1. RICH 2010, Cassis, May 2-7, 2010 Itzhak Tserruya
The Hadron Blind Detector of the PHENIX Experiment at RHIC
RICH 2010, Cassis, May 2-7, 2010 Itzhak Tserruya

2. Outline Motivation Detector Concept Design and Construction Operation
Performance
Summary
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

3. 1. Motivation
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

4. combinatorial e+ e - pair
Motivation (I)
Electron pairs (or dileptons in general) are unique probes to study the hot and dense matter formed in relativistic heavy ion collisions at RHIC:
best probe for chiral symmetry restoration and in-medium modifications of light vector mesons , ω and 
sensitive probe for thermal radiation:
QGP qqbar  *  e+e-
HG +-    *  e+e-
Experimental challenge:
huge combinatorial background created by e+e- pairs from copiously produced 0 Dalitz decay and  conversions.
 e+ e -
po   e+ e -
combinatorial e+ e - pair
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

5. Baseline PHENIX detector: S/B ratio of 1/200 at m = 500 MEV/c2
Background sources in PHENIX
Main sources contributing to the combinatorial background:
0  e+ e- 
 e+ e- 
It often happens that only one electron is detected in PHENIX and the other is lost due to:
limited geometrical acceptance
low pT particle curling in the magnetic field or not reconstructed.
~12 m
Baseline PHENIX detector: S/B ratio of 1/200 at m = 500 MEV/c2
Main goal of the HBD: recognize tracks from conversions and 0 Dalitz decays thereby considerably reducing the combinatorial background

6. 2. Concept
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

7. Main task of the HBD: distinguish single vs double e-hit
Upgrade Concept
Strategy
Exploit the fact that 0 Dalitz decays and  conversions have a very small opening angle to identify them.
Create a field free region close to the vertex to preserve the opening angle of close pairs.
Identify electrons in the field free region
reject close pairs.
Hardware realization
* HBD in inner region
* Inner coil  B0 for r  60cm
Main task of the HBD: distinguish single vs double e-hit
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

8. HBD Concept HBD concept: ♣ windowless Cherenkov detector (L=50cm)
♣ CF4 as radiator and detector gas
♣ Proximity focus:
detect circular blob not ring E
hadron
UV-photon
50 cm
~1 cm
CF4 radiator
detector element
5 cm
beam axis
Detector element:
♣ Triple GEM stack with pad readout
♣ CsI reflective photocathode evaporated on top face of top GEM
♣ RB mode of operation to repel ionization charge
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

9. Very attractive features…
Unprecedented N0
Bandwidth eV ( nm)  N0 ≈ 800 cm-1 in an ideal detector with no losses
Reflective photocathode  No photon feedback
Hadron blind:
Most of the ionization charge in the drift region is repelled away from the GEM stack and
collected by the mesh
Hexagonal pads with size (a=15.5 mm area =6.2 cm2 )
comparable to Cherenkov blob size (10.2 cm2)
hadrons: single pad hit, electrons: 2-3 pads per hit
Low granularity
~1000 pads per central arm acceptance
Low gain
primary charge of e/pad  gain of 5x103 is enough
…but many open questions  extensive R&D
A. Kozlov et al, NIM A523 (2004) Z. Fraenkel et al, NIM A546 (2005) 466,

10. 3. Design and Construction
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

11. The Detector Detector designed and built at the Weizmann Institute
Each arm has 12 (23x27cm2) triple GEM stacks:
Mesh electrode
Top gold plated GEM for CsI & two standard Gems
Pad electrode
Two identical arms
All panels made of honeycomb & FR4 structure
FEEs
Readout plane with 1152 hexagonal pads made of Kapton in one single sheet to serve as gas seal
Low material budget: vessel 0.92%, gas 0.54%,
electronics ~1.5%  total under 3% X0.
Side panel
Readout plane
~ 350 gluing operations per arm
Leak rate 0.12cc/min (~1 vol /year)!
Mylar window
HV terminals
Low dead/inactive area of 6%
(0.5 mm tolerance between
adjacent modules)
Service panel
Triple GEM module with mesh grid
Sealing frame

12. HBD Construction Jigs 6 active panels glued together
Jig #1: Glue active panels together and to the PCB cathode board
Jig #2: Glue the rest of the panels to complete the HBD box
6 active panels glued together

13. The people behind the detector

14. Detector assembly
CsI evaporation and final detector assembly in clean tent at Stony Brook
Laminar Flow Table for GEM assembly
CsI Evaporator and quantum efficiency measurement
(on loan from INFN)
Can make up to 4 photocathodes in one shot
High Vacuum GEM storage
6 men-post glove box, continuous gas recirculation & heating
O2 < 5 ppm
H2O < 10 ppm
Class ( N < 0.5 mm particles/m3)
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

15. CsI evaporation facility
Absolute QE of CsI photocathodes
Excellent reproducibility.
Excellent stability
4 photocathodes produced per shot together with chicklets for QE monitorin
For each GEM 3 measurements are taken at 160 nm across X axis

16. The SB plant
the crew
Jason
Gabby
Greg
Ben
Matt
Ben-II

17. 4. Operation
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

18. HBD – short history The HBD was commissioned during the 2007 RHIC run
The detector encountered severe high voltage problems
which ultimately damaged many GEMs:
Minor GEM sparks induced larger, more damaging sparks due to large stored energy in filter capacitors
Spark in one module would induce sparks in other modules by optical coupling
Problem exacerbated by LeCroy 1471N PS that reapplied HV after a trip causing HV spurious HV spikes
The main problem with GEM sparking was due to dust !
Detector rebuilt (minimum exposure to glove box, Zener diodes, relay box )
Detector successfully operated in Run-9 (pp collisions) and in Run-10 (Au+Au collisions – still ongoing) each one about 6m long.
Fe55 x-ray
UV lamp
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

19. כ"ב/כסלו/תשע"ט
Readout and noise
Excellent electronic noise in all modules. Typically  = 1.5 ADC counts corresponding to 0.15 fC or 0.2 p.e. at a gain of 5000.
Allowed to operate the detector at a low gain of 3000 to 5000.
The entire readout chain worked reliably and smoothly (BNL & Columbia Univ)
Mean
Pedestal rms vs HV
Sigma
Mean
Sigma

20. Gain determination using scintillation hits
כ"ב/כסלו/תשע"ט
Exploit scintillation hits (identified as single pad hits not belonging to tracks) to determine the gain on a pad by pad basis.
Forward Bias
Scintillation
Ionization
Gain determination:
Fit scintillation component with an exp fctn:
1/slope = G . <m>
<m> = avrg nr of scintillation photons in a fired pad)
In pp collisions <m>  1
In Au +Au collisions, assuming the nr of scintillation photons per pad follows a Poisson distribution:
<m> =
Zoom
Reverse Bias
Scintillation: unchanged
Ionization: suppressed
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

21. Correction for P/T Variation
Gain monitoring: gain is calculated for each module and each run.
Gain variations due to p/T changes, automatically compensated by varying the operating HV in 5 pre-determined for p/T windows. T1 T2 T3
CONFIG
CHANGE
APPLY
HV CHANGE
shown above
IR ACCE SS
RUNS PROCESSED:
ES3
WN5

22. Adjusting the Reverse Bias using scintillation light
Requires high precision in adjusting the mesh voltage with wrt the GEM stack, to preserve the p.e. collection efficiency
Requires setting ~ 4KV PS to ~ 5 V precision ( 0.1 %)
Operating Point
Electrons
hadrons
Exploit the scintillation that remains unchanged when switching from FB to RB
Scan voltage between mesh and top GEM for each module
Method invented in Run-9
Simple, fast and precise
ES1 =-10V
Red (+5V), black (0V)‏
green(-5V),blue(-10V),
rest(-15V and lower)‏
Hits per event
Pulse height [ADC counts]

23. GasTransmission
Continuously monitored using a monochromator system – Recent scan on April 20
Gas flowing in recirculation mode, with scrubbing at 4.5 lpm
Gas transmission stable over the entire run
Input gas upper 90%
Output gas lower 90% 29

24. 5. Performance
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

25. Position Resolution
כ"ב/כסלו/תשע"ט
Hadron tracks reconstructed in central arms projected to HBD.
Position resolution:
z ≈  ≈ 1.5 cm
Dictated by:
pad (hexagon) size:
a = 1.55 cm 2a/√12 = 0.9 cm
vertex resolution ~ 1 cm
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

26. Hadron Blindness
כ"ב/כסלו/תשע"ט
Hadron suppression illustrated by
comparing hadron spectra in FB and RB
(same number of central tracks)
Pulse height
Hadron rejection factor
Pulse height
Strong suppression of hadron signal at reverse drift field
Larger rejection by combining pulse height and cluster size
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

27. Electron - hadron separation in RB
כ"ב/כסלו/תשע"ט
HBD in RB mode
Reconstructed tracks in Central Arms are projected to HBD
Single electrons selected by Dalitz open pairs (m< 150 MeV/c2)
Hadrons dE/dx signal in the small (~ 100 mm) region above the top GEM and in the first transfer gap very small wrt electron Cherenkov signal

28. Single electron detection efficiency
כ"ב/כסלו/תשע"ט
Single electron detection efficiency
1. Single electron efficiency using a sample of open Dalitz decays (V cut rejects conversions; not very effective for conversions in the radiator gas ) :   90 %
2. Single electron efficiency derived from the J/psi region ( m > 3.5 GeV/c2 after background subtraction):  = 90.6  9.9 %

29. Single vs double electron separation
כ"ב/כסלו/תשע"ט
Fully reconstructed 0 Dalitz pairs (m < 150 MeV/c2) in the central arms Matched to HBD into two separate clusters or one single cluster.
Double electron response (0 Dalitz close pairs)
Single electron response (0 Dalitz open pairs)
~ 40 p.e. per two electron track
~ 22 p.e. per single electron track
Agrees with expected yield taking into account p.e. collection efficiency and transmission loss in the gas
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

30. Combinatorial Background rejection
כ"ב/כסלו/תשע"ט
With the performance results shown in the previous slides we should be very close to the design values for the CB rejection.
**Preliminary ** rejection numbers:
- matching to HBD  2.2
- double hit cut
- close hit cut
- single pad cluster cut ~2
6.5

31. Summary
A novel HBD detector has been developed, installed and is successfully being operated in the PHENIX set-up
Analysis of data taken in pp and Au+Au collisions show:
Clear separation between e and h
Expected Hadron rejection factor
Excellent electron detection efficiency
Good separation of single vs double electron hits
Considerable reduction of the combinatorial background
Looking forward to the physics
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010

32. HBD Team Brookhaven National Lab
B. Azmoun, R. Pisani, T. Sakaguchi, A.Sickles, S. Stoll, C. Woody
Columbia University (Nevis Labs)
C-Y. Chi
Stony Brook University
Z. Citron, M. Connors, M. Durham, T. Hemmick, J. Kamin, B. Lewis, V. Pantuev, M. Prossl, J.Sun
University of California Riverside
A. Iordanova, S. Rolnick
Weizmann Institute of Science
Z. Fraenkel*, A. Kozlov, A. Milov, M. Naglis, I. Ravinovich, D. Sharma, I.Tserruya
* deceased

33. Event display
Itzhak Tserruya
RICH 2010, Cassis, May 2-7, 2010