The Hadron Blind Detector

1
The Hadron Blind Detector

• Matt Durham
• Stony Brook University
• WWND 2008

2
Outline

• Dielectron Continuum Measurement at PHENIX
• Limitations from combinatorial background
• The Hadron Blind Detector
• Operation principles
• HBD in Run 7
• Performance and problems
• Current performance studies

Matt Durham, WWND08
3
Dielectron Spectrum Cocktail
Leptons do not interact strongly with the medium
created in RHIC collisions a penetrating probe
Modifications from Chiral symmetry
restoration Modification of vector mesons Thermal
Radiation Charm bound states
Matt Durham, WWND08
4
Dielectron Continuum Measurement
Low-mass Continuum 150 lt mee lt 750
MeV ENHANCEMENT 3.4 0.2(stat) 1.3(sys)
0.7(model)
arXiv 0706.3034
Large systematic error due to low S/B ratio
Matt Durham, WWND08
5
The PHENIX Experiment

• Currently, electrons are tracked by drift chamber
  and pad chamber
• The Ring Imaging Cherenkov Counter is primary
  electron ID device
• Electromagnetic calorimeters measure electron
  energy

p
g
e
e-
2007-12-14
Torsten Dahms - Stony Brook University
5
Matt Durham, WWND08
6
Background Sources?

• Typically only 1 electron from a pair falls
  within the PHENIX acceptance.
• The magnetic field bends the pair in opposite
  directions.
• Some spiral in the magnetic field and never
  reach tracking detectors.

• To eliminate these problems
• Detect electrons in field-free region near
  beampipe
• Need gt90 efficiency

Matt Durham, WWND08
7
Separating Signal from Background
Mass spectrum from pion Dalitz decays peaked
around 2me
Spectrum from photon conversion tightly peaked
around 2me
Heavier meson decays have large opening angles
Opening angle can be used to cut out photon
conversion and Dalitz decays ?MUST BE ABLE TO
DISTINGUISH SINGLE AND DOUBLE HITS
Matt Durham, WWND08
8
Gas Electron Multiplier (GEM)
Thin ( 50 µm) Kapton insulator clad with copper
on each side Holes are chemically etched into the
GEM When a voltage is applied between the
two sides, the high density electric field causes
charged particles to avalanche
140 µm
70 µm

• F. Sauli ,NIM A 386 (1997) 531

300-500V
Gas gain 10-20
C. Altunbas et al, NIM A, 490 (2002) 177-203
Matt Durham, WWND08
9
Traditional GEM operation
Hadron Blind GEM operation
Mesh drifts ionization trail away from GEM stack,
but allows Cherenkov photons Operate in CF4 -
radiator and avalanche gas 400-500V across GEMs
Mesh drifts ionization trail towards GEM
stack Usually operated in Ar/CO2 300-400V across
GEMs
Matt Durham, WWND08
10
Final Design
HBD Gas Volume Filled with CF4 (LRAD50 cm)
Windowless Cherenkov Detector Radiator gas
Avalanche Gas
Cherenkov light forms blobs on an image
plane (rBLOB3cm)
e
e-
55 cm
q Pair Opening Angle
PCB pad readout ( 2x2 cm2)
5 cm
Electrons radiate, but hadrons with P lt 4 GeV/c
do not
CsI photocathode covering GEMs
Space allocated for services (gas and HV)
Triple GEM detectors (10 panels per side)
Dilepton pair
Black Kapton window around beampipe
Matt Durham, WWND08
11
HBD Performance at PHENIX

• Both HBD arms were installed prior to Run-7

HBD West
HBD East
Matt Durham, WWND08
12
First, Bad news Mesh to Top GEM Sparking
Two separate power supplies per stackallows for
easy switch between FB and RB
Optical line of sight
coupling between
modules led to multiple
stacks tripping simultaneously.
HV bias
HV
TRIP!!
R1
R1
R1
R2
R2
R2
R2
Damage made GEM stacks inoperable ?Insufficient
active area to do physics in Run 7. Successful
engineering run

• Led to damage of top GEMs
• Grid patterned burns line up with mesh.

R2
2R2
R1 2.2 MO, R2 10 MO, Vin 3720 V,
dV/GEM 500V, Ichain 150 µA
13
Lecroy HV Problems
Inherent flaw in Lecroy 1471 HV modules causes HV
to spike 600 ms after trip! Voltage between
mesh and GEMs is a convolution of 2 power supply
trips and spikes ?NOT WELL CONTROLLED
TRIP
HV SPIKE
External circuit between GEM stacks and HV
supply Disconnects voltages and discharges stacks
Matt Durham, WWND08
14
Response to Hadrons and Electrons
Electrons and hadrons tracked and identified in
the rest of PHENIX were matched back to tracks in
the HBD
GOOD NEWS The spectrum shows a clear separation
between electron and hadrons tracks. Clear
electron ID Factor of 2-3 increase in S/B
BAD NEWS There is no clear separation between
single and double electron hits
The physics is in the photoelectron statistics
Matt Durham, WWND08
15
Where are the photoelectrons going?
Strong reverse bias Photoelectrons are drawn up
to mesh, instead of down through
holes ?Controlled by drift voltage
HV
Poor extraction from surface Photoelectrons not
extracted from within CsI layer down through
holes, decreasing effective QE ?Controlled
by dV across top GEM
Low Transfer Efficiency Some photoelectrons, and
associated avalanche,
follow field lines which terminate on bottom
side of GEM ?Controlled by transfer gap voltage
We can vary each voltage independently and see
how it affects light signal
Matt Durham, WWND08
16
Light Detection Studies
We can take advantage of CF4 scintillation at 160
nm to measure photoelectron detection
Signals coincident with alpha
Coincidence of SSB and pads
By comparison with 55Fe, get 5 photoelectrons per
trigger
Matt Durham, WWND08
17
Reverse Bias Too Strong?
Adjust drift gap voltage (add in batteries that
float on top GEM voltage)
Look at light signal on pads, coincident with
alpha trigger
In Run 7, HBD operated at -30V reverse bias
ensures hadron blindness, but reduces p.e. yield
NIM A546 (2005) 466-480
Get more light with weaker reverse bias, and
still remain hadron blind
Matt Durham, WWND08
18
Photoelectron Extraction from CsI
Adjust voltage across top GEM ?Increases
gain But we can track that by looking at 55Fe
signal
Slopes of gain curves very similar
Photoelectron collection not affected by GEM
voltage
We run at minimum gain to get good S/N
Matt Durham, WWND08
19
Transfer Efficiency
Adjust first transfer gap voltage while keeping
all else constant
Light signal increases with transfer voltage, up
to 800V 55Fe signal also increases
We are collecting more signal without stressing
GEMS for more gain
Matt Durham, WWND08
20
Transfer Efficiency II
Number of photoelectrons does not increase
25
Effective gain increases by 25
FREE GAIN More signal without increasing stress
on GEMs Similar result for second transfer gap
Matt Durham, WWND08
21
Improvement in Significance
Run 7
ltpegt 15
Assuming Run-4 sized data set
Matt Durham, 1 April 2008
22
Rebuild Status

• HBD West vessel is fully instrumented
• 10 new stacks, fully tested
• Glovebox atmosphere is being heated to drive
  water out of vessel
• Gas flow system at BNL upgraded
• CF4 flow at 5l/m for increased transparency
• 15 gain in light

First vessel expected at BNL July East arm
delivered end of summer
Matt Durham, WWND08
23
Summary

• Dileptons are an interesting probe of diverse
  physics at RHIC
• Severe limitations from combinatorial background
• HBD can significantly reduce the background
• Problems from Run-7 are being addressed and fixed
• Flaw in Lecroy power supply overcome
• Lower operating voltage on GEMs stability
• More light yield greater pair rejection
• Run-9 HBD does physics

Matt Durham, WWND08
24
HBD Collaboration

• Brookhaven National Lab B. Azmoun,
  A.Milov, R. Pisani, T. Sakaguchi, A.
  Sickles, C. Woody
• Columbia University C.-Y. Chi
• Stony Brook University W. Anderson, Z.
  Citron, J. M. Durham, T.Hemmick, J. Kamin,
  V. Pantuev
• Weizmann Institute of Science A. Dubey, Z.
  Fraenkel, A. Kozlov, M. Naglis, I.
  Ravinovich, D. Sharma, I. Tserruya

THANK YOU
Matt Durham, WWND08
25
BACKUPS
26
Photoelectron Statistics Simulation
10

• Generate Poisson distribution around mean number
  of photoelectrons produced
• Double that number to simulate overlapping
  electron hits
• Smear with exponential distribution (avalanche
  process)
• Adjust mean number of photoelectrons to observe
  peak separation

20
30
Matt Durham, WWND08
27
Light from Cube
Data Simulation
Signals coincident with alpha
Sample distribution 5 times and sum
-Represents single hit w/ mean of 25pe Sample
data 10 times and sum -Represents
double hit Sum single and doubles VERY GOOD
AGREEMENT
Matt Durham, WWND08