Adam Jacholkowski

1
Silicon Tracking at WA97 and NA57

• introduction to WA97/NA57
• setup physics
• silicon pixels hardware aspects
• alignment
• pattern recognition track fit
• vertex finding
• summary final remarks

OUTLINE
2
INTRODUCTION(1)
The NA57 Experiment
(continuation and extension of WA97)
Study of the dependence of hyperon
enhancements on

• Interaction volume - centrality down to
  Nwound 50
• Collision energy - data at two beam momenta -
• 158 and 40
  A GeV/c

• Data samples

3-4 Tb in CASTOR (mass storage)
WA97 p-Be sample used as reference data at 158 A
GeV.
3
INTRODUCTION
INTRODUCTION(2)

• NA57 example of the enhancement study
• result for more see
• G. Bruno plenary talk at QM2004

• Example

systematic error
statistical error
Enhancement def.
INCREASES WITH STRANGENESS CONTENT !
central rapidity (one unit) yield
4
WA97 (predecessor of NA57) set-up in the OMEGA
magnet

• Target Be, Pb
• Beam p, Pb, 158 A GeV/c momentum
• Magnetic field 1.8 T
• Silicon telescope tracking device (7 pixel and
  10 microstrip planes, 5cm x 5cm)
• Pad chambers lever arm
• Scintillation petals lead run centrality trigger
  (40 sinel)
• Multiplicity detectors off-line event centrality
  analysis

L
( 0.5 M pixels)
a
d
L 30 cm d 60 cm (Pb-Pb), 90 cm (p-A) a 40
mrad (Pb-Pb), 48 mrad (p-A)
5
WA97
B0 event
6
NA57 SETUP (Pb - Pb run)
( 1.0 M pixels)
1.4 T
Apparatus
X
Target 1 Pb Scintillator Petals centrality
trigger MSD multiplicity silicon
detector Tracking device silicon pixel
planes (5 x 5 cm2 cross
section) Lever arm double side mstrips
5 cm
7
(No Transcript)
8
(No Transcript)
9
O3 pixel (single) card
O2 pixel plane (box)
5 cm
10
Single pixel cell of the LHC1/?3 chip
LHC1 A semiconductor pixel detector readout chip
with internal, tunable delay providing a binary
pattern of selected events Erik H. M. Heijne et
al Nucl. Instr. Methods A 383 (1996) 55
( RD19 WA97 collaboration)
11
Pixel Maps (full planes) 1
Z
O3Y
(98256 pixel cells)
Y
as seen by the beam (along X)
double length pixels (chip border)
10 000 events
12
Pixel Maps (full planes) 2
O2Z
(73656 pixel cells)
only one card switched ON
10 000 events
13
Dead Time (1)
LDC Local Data Collector (group of pixel cards)
ms
DT proportional to amount of fired pixels
(hits noise)
14
Dead Time (2)
Risk of saturating DAQ in case of high level of
noise, but exaggerated noise suppression ?
lowering planes efficiency
compromise
Importance of masking noisy pixels and
(chips) efficiency monitoring
Data Base
15
MAIN ANALYSIS STEPS

• ALIGNMENT and CALIBRATION ? DATA BASE
• GEOMETRICAL RECONSTRUCTION ? clusters, tracks
• V0 FINDING ? ? and K0 candidates
• CASCADE RECONSTRUCTION (V0s tracks)
• PARTICLE SIGNALS EXTRACTION (selection cuts)
• ? Gold-Plated ntuples
• CORRECTING for ACCEPTANCE and LOSSES ? EVENT-BY-
• EVENT WEIGHTING (and/or de-convolution)
• EXTRAPOLATION TO FULL Pt AND ONE UNIT OF
  rapidity
• NORMALIZATION (beam flux,target)
  YIELDS
• MULTIPLICITY RECONSTRUCTION ? CENTRALITY

THIS TALK
16
ALIGNMENT(1)

• Starting point optical bench survey
  measurements internal pixel ladder positions
  (known from construction)
• Internal pixel alignment cross checks using
  strips tracks (WA97) and exploiting ladder
  overlaps
• Transverse longitudinal alignment using
  straight tracks (special B0 and telescope in
  proton beam runs)
• Small correction tilt angles relative to the
  telescope axis
• Cross-alignment of the Z and Y planes
• Alignment data taken periodically and/or after
  each intervention on the optical bench ( results
  stored in the DB)

17
Alignment (2)
mm
Z
Y
microns
Z
Y-plane tilt test
Single Y (vertical) ladder
18
Parabolic Approximation (used in Pat. Rec.)
10 µ diff.
circle parabola
Sagitta L2 /8? ¼ cm (50 pixels)
19
Polynomial Parameterization Fit (example)
20
TRACK RECONSTRUCTION PRECISION
(3 points parabolic approximation)
s0 pitch/sqrt(12)
B in kGs, p in GeV/c, L in cm
R.L. GLUCKSTERN Nucl. Instr. Methods 24(1963)
381
21
NA57 case ? 0, L 30cm, B 14kGs,X0
30cm/(9x 0.012) 277cm, pitch 50µm
meas
msc.
?p/p meas and msc errors equal at p 12.2 GeV/c
(4.5) p 12.2 ? ?f 0.25 0.17mrad ? 0.3
mrad
22
Pattern Recognition(1)
Semi-combinatorial, using predefined plane
configurations of the compact part only

• Parabolic track model (very good approximation!)
  in the bending plane
• Starting from 3 points (e.g. in the first and the
  last plane in one of the intermediate planes)
  then adding other points lying within the
  predetermined limits relatively to the
  predictions
• Constraints Npoints Nmin (for example 6-7 out
  of 9-11 possible) with a requirement of a minimum
  number of points in each type of pixels (Z or
  Y-like)

23
Pattern Recognition(2)
WA97
x
x
x
(X-)
(p-)
24
Pattern Recognition (3)

• 2D - PR matching in WA97, 3D - PR in NA57
• Hit sharing level controlled according to the
  chosen tolerance ambiguities resolved on the
  basis of ?2
• Track finding efficiency bigger than 95, while
  Kalman Filter e 50 ! ( sparse points
  multiple scattering)
• Ghosts kept at a negligible level (below 1 )
• PR optimization - multiplicity dependent
  (different in Pb-Pb and p-Be)

25
Track Fit(Quintic Spline)
H. Wind Nucl. Instr. Methods 115 (1974) 431
26
ORHION reconstruction programme

• Fortran code developed under Patchy new versions
  kept backward compatible (useful for reprocessing
  !)
• Working both on real and simulated (GEANT MC)
  data
• Internally split into different (main) sub
  processes
• OR steering
• ST pattern
  recognition TF track fit
  XC lever arm track improvement
• V0 secondary vertices finder
• DST-output files of different formats ? input
  for the analysis programs

27
aspect ratio 9 !
p-Be 40 GeV/c ? event
ORHION
cm
O3YO3ZO2YO2ZO3YO2YO2ZO2Y O2Z O3YO3Y
planes sequence
28
aspect ratio 9 !
p-Be 40 GeV/c ? event
ORHION
cm
O3YO3ZO2YO2ZO3YO2YO2ZO2Y O2Z O3YO3Y
planes sequence
29
Vertex Finding

• Primary vertex from secondary tracks
  extrapolation (the µ-strips beam telescope used
  only in the WA97 p-Be run )
• - event-by-event (WA97) or
• - run-by-run (to handle more peripheral
  collisions in NA57)
• V0 finding pairs of oppositely charged tracks
  extrapolated first to a ref. plane, then search
  for the point of nearest approach using helix
  parameterization
• The nearest approach distance a crucial
  parameter in selecting clean signals (removing
  background)
• a typical cut value dmax/2 (alias close)
  0.04 cm

30
HYPERON DETECTION
X
Plus many other associated tracks
Each hyperon (particle) assigned to a centrality
class according to MSD? Ncharged
byp
byL
31
Mass Resolution ?
158 A GeV/c
40 A GeV/c
32
Mass Resolution K0s
158 A GeV/c
40 A GeV/c
33
Summary

• WA97 ? first application of pixel technology (O2
  then O3/LHC1 chips) in conjunction with strips
• NA57 ? pattern recognition and tracking entirely
  based on pixel detectors
• ? WA97 and NA57 experience ? pixel detectors
  a powerful tool for high precision tracking (3D)

34
Final Remarks (1)
Technology still in rapid evolution, future ?
CERN LHC experiments/NA60
35
Message to CBM
Final Remarks (2)
PIXEL TECHNOLOGY is a powerful tool for
physics once good care is taken of all
necessary elements of hardware (calibration) and
software (alignment, noise and efficiency
control) environment
36
THE NA57 COLLABORATION
Physics Department, University of Athens,
Greece Dipartimento IA di Fisica dell'Università
e del Politecnico di Bari and INFN, Bari, Italy
Fysisk Institutt , Universitetet i Bergen,
Bergen, Norway Høgskolen i Bergen, Bergen,
Norway University of Birmingham, Birmingham, UK
Comenius University, Bratislava, Slovakia
University of Catania and INFN, Catania, Italy
CERN, European Laboratory for Particle Physics,
Geneva, Switzerland Institute of Experimental
Physics Slovak Academy of Science, Kosice,
Slovakia P.J. Safárik University, Kosice,
Slovakia Fysisk institutt, Universitetet i Oslo,
Oslo, Norway University of Padua and INFN,
Padua, Italy Collège de France, Paris, France
Institute of Physics, Prague, Czech Republic
University La Sapienza'' and INFN, Rome, Italy
Dipartimento di Scienze Fisiche E.R.
Caianiello'' dell'Università and INFN, Salerno,
Italy State University of St. Petersburg, St.
Petersburg, Russia IReS/ULP, Strasbourg, France
Utrecht University and NIKHEF, Utrecht, The
Netherlands.
37
CENTRAL RAPIDITY YIELD MEASUREMENT
INTRODUCTION
FULL Pt RANGE
ONE UNIT OF RAPIDITY
T from Max-Log-Likelihood
38
Hyperon reconstruction?(cowboy) acceptance
d
?
X-
p-
39
Pixel maps (single cards)
Importance of masking noisy pixels and
(chips) efficiency monitoring
Data Base
(36828 pixel cells)
(49128 pixel cells)
40
EXAMPLE OF ORHION PROCESSING(Pb-Pb 2000
Background)

• Background data representative sample of all
  data
• (each 200th event 5), 24 files
• Parallel running at CERN (Linux Batch) ? overall
• 1 day and 1 night human time
• About 3-4 NCU hours per file
• 1440-1920 NCU hours
  of full 2000
• data ( 230 Mevts ) ORHION processing
• (distributed between all labs !!)

Slightly less for Pb-Pb at 40 GeV/c, then still
less for p-Be (one week)
41
Selection of Hyperons (and KS0)

• Cleaning of the signals via geometrical cuts

close
? impact
X-? vertex
42
WEIGHTING PARTICLES

• A weight is associated with each selected
  particle to
• correct for acceptance, efficiencies and cuts
  ( few thousands)

spread
2 different selections (cuts)
43
WEIGHTING PROCEDURE (2)

• Weights are calculated by Monte Carlo
• - generated hyperons (Ngen) are traced
  through a
• GEANT simulation of the NA57 apparatus
• - track hits are merged with true events
• - resulting events are processed through the
• reconstruction and analysis chain
• - reconstructed hyperons are counted
  (Nrec)
• Simulation thoroughly checked against real data

44
WEIGHT STATISTICS at 158 GeV/c
Particle
3340 2350 2718 6444 936 432 192
weighted
x400 x400 x50 x1 x1 x1
x1
collected
Most expensive cascade particles1-3 NCU hours
on LXPLUS, 10000 ?s and 10000 K0s ? about 40K
NCU hours, alias 1 working month (estimate)

• 2001 p-Be data weighting ( 4000 ?s only) just
  in one week at CERN

45
DECONVOLUTION
An alternative method to weighting (which is
precise but CPU expensive), applicable to high
statistics samples
F. Antinori et al Transverse mass spectra of
strange and multi-strange particles in Pb-Pb
collisions at 158 A GeV/c Eur. Phys. J. C 14,
633-641 (2000)
46
Energy multiplicity dependence
logarithmic scaling