1. A Hadron Blind Detector for the PHENIX Experiment at RHIC Babak Azmoun for the PHENIX Collaboration Brookhaven National Laboratory October 13, 2004

2. 11/13/04Babak Azmoun, BNL2 The PHENIX HBD Team Weizmann Institute of Science; Rehovot, Israel – I.Tserruya, Z. Fraenkel, A. Kozlov, M. Naglis, I. Ravinovich, L. Shekhtman Brookhaven National Lab; NY, USA – B.Azmoun, C.Woody Stony Brook University; NY, USA – A. Milov, A. Drees, T. Hemmick, B. Jacak, A. Sickles University of Tokyo; Tokyo, Japan – T.Gunji, H. Hamagaki, M. Inuzuka, T. Isobe, Y. Morino, S. Oda, K. Ozawa, S. Saito, T. Sakaguchi Waseda University; Tokyo, Japan – Y.Yamaguchi Riken; Wako, Japan – S. Yokkaichi KEK; Tskuba-shi, Japan – S. Sawada

3. 11/13/04Babak Azmoun, BNL3 Outline Outline  Motivation (Physics case)  System specifications and detector concept  GEM +CF4+CsI R&D  Hadron blindness  Hadron rejection (alpha particles - in lab)  Pion rejection (KEK)  Detection efficiency  CsI quantum efficiency and expected N 0  Optical Properties of CF 4 / Effects of gas impurities  Beam test @ PHENIX & KEK  Summary

4. 11/13/04Babak Azmoun, BNL4 Motivation  In ordinary matter, the Standard Model of particle physics states that Chiral symmetry, which is a symmetry between light quark flavors, is normally broken due to constituent quark masses. However, at high temperatures and/or high baryon densities, such as those produced in relativistic heavy ion collisions at RHIC, this symmetry may be restored.  Low mass e + e - pairs are the best probe for Chiral Symmetry Restoration  Effects of CSR may have been seen in CERN SPS results  The same effects are predicted to occur at RHIC, but will be difficult to detect  Nevertheless, the RHIC program would be incomplete without a low mass lepton pair measurement  PHENIX is the only experiment at RHIC that could perform this measurement

5. 11/13/04Babak Azmoun, BNL5 Low Mass Electron Pairs at RHIC Strong enhancement of low-mass pairs persists at RHIC Contribution from open charm becomes significant Possibility to observe in- medium modification of the intermediate vector mesons (  ): Dropping of  invariant mass (Rapp & Wambach) Broadening of vector meson (  ) invariant masses (Brown et.al) Thermal radiation from hadron gas (small…) R. Rapp nucl-th/0204003      *  e + e -

6. 11/13/04Babak Azmoun, BNL6 Experimental Challenges at RHIC Large combinatorial pair background due to copiously produced photon conversions and Dalitz decays : Need rejection factor > 90% of  e + e - and     e + e - Would like to improve S/B by ~ 100 - 200  e + e -    e + e - S/B ~ 1/500 “combinatorial pairs” total background Irreducible charm background all signal charm signal Both members of the electron pair are needed to reconstruct the Dalitz pair or conversion. Single electron pair members contribute to background: Limited by: – Low p T acceptance of outer PHENIX detector: ( p T > 200MeV) – Limited geometrical acceptance of present PHENIX configuration

7. 11/13/04Babak Azmoun, BNL7 Upgrade Concept: Utilize HBD to Identify and Reject background Upgrade Concept: Utilize HBD to Identify and Reject background Hardware * Compensate magnetic field with inner coil to preserve (e + e - ) pair opening angle (foreseen in original design  B  0 for r  50- 60cm) * Compact HBD in inner region (possibly to be complemented by a TPC or other tracking detector in the future). Strategy * Identify signal electrons (low mass pairs) with p>200 MeV in outer PHENIX detectors * Identify low-momentum electrons ( p<200 MeV) in HBD * Reject pair if opening angle < 200 mrad (for a 90% rejection). Specifications * Electron efficiency  90% * Double hit recognition  90% * Modest  rejection ~ 200

8. 11/13/04Babak Azmoun, BNL8 HBD Layout: Windowless Cherenkov Detector e-e- e+e+ Dilepton pair Beam Pipe HBD Gas Volume: Filled with CF 4 Radiator (n CF4 =1.000620, r RADIATOR =50cm) Cherenkov blobs on image plane (  max = cos -1 (1/n)~36 mrad  r BLOB ~3.6cm) Triple-GEM micropattern readout detector, (8 panels per side) Space allocated for services Radiator gas = Working Gas Electron Pairs produce Cherenkov light, but Hadrons with P < 4 GeV/c do not:   (Dileptons)>1/n   (Hadrons) <1/n  Pair Opening Angle

9. 11/13/04Babak Azmoun, BNL9 The HBD Detector element: multi - GEM High gain (CF 4  ~10 4 ) Reduced ion feedback Reflective CsI photocathode (CsI coating on top surface of upper-most GEM foil) No photon feedback Proximity focus  detect blob Low granularity (pad size~10cm 2 ) Concept: Windowless Cherenkov detector. Same radiator and detector gas. preferred option: CF 4 (~50cm) Large bandwidth and large N pe (Bandwidth not limited by a window)

10. 11/13/04Babak Azmoun, BNL10 R&D Program GEM Systematics (using CF 4 ): gain Vs voltage, energy resolution, gain stability, etc. Optical properties of CF 4 : transmittance, effects of gas contaminants CsI: quantum efficiency, compatibility with CF 4, aging, etc. Simulation studies Prototype/Beam Test: N 0 measurement (N pe ), hadron blindness (photoelectron efficiency &  -rejection), performance evaluation of GEM detector in high multiplicity environment. TOPICS

11. 11/13/04Babak Azmoun, BNL11 Gain Curve: Triple GEM with CsI and CF 4 : measured with Fe 55 and with UV lamp Fe 55 x-ray UV lamp GEMs work with CsI and CF4! Pretty good agreement between gain measured with Fe 55 and UV lamp. Gains in excess of 10 4 are easily attainable. Voltage for CF 4 is ~140 V higher than for Ar/CO 2 but slopes are similar for both gases. Gain increases by factor ~3 for ΔV = 20V

12. 11/13/04Babak Azmoun, BNL12 Aging CsI photocathode: * In spite of large ion back-flow there is no dramatic deterioration of the CsI QE. * For a total irradiation of ~10mC/cm 2, the QE drops by only 20%. (The total charge in 10 years of PHENIX operation is conservatively estimated to 1mC/cm 2.) GEM foils: * During the whole R&D period we never observed aging effects (e.g. loss of gain) on the GEM foils. Total irradiation was well in excess of 10mC/cm 2. Stability measurements performed during day 3 (4 mC/cm2 ), day 4 (3 mC/cm2 ), day 5 (2 mC/cm2 ).

13. 11/13/04Babak Azmoun, BNL13 Charge Collection in Drift Gap : Mean Amplitude Rate At E D = 0: - signal drops dramatically as anticipated. - rate also drops dramatically large hadron suppression

14. 11/13/04Babak Azmoun, BNL14 E D = 0 Single Photoelectron Detection Efficiency measure detector response vs E D at fixed gain Very efficient detection of photoelectrons even at negative drift fields !! I PE pA ETET E D (+) ETET EIEI G G G TT TT II DD

15. 11/13/04Babak Azmoun, BNL15 At slightly negative E D, photoelectron detection efficiency is preserved whereas charge collection is largely suppressed. Hadron Blindness (I): UV photons vs.  particles

16. 11/13/04Babak Azmoun, BNL16 CsI absolute QE: set-up Bandwidth: 6.2 – 10.3 eV PMT-2 and CsI have same solid angle C1 optical transparency of mesh (81%) C2 opacity of GEM foil (83.3%) All currents are normalized to I(PMT-1) Rotatable UV mirror QE(CsI) = QE(PMT-2) x I(CsI) / I(PMT-2) x C1 x C2

17. 11/13/04Babak Azmoun, BNL17 CsI QE: results Cherenkov Spectrum: Output~1/ 2 CsI QE tracks Cherenkov Spectrum into the deep UV- VUV regime QE measurement is limited by dynamic range of Spectrometer (due to LiF windows):  Spectrometer cutoff ~114nm  ~10.8eV So… Folding the measured CsI QE into the Cherenkov Spectrum gives an effective N 0 :  Measured in range 6.2-10 eV = 414 cm -1 in CF 4 But… N 0 extrapolated to 11.5 eV (CF 4 cutoff) = 915 cm -1  Optimum expected value ~ 940 cm -1 (Original PHENIX HBD proposal)

18. 11/13/04Babak Azmoun, BNL18 VUV Transmission in the presence of gas Impurities Clear correspondence between water and oxygen levels and the degree of absorbance Tolerance Level: [H2O] max ~ 15-20ppm [O2] max ~ 5ppm  Transmittance: T = I s /I vac (PMT Current Ratio: Sample-gas scan/Vacuum-Ref. Scan) Data in prevailing literature agrees with our data: Interaction Cross Sect. in H2OAttenuation Coefficient in O2

19. 11/13/04Babak Azmoun, BNL19 HBD Beam Test @ KEK  Low number of pe’s is due to absorption within the gas (presumably because of O 2 and H 2 0 impurities)  It was therefore not possible to make a definitive measurement of N 0 E D = +1 KV/cm [cm][15.3] S1 100x45mm S3 50x45mm S22 10x20mm C1C2 [947] [845] PbGl Cal Additional detectors including TPC, CNS-HBD, and an array of Silicon strips detectors 50cm CF4 Radiator Retractable Fe 55 source D 2 lamp HBD GEM   (~98%), e - (~2%) ~ 19 e Pions Electrons E D = -0.3 KV/cm Pions ~ 3e 6 pe (efficiency ≠ 100%) Electrons 35 e=26e (1.38*19e) + 9pe

20. 11/13/04Babak Azmoun, BNL20 HBD Response Simulation Total signal: 62 e = 29 dE/dx + 33 Cherenkov Blob size: single pad 12% more than one pad 88% Normal case, no absorption in CF4, no lamp shadowing, realistic losses and conservative N 0 = 840 cm -1 Total signal: 38 e = 29 (dE/dx) + 9 (Cherenkov ) Blob size: single pad response =78%  very similar to data Includes 20 cm absorption length in CF4, lamp shadowing, realistic losses and conservative N 0 = 840 cm -1

21. 11/13/04Babak Azmoun, BNL21 E D = -0.3 KV/cm   Applying a pion discriminator threshold @ ch.~60, greatly minimizes the number of pions detected, while maintaining ~100% of the pe signal Pion rejection factor will depend on the threshold that one can safely apply without the loss of the photolectron signal, and in turn depends on the number of photoelectrons detected. Pion Rejection Cut @ ADC ch.~60 Expected electron spectrum ~ 33 pe (qualitatively simulated: yield is not to scale) Measured Pion spectrum  blob signal is shared among 3-4 pads  ~10 pe’s per pad  Pion signal is contained within a single pad  The total pe signal will also receive a noise contribution from adding the noise from the 3-4 pads,  the peak separation here may be a bit idealized in this respect

22. 11/13/04Babak Azmoun, BNL22 Test of triple-GEM detector in PHENIX IR Std. Conical, Segmented CERN Foils Active area of pads ~1.0x1.2cm 2 ADC pulse height distr. (Ar/CO2) in Lab ADC Pulse height distr. (CF 4 ) during Full Luminosity A-A Collisions in RHIC  The triple GEM detector performed smoothly within the PHENIX IR using both Ar/CO 2 (70/30) and CF 4 working gases and exhibited no sparking or excessive gain instabilities.  The operation of the GEM and the associated electronics were not hindered by the presence of the ambient magnetic field generated by the central magnet within the IR.  However, in close proximity to the beam pipe (50cm), the detector was sensitive to beam related background.  The fundamental implication of these tests is that the incorporation of a GEM detector among the inner PHENIX detectors is quite feasible when considering how stable the GEMs’ performance was in such a high multiplicity environment. Added Background Gain Variation: +/- 10%, as in the lab PHENIX IR

23. 11/13/04Babak Azmoun, BNL23Summary GEM: High gain/stable operation in pure CF 4. Applicable in the high multiplicity environment of PHENIX Flexible design: provides surface for CsI evaporation, multi-stages are possible, and large active area with thin profile CsI: High Quantum efficiency in VUV region where the Cherenkov spectrum peaks CF 4: High N 0 (Cherenkov radiator)…to be confirmed Transparent in the VUV High gain capability (Detector working gas) Compatible with CsI HBD: Exploits the concept that using GEMs, one can build a detector which is very efficient for detecting Cherenkov photons while being very insensitive to direct ionization from charged tracks. Future: N 0 measurement, Measure CF 4 Scintillation, Test of realistic prototype in beam

24. 11/13/04Babak Azmoun, BNL24 Additional Slides

25. 11/13/04Babak Azmoun, BNL25 Readout Board and Preamps Hybrid Preamps with line drivers ~ ¾” honeycomb Read pads ~ 3x3 cm 2 Preamp signals to shaper + ADC wires GEMs Being developed by BNL Instrumentation Based on IO-535 ± input signal ± 2.5 V output Need almost one full rack for the readout electronics

26. 11/13/04Babak Azmoun, BNL26 Charge Collection in Drift Gap: (I) Am 241  -spectra ETET E D (+) ETET EIEI G G G TT TT II DD Am 241 NPNP

27. 11/13/04Babak Azmoun, BNL27 CF 4 Transparency to UV Photons CF 4 is transparent

28. 11/13/04Babak Azmoun, BNL28 Micropattern Readout Detector: Gas Electron Multiplier (GEM) General: Invented by F. Sauli @ CERN in ~1995 High precision micropattern readout w/ high/stable gain operation Convenient geometry: possibility of multiple stages Applicable in high rate environments Electron avalanche takes place inside GEM holes: NO photon feedback to limit gain Structure: 50  m thick Kapton foil with 5  m copper cladding on both sides which act as electrodes. The foil is perforated with bi-conical shaped holes on the order of 50  m in diameter and a 120  m pitch distance. Operation: Amplification of charge from an incident electron is via an intense electric field inside GEM hole. The field is generated by applying a potential between the electrodes of a few hundred volts. The total charge collected from avalanche is on the order of thousands of times larger than charge of the incident electron, for a multi-stage GEM. G eff ~ exp(  V FOIL ) e- Avalanch (~40kV/cm) Charge transfer/Collection (1kV/cm)

29. 11/13/04Babak Azmoun, BNL29 GEM R&D Detector Box The GEM detector was extensively tested in the lab using both pure CF 4 and Ar/CO 2 gas mixtures for comparison. Set-Up Summary of Measurements Gain measured in various gases: although the volt. potential across CF 4 is higher for the same gain w.r.t. other gases, CF 4 still produces sufficiently high gains (~10 3 -10 4 ). Gain Vs Physical parameters: (e.g., temperature & pressure) vary the gain by up to ~100%. FWHM energy resolution: for CF 4 ~38% in a triple GEM structure, and ~20% for Ar/CO 2 (70/30) Gain Stability in Time: initial gain drift of ~50% is observed, but gain remains stable to within 10% after ~ 1hr. of continual operation (CERN GEM’s) Charge collection and transfer efficiency through GEM stack: measured w.r.t. GEM field config.  optimized field config. Sparking probability in CF4 is low: discharge probability isn’t due to the gas, but to the quality of the GEM foils Ion feedback: high, but harmless to CsI Aging Studies: After accumulating 10mC/cm 2, no significant change (~20%) in the CsI p.c. or the GEM’s behavior Gain uniformity across GEM surface: ~25% variation in gain using 10x10cm GEM foil GEM foils of 3x3 and 10x10 cm 2 produced at CERN

30. 11/13/04Babak Azmoun, BNL30 Ion Back Flow Ions seem to follow the electric field lines. In all cases, ion back-flow is of order 1!!! Mesh GEM1 GEM2 GEM3 PCB 1.5mm 2mm Absorber CsI pA Hg lamp E=0 Fraction of ion back-flow defined here as: I phc / I PCB Independent of E t Depends only on E I (at low E I some charge is collected at the bottom face of GEM3) Independent of gas

31. 11/13/04Babak Azmoun, BNL31 Total Charge in Avalanche in Ar-CO 2 and CF 4 measured with Am 241 When the total charge in CF 4 exceeds 4 x 10 6 a deviation from exponential growth is observed leading to gain saturation when the total charge is ~2 x 10 7. Charge saturation in CF 4 !!!

32. 11/13/04Babak Azmoun, BNL32 Discharge Probability vs. ΔV GEM vs. Gain Stability of operation and absence of discharges in the presence of heavily ionizing particles is crucial for the operation of the HBD. Use Am 241 to simulate heavily ionizing particles. In Ar-CO 2, the discharge threshold is close to the Raether limit (at  10 8 ), whereas in CF 4 the discharge threshold seems to depend on GEM quality and occurs at voltages  V GEM  560-600V CF 4 more robust against discharges than Ar/CO 2. HBD expected to operate at gains < 10 4 i.e. with very comfortable margin below the discharge threshold alpha