1. Last update at the RPVLL meeting: https://indico.cern.ch/event/276607
Search for (semistable) charged heavy particles using the ionization in the Pixel Detector
A.Gaudiello1, C. Gemme1, G.Gagliardi1, E. Guido1, S. Passaggio1, L. Rossi1, C. Schiavi1
1INFN Genova
Last update at the RPVLL meeting:

2. Motivations for the searches with dE/dx
The ATLAS Pixel detector is able to measure the particle ionization in the very first cm of the trajectory and to tag tracks that have high release due to low bg particles.
The measurement is possible if the track is reconstructed and it is not sensitive to possible decays or interactions with material after the silicon layers. 
CMS analysis only for stable, using ID only, ID+MS, MS only: doi: /JHEP07(2013)122
Bethe-Bloch empirical fit of low mass particles allows to measure the particle mass from its energy release and momentum measurement
This calibration done in 2010 is very stable over the time.
Proton mass

3. Pixel-only Analysis 2012 Benchmark signals:
Stable squarks and gluinos R-hadrons, as in 2011, in extended mass range GeV;
Metastable squarks and gluinos R-hadrons with lifetime 10,1,0.1,0.01, ns and mass 400,600,800, 1000 GeV and decaying in a heavy or light neutralino.
Stable ( GeV) and metastable Charginos (same samples as kinked tracks analysis, lifetime ns).
Two main changes in 2012 wrt 2011 analysis:
In the DESDM RPVLL stream, the online skimming of the JetTauEtmiss stream has raised the high-pT track threshold from 50 to 80 GeV.
The un-prescaled triggers have changed wrt to 2011.
EF_xe80T_tclcw_loose; // All data periods

4. Stable squarks and gluinos R-hadrons
49%
31%
20%

5. Trigger selection and Efficiency
EF_xe80T_tclcw_loose triggers ~ 98% of the OR of the unprescaled MET triggers in signal events.
We prefer to adopt one single trigger for the full data period to make the study of the systematics simpler.
Trigger efficiency is ~17% on a wide range of masses (slightly larger than in 2011).
2012
2011

6. Event Selection
Cuts have been revised to optimize S/√B and populate the bkg samples.
Event level (after RPVLL filter and squeezing):
GRL*
offline confirmation of the MET trigger: Etmiss RefFinal > 130 GeV (was 80)
at least 1 primary vertex (PV) with at least 5 tracks
SUSY recommended event cleaning
Track level (at least one track with)
High-pT: pT>80 GeV, primary, n.Pixel hits >=3, nBLayer hits >=1 and n.SCT hits >=6
Isolated: no primary with pT>1 GeV in a cone of radius ΔR<0.25
Electron veto,
p-cuts: p>150 GeV & Δp/p<50% (was p>100 GeV)
High ionization: dE/dx > |η| |η| |η|3 MeV/g cm-2 *
(A-G)data12_8TeV.periodAllYear_DetStatus-v54-pro13-04_DQDefects _PHYS_StandardGRL_All_Good.xml
(H-L)data12_8TeV.periodAllYear_DetStatus-v61-pro14-02_DQDefects _PHYS_StandardGRL_All_Good.xml

7. Small sensitivity to Pile-up
Period J
Being the pixel detector at high granularity, it is not very sensitive to pile-up.
Anyhow MC needs to be properly reweighted with PU: effect is mainly on the skimming of the track.
high
low
Period J
mu<16
Eff rel
Eff tot
mu>30
trigger&&mu
677802
127810
75836
MET
78912
0.12
1.2E-01
15931
8137
0.11
1.1E-01 PV
78838
1.00
15915
8128
skim
55096
0.70
8.1E-02
11240
0.71
8.8E-02
5377
0.66
7.1E-02
isolation
6307
9.3E-03
1354
1.1E-02
566
7.5E-03
electronVeto
4249
0.67
6.3E-03
896
7.0E-03
398
5.2E-03
pCut
1818
0.43
2.7E-03
393
0.44
3.1E-03
181
0.45
2.4E-03
ionization 29
0.02
4.28E-05 6
4.69E-05 2
0.01
2.64E-05
jetCleaning

8. Slight increase in signal efficiency (7.7% vs 5.8%)
MC 2011 gluino 1000 GeV
MC 2012 gluino 1000 GeV
data 2012 (period J, ~3ifb)
data 2011 (5ifb) *
* Squeezed = Ef_xexx_yy and track pT>80 GeV, 2 pix, 6 SCT
Slight increase in signal efficiency (7.7% vs 5.8%)
Higher rejection for background (314 vs 40*5fib/3ifb candidates)

9. Outlook for stable
We are studying and revising the definition of the background samples. Analysis cuts could slightly change to allow enough statistics in the bkg samples.
Bkg1 (low ionization) as data but dE/dx <1.8
Bkg2 (low momentum) as data but p<150 GeV and pT>10 GeV
Next-to-next:
Study systematics.
Move to analysis for stable charginos.

10. Metastable squarks and gluinos R-hadrons
R-hadrons Decay in neutralinos (either close to the R-hadron mass or lighter).
Current selection is pretty efficient on MET.
Inefficiency comes if lifetime is too short to generate a reconstructable track.

11. “Reconstructable” charged R-hadrons
Ratio of
Charged R-hadrons associated (DR<0.01) to a track with pT>5 GeV (no other request!) and generated charged R-hadrons.
Clearly no/poor efficiency for rest lifetimes < 1 ns independently of any selection optimization.
In the next 2 slides, properties of the generated R-hadrons that are associated to reconstructed tracks. Not applied yet any selection cut.

12. “Reconstructable” charged R-hadrons: Properties IpT
nSCT
1 ns R-hadrons are reconstructed also at lower SCT hits than normal. p
1 ns R-hadrons are reconstructed if they are more boosted

13. “Reconstructable” charged R-hadrons: Properties II
dE/dx
mass
1 ns R-hadrons has broader mass distribution, Consequence of the high-p boost selection.

14. Event Selection
Cuts have been revised to optimize S/√B and populate the bkg samples.
Event level
Trigger: EF_xe80T_tclcw_loose
GRL*
offline confirmation of the MET trigger: Etmiss RefFinal > 130 GeV (was 80)
at least 1 primary vertex (PV) with at least 5 tracks
SUSY recommended event cleaning
(A-G)data12_8TeV.periodAllYear_DetStatus-v54-pro13-04_DQDefects _PHYS_StandardGRL_All_Good.xml
(H-L)data12_8TeV.periodAllYear_DetStatus-v61-pro14-02_DQDefects _PHYS_StandardGRL_All_Good.xml

15. Metastable R-hadrons: MET@generation
gluino 600 Neutralino 500:
Met ~ 90 GeV
All the energy of the neutralinos escape, but there are 2 gluinos back-to-back.
When the neutralino is very light, there is ymuch more MET, because there are high pt jets coming out at random angles, and the escaping pt does not cancel on the two sides.
gluino 600 Neutralino 100:
Met ~ 220 GeV

16. Metastable R-hadrons: MET@afterTrigger
gluino 600 Neutralino 500:
Met ~ 90 GeV
600->500
stable
10 ns
1 ns
EF_xe80T_tclcw_loose
0.25
0.47
0.60
MET>130
0.73
0.52
gluino 600 Neutralino 100:
Met ~ 220 GeV
600->100
stable
10 ns
1 ns
EF_xe80T_tclcw_loose
0.25
0.84
0.91
MET>130
0.73
0.82
0.86

17. Event Selection
Cuts have been revised to optimize S/√B and populate the bkg samples.
Event level
Trigger: EF_xe80T_tclcw_loose
GRL*
offline confirmation of the MET trigger: Etmiss RefFinal > 130 GeV (was 80)
at least 1 primary vertex (PV) with at least 5 tracks
SUSY recommended event cleaning
Track Level (at least one track with)
High-pT: pT>80 GeV, primary, n.Pixel hits >=3, nBLayer hits >=1 and n.SCT hits >=4 (was 6 in stable)
Isolated: no primary with pT>1 GeV in a cone of radius ΔR<0.25
Electron veto,
p-cuts: p>150 GeV & Δp/p<50% (was p>100 GeV)
High ionization: dE/dx > |η| |η| |η|3 MeV/g cm-2

18. Metastable R-hadrons: pT of skimmed tracks
gluino 600 Neutralino 500:
Lower momentum if gluino is unstable because R-hadron is not reconstructed.
Close only if lifetime is 10 ns, mainly is X0 is light (there might be other high-p charged tracks in the decay)
gluino 600 Neutralino 100:

19. Metastable R-hadrons: dE/dx of skimmed tracks
gluino 600 Neutralino 500:
Visible that for short lifetime there is less contribution of R-hadrons (high ionization)
gluino 600 Neutralino 100:

20. Cut flow for 600 GeV (meta) stable Rhadrons
600->500
STABLE
10 ns
1 ns
0.1 ns
0.01 ns
0.001 ns
generated
20000
9400
10000
9999
GRL
1.00
cleaning
19925
8503
0.90
9979
9980
9983
9984
trigger
5005
0.25
3985
0.47
6020
0.60
6153
0.62
6193
6163
MET>130
3662
0.73
2070
0.52
3125
3350
0.54
3561
0.58
3649
0.59 PV
3659
2060
3111
3326
0.99
3527
3634
skim SCT 4
1919
915
0.44
332
0.11
163
0.05
175
isolation
1637
0.85
817
0.89
184
0.55 1
0.01 2
electronVeto
1622
814
pCut 150
1593
0.98
729
114
0.50
ionization
1160
499
0.68 61
0.00
jetCleaning
0.058
0.0531
0.0061
0.0001
600->100
9900
9000
9500
8133
0.82
8954
9465
9960
9964
6827
0.84
8174
0.91
8706
0.92
9149
9182
5602
7058
0.86
7695
0.88
8269
8322
5522
7023
7613
8171
8214
2104
0.38
472
0.07
266
0.03
790
0.10
1061
0.13
1904
318
0.67 11 21
0.02
1900 8 15
0.71
1747
187 4 7
1282
120
0.64
0.1295
0.0133

21. Matching with Jets or muon MS
10 ns
600 100
1 ns
Rhadron 600 Stable
10 ns
600500
1 ns
MuSpectro: candidate track geometrical matched with a muid
Jets only: candidate track geometrical matched with a Jet but not with a muid
Others: candidate track geometrical not matched with a Jet or with a muid
 For 10 (1) ns lifetimes the pixel-only analysis is the dominating (only) search method.

22. Outlook for metastable
As for the stable case, study and revise the definition of the background samples mainly to populate the samples sufficiently.
Bkg1 (low ionization) as data but dE/dx <1.8
Bkg2 (low momentum) as data but p<150 GeV and pT>10 GeV
Next-to-next:
Generate several lifetime samples,
as in the disappearing tracks search,
Study systematics.
Show limits vs lifetime and mass
Move to analysis for metastable charginos.
as illustration

23. Backup
04/11/2013
C. Gemme – INFN Genova

24. Cleaning in SUSY Implemented SUSY cleaning: at the EVENT level
if ((nt->larError)==2) continue; // Removing LAr error events
if ((nt->tileError)==2) continue; // Removing Tile error events
if ((nt->coreFlags&0x40000) != 0) continue; // Remove incomplete events due to TTC reset
larError, tileError and coreFlags NOT SQUEEZED yet.
// TTileTripReader::checkEvent(unsigned int run, unsigned int lbn, unsigned int event) \\Tile trips in periods G-J ????? Also information are missing in SUSY instructions
recommend to veto events which have any jet pt>40 && & BCH_CORR_JET > 0.05 && DeltaPhi(MET,Jet)<0.3  cuts 10% of GRL events
and with any looser (=BadlooseMinus, 0.15% of jets) jets with pt>20 GeV and any eta  cuts <1% of GRL events (SIMILAR TO THE JET Cleaning that we have used in the candidates events in 2011)