Outline

1
Measuring plane efficiencies
Giuseppe E Bruno Università di Bari and INFN -
Italy

• Outline
• Aim of such a measurement
• A look into the CMS framework
• A first proposal

2
Plane efficiencies whats for ?

• from data to a paper
• data taking ? raw data
• reconstruction ? ESD ? standard AOD
• analysis
• standard AOD? user AOD
• selection of candidates
• computation of corrections for acceptance and
  reconstruction inefficiencies
• application (e.g., with a deconvolution) of the
  corrections (and flux factors for normalisation)
  to the raw distributions to get the physical
  distributions
• estimate of the systematics errors (here
  corrections always play a role)
• writing of the paper

3
Plane efficiencies whats for ?

• computation of corrections for acceptance and
  reconstruction inefficiencies

A framework is being developed for this S.
Arcelli (Bologna), I. Kraus (CERN), A.
Mastroserio (Bari), R. Vernet (Catania)
This framework relies on the Monte Carlo
simulation (AliRoot) the detector response
functions are an ingredient of the simulation
4
Plane efficiencies whats for ?

• In order to compute properly the corrections, the
  Monte Carlo description of our detectors should
  be as realistic as possible.
• fast simulation (digits created according to
  tables of probabilities ? plane efficiencies)
• response models driven by physics (e.g. in ITS,
  the Gaussian diffusion, plus coupling effects,
  etc.)

Can be used in few cases, and for
systematic errors evaluation
Advantages fast, history (deterioration) of the
detectors reproduced naturally Drawbacks poor
description of the detector spatial precision,
digit correlations ?, some analyses sensitive
to the chosen granularity for eff. determination
Standard case
The measured plane efficiencies would be THE
REFERENCES of the models
5

• The aim of this work is to measure the
  efficiencies of the layers used for tracking,
  with the tracks themself
• ITS is the goal
• Eventually, the framework would be extended to
  other detectors (TRD, TOF, )
• TPC would do it differently (not using tracks)
• Of course, each layer will be divided in basic
  blocks (e.g., the chip for SPD)

• An estimate of the Plane efficiency can be
  done also looking at the map of hits (i.e. the
  distribution of digits)
• It will give the relative efficency within a
  block
• I exspect it to be one of the QA measurements

6
A look into CMS code

• It is not a framework, they have a code for
  cosmics
• CMS basically has one type of detector for
  tracking the silicon micro-strips (the 3 pixel
  layers neither have been installed nor are in the
  code for plane efficiency)
• Their basic block is one module barrel 8K,
  tot.15K
• A space point is searched to be compatible with
  the track prediction, obtained without the
  detector under study
• To avoid border ambiguities, only tracks
  impacting on the 92 (TIB) and 87 (TOB)
  innermost area of the detectors are considered ?
  the resulting efficiency is extrapolated to the
  full detector

7
The Method

• The Layer under study is removed from the
  tracking
• The interpolation of the track on that layer is
  used to evaluate in which module the hit is
  expected.
• The all rec hit collection is used to evaluate
  if the hit is in the expected module (S) or not
  (B).

Track RecHit all RecHit
S SB
Eff
The layer under study is removed from the tracking
8
Further constraints to evaluate efficiency
To evaluate the module efficiency the active area
of the modules has been reduced to avoid
artificial inefficiency due to the uncertainty
to the module assignment in the border of active
area and in the bonding region.
TOB Modules
Bonding
5mm
10mm
Area considered for modules
9
A look into CMS
The innermost tracking is accomplished with three
layers of silicon pixel detectors (at radii of
4.4, 7.3 and 10.2 cm) with a total area of
approximately 1m2, composed of 66M 100150µm2
area pixels. The remaining tracking layers are
composed of 9.3M single- and double-sided silicon
microstrip detectors covering a total of 200m2 of
detectors, organised in an inner barrel (TIB)
with 4 layers within the 2050 cm radius range,
an outer barrel (TOB) with 6 layers within the
55-120 cm radius range, and two endcap detectors
(TEC and TID).
10
A first proposal for ALICE ITS

• The plane efficiency will be measured, using high
  quality tracks by removing one layer at the time
  from the tracker
• In AliITSTrackerV2 such a functionality was
  foreseen
• Implementing it in AliITSTrackerMI is under study
• At the moment the tracker is aware only of
  unsensitive regions those get assigned virtual
  rec-points (Q0)
• The tracker is still not aware about dead/bad
  channels (to be read from DB)
• Proposal a unique tool to remove regions from
  a layer e.g. dead channels, unsensitive regions,
  the full layer.

11
A first proposal for ALICE ITS

• Im assuming to have access to the rec-points but
  not to the raw digits (this may have an influence
  for the SSD, if we want a resolution better than
  module by module)
• The basic blocks (granularity) should be defined
  according to advices from detector experts and
  PWGs, taking into account the required statistics
  for evaluating the efficiencies within a given
  tollerance
• (an estimates available in the next slides)
• The output (efficiency tables) would be written
  into the DB for periods of validity
• framework functionality automatic update when
  needed
• Dead and noisy channels
• not a goal of this work
• the framework aware of them

12
Proposal for SPD
D. Elia (Ba)

• Plane efficiency to be avaluated chip by chip
  with a statistical error of 0.5
• Number of required high quality tracks by
  assuming Eff90
• layer 1 400 chips ? 1.8M
• layer 2 800 chips ? 3.5M

13
Proposal for SDD
F. Prino (To)

• layer 3
• 14 ladders
• 1 ladder6 detector
• tot. 84 detector
• layer 4
• 22 ladders
• 1 ladder8 detector
• tot. 176 detector
• each detector divided in 8(chips)2(drift
  direction)
• layer 3 1344 zones
• layer 4 2288 zones

? 6.0M tracks
? 10.2M tracks
quite demanding !
14
Proposal for SSD
E.Fragiacomo (Ts)

• layer 5 34(ladders)22748 modules
• layer 6 38(ladders)25950 modules
• 1 module6(p-side)6(n-side)12 chips
• There is not simple segmentation if we want to go
  finer than the module
• chip by chip ?
• too many elements (layer 6 11400) !
• stereo geometry how to share the inefficiency
  between n- and psides ?
• The module looks the most reasonable choice

15
Proposed segmentation

• pixel chip by chip n. of tracks (M)
• layer 1 400 zones ? 1.8
• layer 2 800 zones ? 3.5
• drift chip by chip (/2)
• layer 3 672 zones 3.0 (6.0)
• layer 4 1144 zones 5.1 (10.2)
• strip module by module
• layer 5 748 zones 3.3
• layer 6 950 zones 4.2

about 2 (4) M events (assuming 3 good
tracks/event) needed for evaluating the ITS
efficiency (relative error 0.5)
16
Proposal for ITS

• The measured Eff. will be used as reference for
  the response function
• as an overall factor, if the response model is
  100 efficiency
• as the reference to tune the response model, if
  not
• For special analyses and (eventually) for
  systematic error evaluation
• fast simulation the chip (module) efficiencies
  are the probabilities to have a digit
• sub-chip (sub-module) efficiencies (e.g. pixel by
  pixel for the SPD) may be obtained by combining
  the chip_by_chip efficiency (from this framework)
  and the hit maps (from QA ?).

17
Conclusions ?

• I would appreciate your feedback

18
SPD relative track occupancy
Assuming Vertex_z0
144 mm
144 mm
SPD2
76mm
SPD1
39mm
At the border
SPD1
SPD2
19
SPD regions uncovered by tracking
SPD1
39mm
7.0
6.5
6.5
r3.9cm
SPD1
SPD1
12.0
4
4
r7.6cm
SPD2
SPD2