1. A.Rozanov CPPM 8.10.07 Highlights of b-layer upgrade workshop
Outline
V.Kostioukhine, P.Nevski. A.Rozanov
Motivations for improvements in b-layer
Test layout for MC studies
Additional b-layer at R=3.7 cm
Material budgets
Pixel Occupancy
A.Rozanov CPPM Highlights of b-layer upgrade workshop

2. A.Rozanov CPPM 8.10.07 Highlights of b-layer upgrade workshop
b-tagging for SM Higgs
MH(GeV) 5s
b-tagging role in:
Discovery channel
ttH with H-> bb
Supplementary channels
WH with H-> bb
ttH with H-> WW*
Background channels
ttjj
ttbb
A.Rozanov CPPM Highlights of b-layer upgrade workshop

3. Performance contributions
Resolutions Rphi, Z
Pitch, cluster size, charge sharing (tilt,Lorenz),analog vs digital readout
Pattern recognition: efficiency, fakes, ambiguous hits, shared hits
Material: multiple scattering, secondary, conversions, b-layer radius
A.Rozanov CPPM Highlights of b-layer upgrade workshop

4. Pileup influence on b-tagging:
Wrong selection of primary vertex (dominant, but manageable in analysis…)
Tracks from pileup vertices in jets (with big Z impact parameters of course)
Zvertex resolution for current ID is 50m
<Nvrt> - mean number of pileup vertices in ±150m from primary vertex
IR = 5.6cm Nint = 23
IR = 2.1cm Nint = 88
IR = 2.1cm Nint = 176
IR = 3.5cm Nint = 510
<Nvrt> at Z=0mm Z=20mm
0.05
0.048
0.49
0.31
0.99
0.63
1.71
1.48
A.Rozanov CPPM Highlights of b-layer upgrade workshop

5. A.Rozanov CPPM 8.10.07 Highlights of b-layer upgrade workshop
Pileup
Minimal distance between primary and pileup vertices at Z=0
IR = 5.6cm Nint = 23
IR = 2.1cm Nint = 88 mm mm
Better Z track and primary vertex resolution is needed to improve primary vertex reconstruction and rejection of pileup tracks (track - vertex association)
A.Rozanov CPPM Highlights of b-layer upgrade workshop

6. Performances versus jet pT, η
Non-uniform performances:
even if very good at ||=0 and 50<P<150 GeV, still not satisfactory at ||~2 and P>250GeV
tagging b-jets can bias kinematics
IP2D
IP3D+SV1
εb= 60%
IP2D
IP3D+SV1
εb= 60%
Worsening of b-tagging performance at ||>1 is due to material.
Any way to improve?
A.Rozanov CPPM Highlights of b-layer upgrade workshop

7. A.Rozanov CPPM 8.10.07 Highlights of b-layer upgrade workshop
Pixel layout
||=2.0
112mm
||=2.5
A.Rozanov CPPM Highlights of b-layer upgrade workshop

8. A.Rozanov CPPM 8.10.07 Highlights of b-layer upgrade workshop
Setup + software
ATLSIM (Geant3 simulation) rel DC1 model of ID (pixel). xKalmanOO from HEAD. Private version of b-tagging software.
One pixel layer is added (Norbert’s proposal parameters): R=37mm 16 staves (old DC1 design) Zpitch=250µm (no long pixels and gap between chips) Tilt=100 Sensor thickness=250µm Sensor and geometry are old  overlaps are not well defined (will be corrected soon)
A.Rozanov CPPM Highlights of b-layer upgrade workshop

9. A.Rozanov CPPM 8.10.07 Highlights of b-layer upgrade workshop
Layout
Rb1= 37.0mm Rb2= 50.5mm R1 = 88.5mm R2 = 122.5mm
To reduce thickness of new layer:
Support shell is absent
Thickness of stave (nonsensitive) is decreased
A.Rozanov CPPM Highlights of b-layer upgrade workshop

10. Layout Material budget
To reduce thickness of new layer:
Old b-layer R=5.05 cm 2.2 % X0
New b-layer R=3.7 cm % X0
A.Rozanov CPPM Highlights of b-layer upgrade workshop

11. A.Rozanov CPPM 8.10.07 Highlights of b-layer upgrade workshop
Layout
Geant3 simulation of 3GeV pion in new pixel layout
4 hits in pixel det.
A.Rozanov CPPM Highlights of b-layer upgrade workshop

12. Pixel occupancy at ||=0
Geant3 simulation of new pixel layout. 24 ev pileup (1034), IR=5.6cm, WH(120)µuu events x60.0 mm2 sensor with 250µm Z pitch. Number of hits in module at ||=0
New layer R=37mm
Existing layer R=50.5mm
A.Rozanov CPPM Highlights of b-layer upgrade workshop

13. Pixel occupancy at ||=2.5
Geant3 simulation of new pixel layout. 24 ev pileup (1034), IR=5.6cm, WH(120)µuu events x60.0 mm2 sensor with 250µm Z pitch. Number of hits in module at ||=2.5
New layer R=37mm
Existing layer R=50.5mm
A.Rozanov CPPM Highlights of b-layer upgrade workshop

14. Amount of silicon hits (Pixel+SCT) for “good” tracks 3L 4L
Reconstruction
ATRECON clusterisation and xKalmanOO from CVS HEAD. No special tuning for 4layer pixel. As reference a 3layer pixel (new b-layer removed) is used.
Amount of silicon hits (Pixel+SCT) for “good” tracks 3L L
Additional hit is clearly visible  reconstruction sees additional layer
A.Rozanov CPPM Highlights of b-layer upgrade workshop

15. Primary vertex (H(120)->uu events)
XY(x100) resolution 3L L
~10% in XY ~30% in Z improvement in resolution due to 4th layer with 250µm pitch
Z resolution 3L L
A.Rozanov CPPM Highlights of b-layer upgrade workshop

16. Modified layout with single b-layer
Rb1= 37.0mm Rb2= absent R1 = 88.5mm R2 = 122.5mm
A.Rozanov CPPM Highlights of b-layer upgrade workshop

17. Performance equivalences in old studies
Decrease of the b-layer radius from 5 cm to 4 cm (20% effect)
Decrease of the z-pitch in b-layer from 400 um to 200 um
Decrease of the material in b-layer by 0.6%
Loss of 1% / 2% pixel module/chip inefficiency in b-layer
Partial (2/3) missing the intermediate pixel layer (R=9cm)
Pile-up of 1034 cm2s-1 without muon pointing to primary vertex.
A.Rozanov CPPM Highlights of b-layer upgrade workshop

18. 2 mass for WH(120)->uu events 3L 4L
Secondary vertices
2 mass for WH(120)->uu events 3L L
K0 peek is the same for both layouts  no big problems in pattern recognition
R of interaction vertices (material) for WH(120)->uu events after conversion removal 3L L
Additional peek due to new layer is clear  good !
Reconstructed number of interactions on beam pipe is smaller  problem ???
A.Rozanov CPPM Highlights of b-layer upgrade workshop

19. A.Rozanov CPPM 8.10.07 Highlights of b-layer upgrade workshop
b-tagging: results
Preliminary!!!
WH(120)->uu(bb), no pileup, ATLFAST jets, reconstructed primary vertex. “Shared” pixel treatment switched off.
3 layers
4 layers
60%
70% 2D
58.5 ± 6
18.3 ± 1
38.5 ± 3
12.0 ± 0.5 Z
12.3 ± 0.6
4.6 ± 0.1
13 ± 0.6
5.1 ± 0.1
2D+Z(3D)
92 ± 11
28 ± 1.8
70 ± 7.
18.1 ± 1
3D+SV1
211 ± 39
63. ± 6
170 ± 28
33 ± 2.3
3D+SV2
183 ± 31
54 ± 5
177 ± 30
30 ± 2
Tracking performance seems ok for 4 layers case but b-tagging is worse comparing with current design. Track part of b-tagging is mainly responsible for worsening. Reconstruction tuning gives some improvement in b-tagging for 4L case. Not clear if it allows finally to overcome initial performance.
A.Rozanov CPPM Highlights of b-layer upgrade workshop

20. b-tagging: illustration
Track 2D impact significance with respect to primary vertex u-jets 3L L
Distributions are obtained AFTER conversions, K0/s and material interactions removal (although removal is based on reconstruction so not ideal…)
A.Rozanov CPPM Highlights of b-layer upgrade workshop

21. b-tag:result for single b-layer(37mm)
Preliminary!!!
WH(120)->uu(bb), no pileup, ATLFAST jets, reconstructed primary vertex. “Shared” pixel treatment switched off.
3 layers current
3 layers upgraded
60%
70% 2D
58.5 ± 6
18.3 ± 1
78.5 ± 9
20.0 ± 1. Z
12.3 ± 0.6
4.6 ± 0.1
22 ± 1.2
7.2 ± 0.2
2D+Z(3D)
92 ± 11
28 ± 1.8
178 ± 30
31.1 ± 2
3D+SV1
211 ± 39
63. ± 6
350 ± 80
68.5 ± 7
3D+SV2
183 ± 31
54 ± 5
356 ± 83
67.9 ± 7
Track reconstruction program is the same in both cases
A.Rozanov CPPM Highlights of b-layer upgrade workshop

22. ||>1 with single b-layer(37mm)
WH(120)->uu(bb), no pileup, ATLFAST jets, reconstructed primary vertex. High ||>1 region. “Shared” pixel treatment switched off.
Preliminary!!!
3 layers current ||>1
3 layers upgraded ||>1
60%
60% (||>1)/all
70%
70% (||>1)/all 2D
28.3 ± 6
0.48
11.8 ± .8
0.64
52.7 ± 6
0.67
15.0 ± .9
0.75 Z
9. ± .4
0.73
3.8 ± 0.1
0.83
17.7 ±1.2
0.81
6.6 ± 0.3
0.92
2D+Z(3D)
54 ± 7.
0.59
16.2 ± 1.
0.58
113 ± 20
22.1 ±1.7
0.71
3D+SV1
128 ± 25
0.61
40.2 ± 4.
271 ± 73
0.77
53.8 ± 6
0.79
3D+SV2
140 ± 28
0.76
34.5 ± 4
238 ± 60
50.8 ± 6
With upgraded 3L pixel:
Rejection in high || region is bigger
Drop of efficiency at high || is smaller
Way to improve physics potential of pixel detector at high || ?
(see previous talk)
Less material in new layer (???). T0 cables are still in place.
A.Rozanov CPPM Highlights of b-layer upgrade workshop

23. Limited access scenario
B-layer is replaced without removing all the SQPs and Disk.
This implies to remove one SQP per side, the beam pipe and the existing BL.
The B-layer services are trapped in the gap between the barrel and the disk section and they have to be cut to extract the B-Layer. I have asked to evaluate what this means in term of contamination providing the quantity of activated material that is allowed to be dispersed in the air and on the surfaces.
What can be repaired is limited.
PP0 is accessible. Both leaks in the fitting and connector de-mate can be fixed.
Type0 cables are accessible and they can be repaired if the failure is outside the frame.
Optoboards located on the external OSP can be changed as well as the optoheaters
Disk capillaries can be changed if clough. Not the ones of the barrel
Most of the connectivity failures along the SQPs up to PP1
What cannot be repaired are leaks inside the Disk or in the barrel bi-staves; substitute sector or bi-staves; fix connectivity failure inside frame volume; change most of the optoboard .
D. Giugni
BL Replacement Workshop - Sept 2007- 23
A.Rozanov CPPM Highlights of b-layer upgrade workshop

24. Barrel access
B-layer is replaced disassembling all the SQP’s, the BPSS and the disk
What can be repaired is a lot.
All the SQPs can be refurbished. Accessing the OptoBoard and Optoheaters.
L1 and L2 Type0 cables are accessible up to the ends of the staves.
All the barrel fittings and many of the disk ones.
All the capillaries can be changed if clough.
What cannot be changed are bi-staves in L2 and L1 that would require a further disassembling
D. Giugni
BL Replacement Workshop - Sept 2007- 24
A.Rozanov CPPM Highlights of b-layer upgrade workshop

25. Summary Limited access 5.5 months Barrel access 9.5 months
Local Support access months
D. Giugni
BL Replacement Workshop - Sept 2007- 25
A.Rozanov CPPM Highlights of b-layer upgrade workshop