) strongly encouraged to submit full proposal

23. Results from Orsay’s test bench Single pe σ 2 test =1/2 (σ 2 test,trig + σ 2 test,ref − σ 2 ref,trig ) Test Ref Trig test = PMT:  TOF < 90 ps nphe ~1600 test = 1 SiPM Hamamatsu MPPC 1x1 mm 2 :  TOF ~ 1.8 ns rise time ~ 1 ns nphe ~1 test = 1 SiPM Hamamatsu MPPC 3x3mm 2 : rise time ~5 ns (> capacitance) more noise than 1x1 mm 2 test = 1 APD Hamamatsu 10x10 mm 2 + IC preamp:  TOF ~ 1.4 ns high noise, high rise time test = 1 MCP-PMT Photonis double-plate:  TOF ~ 105 ps will be tested in B field at Saclay (end of November) simulations underway for lifetime Ongoing: tests with extruded scintillator + WLS fibres +SiPM Thi Nguyen Trung Bernard Genolini S. Pisano J. Pouthas

24. Cost estimate for the CND If MCP-PMTs are the final choice: 3 layers, 30 slices, 2 MCP-PMTs per bar → 180 MCP-PMTs, 2 KEuros each ~ 360 KEuros 90 BC408 plastic scintillator bars → 500 euros each → 45 KEuros Electronics/DAQ ~ 500 Euros/channel → 90 KEuros Total cost ~ 500 KEuros 150 KEuros will come from Italian collaborators (Frascati + Genova) Not clear yet extent of the involvement of Glasgow group Declaration of interest from Edimburgh group