Towards a Pulseshape

1
Towards a Pulseshape Simulation / Analysis
Kevin Kröninger, MPI für Physik GERDA
Collaboration Meeting, DUBNA, 06/27 06/29/2005
2
Outline
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
3
SIMULATION
4
Simulation Overview I
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
5
Simulation Overview II

• What happens inside the crystal?
• Local energy depositions translate into the
  creation of electron-hole pairs
• with Edep deposited enery
• Eeh 2.95 eV at 80 K in Ge
• Egap 0.73 eV at 80 K ? ¾ of energy loss to
  phonons
• Corresponds to approximatly 600,000 e/h-pairs at
  2 MeV
• Due to bias voltage electrons and holes drift
  towards electrodes
• (direction depends on charge and detector type)
  
• Charge carriers induce mirror charges at the
  electrodes
• while drifting

ltNgt Edep / Eeh
? SIGNAL
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
6
Drifting Field / Bias Voltage I

• In order to move charge carriers an electric
  field is needed
• Calculate field numerically
• 3-D grid with spatial resolution of 0.5 mm
• Define Dirichlet boundary conditions (voltage,
  ground)
• ? depend on geometry (true coxial? non-true
  coxial? etc.)
• So far no depletion regions, zero charge
  density inside crystal,
• no trapping
• Solve Poisson equation ?f 0 inside crystal
  using a Gauss-Seidel
• method with simultaneous overrelaxiation
• Need approximatly 1000 iterations to get stable
  field
• Electric field calculated as gradient of
  potential

Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
7
Drifting Field / Bias Voltage II
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
8
Drifting Field / Bias Voltage III

• Example non-true coaxial n-type detector

Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
9
Drifting Process
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
10
Mirror Charges Ramos Theorem

• Ramos Theorem
• Induced charge Q on electrode by point-like
  charge q is given by
• Calculation of weighting field
• Set all space charges to zero potential
• Set electrode under investigation to unit
  potential
• Ground all other electrodes
• Solve Poisson equation for this setup (use
  numerical method explained)

Q induced charge q moving pointlike charge f0
weighting potential
Q - q f0(x)
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
11
Weighting Fields I
x
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
12
Weighting Fields II

• Example true coaxial detector with 6 f- and 3
  z-segments

f 0
f 90
f 180
f 270
z
(Slices in f showing ?-z plane)
y
x
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
13
Preamp / DAQ

• Drift and mirror charges yield charge as
  function of time
• Preamp decreases accumulated charge
  exponentially,
• fold in gaussian transfer function with 35 ns
  width
• DAQ samples with 75 MHz ? time window 13.3 ns
• Example

(signal after drift, preamp and DAQ)
(signal after drift)
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
14
Setups / Geometries / Eventdisplays I

• Full simulation of non-true coaxial detector

Charge
Charge
electrode
core
Time
Current
Current
electrode
core
Time
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
15
Setups / Geometries / Eventdisplays II
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
16
Setups / Geometries / Eventdisplays III

• Full simulation of true-coxial 18-fold segmented
  detector

electrodes
core
Charge
Time
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
17
Analysis Approach
18
Pulseshape Analysis in MC Spatial Resolution I

• Is it possible to obtain spatial information of
  hits from
• pulseshapes? In principal YES!
• Risetime of signal (10 - 90 amplitude) is
  correlated with radius
• of hit due to different drift times of
  electrons and holes
• Relative amplitude of neighboring segments is
  correlated to angle
• Events with more than one hit in detector give
  ambiguities
• Studied in Monte Carlo with 2-D 6-fold segment
  detector,
• no DAQ, no sampling

Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
19
Pulseshape Analysis in MC Spatial Resolution II

• Spatial information of radius and angle

Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
20
Pulseshape Analysis SSE/MSE Discrimination I

• Do 0?ßß signals differ from background signals?
• Background mainly photons that Compton-scatter
  multiple hits
• in crystal ? Multisite events (MSE)
• Signal due to electrons with small mean free
  path localized energy
• deposition ? Singlesite events (SSE)
• Expect two shoulders at most from SSE and more
  from MSE
• Count number of shoulders in current
• Apply mexican hat filter with integral 0 and
  different widths (IGEX method)
• Count number of shoulders 2 SSE
• gt2 MSE

Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
21
Pulseshape Analysis SSE/MSE Discrimination II
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
22
Pulseshape Analysis SSE/MSE Discrimination III

• Fraction of SSE and MSE for different filter
  widths

Identified as SSE
Identified as MSE
Fraction of Events
Fraction of Events
Filter width
Filter width
Separation of SSE/MSE in principle possible,
combine with information from neighboring
segments
SSE MSE
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
23
Data to Monte Carlo Comparison
24
Data to Monte Carlo Comparison I

• Data from teststand (see X. Liu)
• Later on used for SSE selection

Work in progress
Source
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
25
Data to Monte Carlo Comparison II

• Teststand data vs. Monte Carlo

Work in progress
Energy MeV
Energy MeV

• General agreement
• No finetuning yet
• Next pulseshapes without
• any additional selection criteria

Energy MeV
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
26
Data to Monte Carlo Comparison III

• Comparison of pulseshapes

Work in progress
Data
Monte Carlo
Charge
Charge
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
27
Data to Monte Carlo Comparison IV

• Comparison of pulseshapes

Work in progress
Data
Monte Carlo
Current
Current
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
28
Data to Monte Carlo Comparison V

• Comparison of pulseshapes

Work in progress
Current
Charge
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
29
Data to Monte Carlo Comparison VI

• Comparison of pulseshapes

Work in progress
Current amplitude
Charge amplitude
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
30
Data to Monte Carlo Comparison VII

• Comparison of pulseshapes

Work in progress
Risetime ns
Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005
31
Conclusion

• First approach towards a simulation of
  pulseshapes
• Different geometries / fields available
• Package available and linked to MaGe
• Pulseshape analysis to further reduce background
  via
• SSE/MSE identification is feasible ? need
  sampling rate
• as large as possible (1 GHz ? 1 ns possible?)
• Data to Monte Carlo comparison using teststand
  data
• yields coarse agreement ? finetune parameters
  of simulation

Kevin Kröninger, MPI München GERDA
Collaboration Meeting DUBNA, 06/27 06/29/2005