Construction and expected performance of the Hadron Blind Detector for PHENIX experiment at RHIC

1. Construction and expected performance of the Hadron Blind Detector for PHENIX experiment at RHIC Alexander Milov (for the PHENIX HBD group) XIX International conference on Ulterarelativistic Nucleus-Nucleus Collisions, Shanghai, China Alexander Milov QM2006, Shanghai Nov 15, 2006

2. People in this project • Weizmann Institute of Science (Israel) A.Dubey, Z.Fraenkel, A. Kozlov, M.Naglis, I.Ravinovich, D.Sharma, L.Shekhtman (on leave from BINP), I.Tserruya (project leader) • Stony Brook University (USA) W.Anderson, A.Drees, M.Durham, T.Hemmick, R.Hutter, B.Jacak, J.Kamin • Brookhaven National Lab (USA) B.Azmoun, A.Milov, R.Pisani, T.Sakaguchi, A.Sickles, S.Stoll, C.Woody (Physics) J.Harder, P.O’Connor, V.Radeka, B.Yu(Instrumentation Division) • Columbia University, Nevis Labs (USA) C-Y. Chi • University of Tokyo (Japan) T. Gunji, H.Hamagaki, M.Inuzuka, T.Isobe, Y.Morino, S.X.Oda, K.Ozawa, S.Saito • RIKEN (Japan) S. Yokkaichi • Waseda University (Japan) Y. Yamaguchi • KEK (Japan) S. Sawada Alexander Milov QM2006, Shanghai Nov 15, 2006

3. Why di-electrons? • Effects of chiral symmetry restoration manifest themselves in terms of in-medium modifications of the line shapes of low mass vector mesons (e.g., mass shifts, spectral broadening) • Lepton pairs are unique probes because they provide direct information undistorted by further interactions. Part of the p+p run no bkg. subtraction Entire AuAu run • ρ (m = 770MeV τ ~ 1.3fm/c)  e+e- • ω (m = 782MeV τ ~ 20fm/c)  e+e- • φ (m =1020MeV τ ~ 40fm/c)  e+e- Alexander Milov QM2006, Shanghai Nov 15, 2006

4. Background sources  e+ e - po   e+ e - “combinatorial pairs” total background S/B ~ 1/500 Irreducible charm background signal charm signal • Main source of the background due to external and internal conversions of the photons coming from π0. π0     e+ e- π0  e+ e- • The goal is to reduce the background by a factor of 100 • Distinct pattern of the background producing decays: small inv. mass small opening angle • A rejection factor of >90% on a close pair will reduce the background to an acceptable level. Alexander Milov QM2006, Shanghai Nov 15, 2006

5. The detector concept B≈0 signal electron qpair opening angle Cherenkov blobs e- partner positron needed for rejection e+ ~ 1 m • Proximity Focused Windowless Cherenkov Detector • Radiator gas = Working gas Primary choice pure CF4 n = 1.00062 (=28) L = 50cm Blind to π0with pT<4GeV/c • Radiating particles produce blobs on an image plane (θmax = cos-1(1/n)~36 mrad Blob diameter ~ 3.6 cm) • To preserve the pair opening angle θpair the magnetic field is turned off (compensated) in the detector • Background processes produce 2 close blobs and single electrons only 1 • Image plane: CsI photocathode on top of tiple GEM stack used for electron amplification separated by 90% transparent mesh from the main volume Alexander Milov QM2006, Shanghai Nov 15, 2006

6. Challenges & Solutions • The space where B can be compensated is limited to ~50cm but the number of p.e. must be high enough to allow for effective amplitude analysis of overlapping and distorted blobs. • Match the CsI Q.E.~70% @ 10eV and pure CF4 bandwidth (6-11.5 eV) to get unprecedented N0 ≈840 cm-1 (x6 larger than any e/π RICH ever built!) • The detector has to let all ionizing particles through without seeing them, but pick up single photoelectrons. • Make CsI + GEMs into a new type of semitransparent photocathode such that it a) is sensitive to the ionization reaching its surface from Cherenkov light b) electric field drives MIP ionization back into the gas volume • The detector must be thin to produce little own background but leak tight to keep water away from absorbing UV light. • Windowless design (CF4 without quencher = gaseous radiator = detector gas). • Combine functions of the detector structural elements (pad plane = gas seal) Alexander Milov QM2006, Shanghai Nov 15, 2006

7. The Image plane HV • Start with a GEM • Put a photocathode on top • Electron from Cherenkov light goes into the hole and multiplies • Use more GEMs for larger signal • Pick up the signal on pads • And why is it Hadron Blind? • Mesh with a reverse bias drifts ionization away from multiplication area • Sensitive to UV and blind to traversing ionizing particles Alexander Milov QM2006, Shanghai Nov 15, 2006

8. The Detector Detector is designed and built at the Weizmann Institute • The detector fits under 3%X0 and it is leak tight to keep water out 0.12cc/min (~1 volume per year)! • Readout plane with 1152 hex. pads is made of Kapton in a single sheet to serve as a gas seal FEEs • Each side has 12 (23x27cm2) triple GEM Detectors stacks: Mesh electrode Top gold plated GEM for CsI  Two standard GEMs  pads Side panel Readout plane Mylar window HV terminals Honeycomb panels Triple GEM module with mesh grid Service panel Sealing frame Alexander Milov QM2006, Shanghai Nov 15, 2006

9. Detector elements • Detector construction involves ~350 gluing operations per box • Dead areas are minimized by stretching GEM foils on a 5mm frames and a support in the middle. • GEM positioning elements are produced with 0.5mm mechanical tolerance. Alexander Milov QM2006, Shanghai Nov 15, 2006

10. Detector assembly “Clean Tent” a.k.a. “The Battle Field of Stony Brook” CsI Evaporator and quantum efficiency measurement (on loan from INFN) Laminar Flow Table for GEM assembly High Vacuum GEM storage 6 men-post glove box, continuous gas recirculation & heating O2 < 5 ppm H2O < 10 ppm Class 10-100 ( N < 0.5 mm particles/m3) Alexander Milov QM2006, Shanghai Nov 15, 2006

11. Photocathode production • CsI evaporation station was given on loan to Stony Brook from INFN/ISS Rome Thank you Franco Garibaldi & Italian team! • Produces 4 photocathodes per shot 240 – 450nm of CsI @ 2 nm/sec Vacuum drops to 10-5 Torr and then to 10-7 Torr (water out of the structure). Contaminants measured with RGA • Photocathode Q.E. is measured “in situ” from in 165-200 nm wavelength range over entire area • Photocathodes transported to glove box without exposure to air • 4 small “chicklets” evaporated at same time for full QE control (120-200 nm) Alexander Milov QM2006, Shanghai Nov 15, 2006

12. Some of the production steps GEMs pre-installed for evaporation Photocathode installation chain: removal from transfer box, gain test, installation into the HBD. First module installed in HBD West Alexander Milov QM2006, Shanghai Nov 15, 2006

13. The GEM stacks • GEMs produced at CERN Tested for 500V in air @ CERN Framed & tested @ WIS for gain uniformity Tested at SUNYSB prior to installation Gain uniformity between 5% and 20% • GEM statistics 133 produced (85 standard, 48 Au plated) 65 standard, 37 Au plated passed all tests 48 standard, 24 Au plated installed GEMs combined into stacks are matched to minimize gain variation over the entire detector • All GEMs pumped for many days under 10-6 Torr prior to installation into detector 20% 5% Alexander Milov QM2006, Shanghai Nov 15, 2006

14. GEM gain stability • During gain mapping, a single pad is irradiated with a 8kHz 55Fe source for ~20 min. Then all other pads are measured (~1.5h) and the source is returned to the starting pad. • Gain is observed to initially rise and then reach a plateau. Rise can be from few % to almost a factor of 2. • Further study show that the gain increase is rate dependent (10-30%) • This does not impose a problem for GEM operation at PHENIX GEMs will reach operating plateau in a few hours Rates are lower then during mapping Secondary rise 1.5 Initial Rise Alexander Milov QM2006, Shanghai Nov 15, 2006

15. Photocathode quality 72 36 Number of photoelectrons • Q.E. needs to distinguish a single electron from a pair. • Absolute Q.E. must be continuously controlled and preserved. At the production stage During transportation and installation During physics data taking • At the production stage the Q.E. is as high as measured in R&D stage and uniform 27 cm Flat position dependence Alexander Milov QM2006, Shanghai Nov 15, 2006

16. Gas transparency Monochromator (120-200 nm) is a part of the HBD gas system Movable mirror D2 lamp H2O & O2 must be kept at the few ppm level to avoid absorption in the gas Turbopump Lamp Monitor Gas Cell Monitor Measure photocathode current of CsI PMTs Heaters are installed on each detector to drive out water from GEMs and sides of detector vessel Alexander Milov QM2006, Shanghai Nov 15, 2006

17. Full scale prototype test Tested in PHENIX with p-p collisions at RHIC April-June ‘06 Pulse height, reverse bias • Full scale detector prototype: 1 GEM + CsI stack module installed in the volume 68 readout channels full readout chain • Pure CF4 gas system • LVL2 triggers to enrich e-sample electrons hadrons e/π rejection ~85% at εe ~90 % Cluster size, reverse bias MIP electrons hadrons Forward Bias+Landau Reverse Bias Alexander Milov QM2006, Shanghai Nov 15, 2006

18. Now HBD East (back side) Installed 10/19/06 HBD West (front side) Installed 9/4/06 Alexander Milov QM2006, Shanghai Nov 15, 2006

19. Summary • The HBD will provide a unique capability for PHENIX to measure low mass electron pairs in heavy ion collisions at RHIC • This detector incorporates several new technologies (GEMs, CsI photocathodes, operation in pure CF4, windowless design) to achieve unprecedented performance in photon detection N0~840 cm-1 • The operating requirements are very demanding in terms of leak tightness and gas purity, but we feel they can be achieved • Tests with the full scale prototype were very encouraging and demonstrated the hadron blindness properties of the detector. • The final detector is now installed in PHENIX and ready for commissioning and data taking during the upcoming run at RHIC Alexander Milov QM2006, Shanghai Nov 15, 2006

20. BACKUPS Alexander Milov QM2006, Shanghai Nov 15, 2006

21. Challenges & Solutions • The space where B can be compensated is limited to ~50cm but the number of p.e. must be high enough to allow for effective amplitude analysis of overlapping and distorted blobs. • Match the CsI Q.E.~70% @ 10eV and pure CF4 bandwidth (6-11.5 eV) to get unprecedented N0 ≈840 cm-1 (x6 larger than any e/π RICH ever built!) • The detector has to let all ionizing particles through without seeing them, but pick up single photoelectrons. • Make CsI + GEMs into a new type of semitransparent photocathode, which a) does not have usual losses for such type of photocathode b) allows multi-stage multiplication to follow it. • The detector must be thin to produce little own background but leak tight to keep water away from absorbing UV light. • Go to windowless design by using CF4 without quenching gas both as a radiator and working gas due to the fact that GEMs have no photon feedback Alexander Milov QM2006, Shanghai Nov 15, 2006

22. PHENIX now e+ e- e- e+ ~12 m Alexander Milov QM2006, Shanghai Nov 15, 2006

23. HBD parameters Acceptance nominal location (r=5cm) || ≤0.45, =135o retracted location (r=22 cm) || ≤0.36, =110o GEM size (,z) 23 x 27 cm2 Number of detector modules per arm 12 Frame W:5mm T:0.3mm Hexagonal pad size a = 15.6 mm Number of pads per arm 1152 Dead area within central arm acceptance 6% Radiation length (central arm acceptance) box: 0.92%, gas: 0.54% Weight per arm (including accessories) <10 kg Alexander Milov QM2006, Shanghai Nov 15, 2006

24. Readout chain 15 mm 19 mm Preamp (BNL IO-1195) 2304 channels total Differential output Noise on the bench looks very good Gaussian w/o long tails 3s cut  < 1% hit probability Alexander Milov QM2006, Shanghai Nov 15, 2006

25. Run Plan • Run 7 (Dec ‘06 – June ’07) • ~ 4 weeks commissioning with Au x Au beams at sNN = 200 GeV • 10 weeks data taking with Au x Au at sNN = 200 GeV • 10 weeks data taking with polarized p-p beams at s=200 GeV • Run 8 (Fall ’07 – Summer ’08) • 15 weeks d-Au at sNN = 200 GeV • 10 weeks polarized p-p at s=200 GeV • Run 9 (Fall ’08 – Summer ’09) • 10-15 weeks heavy ions (different energies and possibly species) • 15-10 weeks polarized p-p at s=500 GeV (including commissioning) • Run 10 (Fall ’09 – Summer ’09) • HBD is removed in order to install new silicon vertex detector in • PHENIX Alexander Milov QM2006, Shanghai Nov 15, 2006

26. Photocathode and gas. • Photocathode: • CsI is an obvious choice. • We are using INFN built evaporator, currently at Stony Brook to do this project. • High area, • High vacuum, • In-situ Q.E. control, • Zero exposure to open air. • Gas CF4 (was not really known): • Has high electron extraction probability • Has avalanche self quenching mechanism • Gas CF4 (well known): • Transparent up to 11.5 eV, makes perfect match to CsI • Is a good detector gas. Alexander Milov QM2006, Shanghai Nov 15, 2006

27. The design. 72 36 Number of photoelectrons Made of 2 units with R~60cm, the volume is filled with CF4 magnetic field is turned off Electrons emit Cherenkov light Cherenkov light is registered by 12 photo-detectors in each unit Signal is read out by 94 pads in each unit, pad size ~ size of a circle Accumulating ~36 photoelectrons from each primary electron, while most other operational RICHes have ~15 or less. High statistics allows to separate 2 close electrons even if their signals overlay! Alexander Milov QM2006, Shanghai Nov 15, 2006

28. Event display (simulation). Alexander Milov QM2006, Shanghai Nov 15, 2006

29. Background sources? • In the decays contributing to the background: • π0  e+ e- γ • π0  γγ  e+ e- γ • Only one electron is detected in PHENIX and another is lost • To cut the background we need a new detector such that: • It sees only electrons • Located at the origin • It does not produce its own background (is thin) • … • … • … ~12 m Alexander Milov QM2006, Shanghai Nov 15, 2006

30. What does it look like • All raw materials (FR4 sheets, honeycomb, HV resistors, HV connectors) ordered and most of them in house • Detector box design fully completed • Jig design underway • Small parts (insert, pins, screws, HV holders..) in the shops • Detector construction to start Nov. 1st • PCB design almost complete • Detailed construction schedule foresees shipment of boxes to SUNY in January 2006. Alexander Milov QM2006, Shanghai Nov 15, 2006

31. Mechanical parts and PCB. PCB final design. Quick MC shows no difference with standard cells Entrance window frames are ready, the window itself to be tight between them Alexander Milov QM2006, Shanghai Nov 15, 2006