[PPT] - Techniques to Search for 2 Charged Long Lived Particles Charged PowerPoint Presentation

SLIDE 1

Techniques and results

in Charged Long-Lived particle searches in ATLAS and CMS in Run 2

NORA PETTERSSON (UNIVERSITY OF MASSACHUSETTS, AMHERST) ON BE HALF OF THE ATLAS AND CMS COLLABORATIONS

SLIDE 2

Techniques to Search for Charged Long Lived Particles

1.

Charged particles that only traverse a certain extent of the tracking detector and subsequentially disappear

 Employ non-standard track reconstruction to find short tracks

 Veto hits in “outer” tracker volume to ensure short tracks

2.

Highly ionizing particles leaving abnormal energy losses in the detector – 𝑒𝐹/𝑒𝑦 measurements

 Utilise the measuring capabilities of the tracking detector

3.

Time of flight measurements using timing information available from the calorimeters and muon spectrometer

4.

Displaced Vertices inside the tracking volume

2

e e e e e e e e e e

SLIDE 3

ATLAS and CMS Experiments

 Two Large experiments at CERN!

 Probably heard all about them in previous talks

 Long-lived particles yield non standard signals

 It is vital to understand the performance of the

detector!

3

SLIDE 4

Disappearing Track (ATLAS)

 Assume a SUSY model where 𝜓1

± (NLSP)

is nearly mass-degenerate with 𝜓1

0 (LSP)

– Long-lived 𝜓1

± decays: 𝜓1 + → 𝜓1 0𝜌+

(soft)

 Common to Wino and Higgsino LSP

scenarios – vital to a large portion of SUSY dark matter searches

4

JHEP 06 (2018) 022

 Gives a signature of a charged track that seemingly disappears after crossing only few layers

f the inner detector

 Need to reconstruct the short tracks (tracklets) using only measurements (hits) expected for the given

lifetime spectrum

 In this case, restrict to the pixel detector and measurements up to ~120 mm

SLIDE 5

Disappearing Track (ATLAS)

 Track reconstruction is done in two steps for this analysis

 Standard algorithms– e.g. to find mainly the primary tracks

 Requiring at least seven measurements in the silicon detector

layers

 A second pass of the tracking

 Using only leftover measurements from the first pass  The hit requirement is significantly looser and aimed at short

tracks: at least four hits in the pixel layers

 Addition of the insertable B-Layer (IBL) improved the

efficiency pixel tracklet reconstruction efficiency

 Up to 60% efficiency to reconstruct tracklets in the pixel

detector volume, up to 300 mm

 Veto is applied to make sure that the tracklets do not

have any hits in the silicon tracker (SCT)

 Effective background and fake removal

5

JHEP 06 (2018) 022 Pixel SCT

SLIDE 6

Disappearing Track (ATLAS)

 Backgrounds arise from hadrons or leptons that may

interact with the detector material as well as combinatoric backgrounds of tracklets made out of random hits

 Producing templates of the tracklet pT distribution varying

n the type of expected background

 Likelihood fit performed on the signal and background

templates

6

JHEP 06 (2018) 022

 Limits are set on the 𝜓1

± mass as a function lifetime  IBL help improving the limits for run-2 due to the increased

reconstruction efficiency for pixel tracklets

 Reinterpretation of this analysis on the Higgsino scenario is

covered in ATL-PHYS-PUB-2017-019

Run-2 improvement

SLIDE 7

Disappearing Track (CMS)

 CMS have a smiliar search for the same model and

topology

 Slightly different analysis strategies

 The disappearing track candidates are required to be

short and to have no hits in the outer layers of the tracking volume

 This suppresses background from random combinations

and from tracking inefficiencies that can create spurious short tracks

 Require strict quality cuts on the short tracks

 Restriction on the impact parameters  Require no missing hits in the inner layers

7

JHEP 08 (2018) 016

SLIDE 8

Disappearing Track (CMS)

 Missing number of outer hits is used to

select short track candidates for the analysis

 Powerful discriminator of signal versus

background

 Reduce QCD background by angle cuts

between the jets and the missing pT

 Remaining backgrounds are:

 Charged leptons that fail lepton

identifications

 Spurious tracks from random hits

 Both are estimated in dedicated regions

enhancing the contributions

8

JHEP 08 (2018) 016

SLIDE 9

Disappearing Track (CMS)

 Limits are set on the cross section of the 𝜓1

± as well as a function of the lifetime

 The limits are set on the cross section for lifetimes between 0.1 and 100 ns

 𝜓1

± masses up to 715 (695) GeV are excluded for lifetimes of 3 (7) ns,  This is the range of lifetimes the analysis is most powerful

 Masses of up to 505 GeV are excluded for the broader range of 0.5 ns to 60 ns

9

JHEP 08 (2018) 016

SLIDE 10

Disappearing Track (CMS)

10

JHEP 08 (2018) 016

Different strategies: CMS optimised for longer lifetimes while ATLAS for shorter lifetimes

ATLAS NB: CMS results are pre- update and are still using a three layer pixel detector while ATLAS results are with a four layer pixel detector

SLIDE 11

Large ionization energy loss (ATLAS)

 Search for long-lived charged particles traversing the

inner detector (ID) and leaving large 𝑒𝐹/𝑒𝑦 deposits

 Interpreted on long lived R-hadrons hypothesised by

split-susy model

 Charge deposits per track length in the pixel layers

provides 𝑒𝐹/𝑒𝑦 measurements

 Adjacent fired pixels are combined into clusters  Cluster size depends on incident angle

 To reduce the tail fractions, a particle’s 𝑒𝐹/𝑒𝑦 is taken

as the average over all the pixel hits, removing one or two measurements with the largest deposits of energy

 IBL helps improving the capability of measuring the

energy loss more precisely

11

e e e e e e e e e e e e

Phys. Lett. B 788 (2019) 96

SLIDE 12

Large ionization energy loss (ATLAS)

 Energy losses are dependent on the mass

and the mass can be calculated for the LLP using the Bethe-Bloch formula

 Use fit range of 0.3 < 𝛾𝛿 < 0.9

 Corresponds well to the LLPs which are

expected to be produced at the LHC

 Fit shown for pions, kaons and protons

 Estimated masses from applying this

method on signal samples of R-hadrons, reproduced the generated mass well up to masses of 1.5 TeV

 Calibrations on protons in data shows

consistent results within 1% of the expectations

12

Phys. Lett. B 788 (2019) 96

SLIDE 13

Large ionization energy loss (ATLAS)

 Fully data-driven background estimation

 Derive shape and normalisations in control regions defined by inverting selections

 Limits set on the production cross section and lifetime of the gluino

 For lifetimes of and above 1 ns: 1290 to 2060 GeV excluded

13

Phys. Lett. B 788 (2019) 96

SLIDE 14

Heavy Stable Charge Particles (CMS)

 Search for heavy stable charge particles (HSCP)

with large ionization energies

𝑒𝐹 𝑒𝑦 and non-unit

charges

 The search considers two techniques

 A tracker-only approach and one where the tracker

information is combined with the muon system (tracker and time of flight (TOF))  Considering three models that exploits the two

different techniques

 For example, split SUSY with R-hadrons that are either

stables or are expected to lose their charge before the muon system

 Staus postulated in mGMSB  Lepton like fermions in a Drell-Yan model

14

Phys. Rev. D 94 (2016) 112004

EXO-16-036

SLIDE 15

Heavy Stable Charge Particles (CMS)



A particle’s energy loss is measured from ionization deposited in the pixel and silicon tracker layers

 Exclude the measurement with the smallers charge

deposit

 Increase the quality and reduce instrumental biases

 Powerful discriminating variable is defined by

comparing the measured values with what is expected of a minimum-ionizing particle

 Provide good separation of SM backgrounds

15

EXO-16-036

J. High Energy Phys. 03 (2011) 024

SLIDE 16

Heavy Stable Charge Particles (CMS)

 No excess observed in either analysis and limits are set on the three models

 For split-susy gluino masses below 1850 GeV are excluded Stop masses below 1250 GeV are excluded  Stau masses below 660 GeV are excluded for the GMSB and below 360 GeV for direct pair production model  Drell-Yan signals with |Q| = 1e (2e) are excluded below 730 (890) GeV

16

EXO-16-036

SLIDE 17

Heavy Stable Charge Particles (CMS)

 No excess observed in either analysis and limits are set on the three models

 For split-susy gluino masses below 1850 GeV are excluded Stop masses below 1250 GeV are excluded  Stau masses below 660 GeV are excluded for the GMSB and below 360 GeV for direct pair production model  Drell-Yan signals with |Q| = 1e (2e) are excluded below 730 (890) GeV

17

EXO-16-036

ATLAS - Gluino at 36.1 fb-1

SLIDE 18

Multi-charged LLP (ATLAS)

 ATLAS have a search dedicated to only multi-charge particles (MCP)

 Results also interpreted on the Drell-Yan production model like the previous CMS analysis

 Assume the particles decay outside the detector so they appear stable and leave muon-like

signatures with large energy loss

 Measure dE/dx in the pixel, transition radiation tracker (TRT), and in MDT subsystem in the muon

spectrometer

 dE/dx from the pixels is estimated as discussed for previous analyses, in the TRT the dE/dx is a mean of the hit-level

energy losses calculate for the each tracks time above threshold, and similar for the MDT an average is taken from all drift tubes crossed

18

arXiv:1812.03673

Emission of many 𝜀 for higher charge broaden the distribution Miss modelling in simulation due to gas-change in the TRT not being propagate to MC

SLIDE 19

Multi-charged LLP (ATLAS)

 Two signal regions are defined

depending on the expected charge

 Needed by the different detector

responses for z= 2 and z> 2

 Expected backgrounds are due to

possible high occupancy in the detector and the presence of large amount of 𝜀-rays

19

arXiv:1812.03673

 Background estimated by ABCD method for z=2 and using

the side bands of MDT / TRT dE/dx distributions for z > 2

 No significant excess observed and limits are set on DY

model and multi-charged lepton-like particles

 From 50 GeV up to 980-1220 GeV are excluded

SLIDE 20

Multi-charged LLP (ATLAS)

 Two signal regions are defined

depending on the expected charge

 Needed by the different detector

responses for z= 2 and z> 2

 Expected backgrounds are due to

possible high occupancy in the detector and the presence of large amount of 𝜀-rays

20

arXiv:1812.03673

 Background estimated by ABCD method for z=2 and using

the side bands of MDT / TRT dE/dx distributions for z > 2

 No significant excess observed and limits are set on DY

model and multi-charged lepton-like particles

 From 50 GeV up to 980-1220 GeV are excluded CMS z=2 and 12.9 fb-1

CMS targeted a wider mass range

SLIDE 21

Displaced Vertex (ATLAS)

 Search for LLP decaying inside the inner detector to several charge particles

 Results interpreted for R-parity violating SUSY where a stop decays to a quark and a muon

 Standard track reconstruction limits the efficiency

 Impose strict cuts on transverse (d0) and longitudinal (z0) impact parameters with respect to the IP

 Use special track reconstruction

 A dedicated second pass of the tracking is ran on leftover hits from the standard tracking with looser

cut on d0 and z0

 The search looks for a displaced vertex with a mass larger than 20 GeV and have a track

multiplicity larger than three and a displaced muon is required to be present with large impact parameters of|d0| > 2

 Additional quality criteria are imposed to reduce backgrounds from detector effects such as fake

muons

 A cosmic veto is applied to reduce the largest background of displaced muons

21

ATLAS-CONF-2019-006

SLIDE 22

Displaced Vertex (ATLAS)

22

More details on DV+μ in a dedicated talk by Karri Di Petrillo ATLAS-CONF-2019-006



Backgrounds are estimated with a fully data driven method



Relies on the fact that variables used to reduce SM background for the displaced vertices and the displaced muons are not correlated



Control regions are defined by inverting parts of the selection



No significant excess observed and limits are set



Stop masses up to 1.7 TeV are excluded for lifetimes of 0.1 ns

 For the range of 10-17 ns the range up to 1.4 TeV is excluded

SLIDE 23

LLP Searches for ATLAS and CMS

23

 Many interesting results from “non” standard search techniques

 Utilise the ATLAS and CMS detectors’ full potential!

 Stay tuned for interesting future developments!!!

CMS Combined ATLAS Combined

SLIDE 24

BACK-UP SLIDES

MORE INFORMATION HERE

24

SLIDE 25

ATLAS LLP Summary of Results

25

SLIDE 26

CMS LLP Summary of Results

26

SLIDE 27

Large 𝑒𝐹/𝑒𝑦 and time of flight (TOF) (ATLAS)

 An other search from ATLAS includes in addition to the dE/dx

measurement als the time of flight

 Search for R-hadrons (split-SUSY), directly produced staus (GMSB), and

charginos (mAMSB) utilising pixel 𝑒𝐹/𝑒𝑦 measurements and TOF

 The 𝛾𝛿 estimated from the particles energy loss uses the same method as

in the previous analysis

 The velocity β of the particles are determined by time of flight

measurements

 Measurements from the tile calorimeter and from the monitored drift tubes

(MDTs) and resistive-plate chambers (RPCs) in the muon spectrometer

 Both TOF measurements are combined into an average 𝛾𝑈𝑃𝐺 factoring in the

resolution of the two systems  𝑎 → 𝜈𝜈 events are used to derive the resolution on the β-distribution for

the two detectors

27

arXiv:1902.01636

SLIDE 28

Large 𝑒𝐹/𝑒𝑦 and TOF (ATLAS)



Few signal regions are defined and optimised for the different expected signatures of the three models under consideration



All five regions are mutually exclusive



No excess observed and limits are set on R-hadrons and direct pair-production of staus/charginos



Lower limits on the mass of long lived gluinos, sbottom and stop R-hadrons are set at 2000 GeV, 1250 Gev and 1340 GeV



Lower limits on the mass of long lived staus and charginos are set at 430 GeV and 1090 GeV

28

arXiv:1902.01636

SLIDE 29

Disappearing Track (ATLAS)

 Backgrounds arise from hadrons or leptons that may interact with the detector material as well

as combinatoric backgrounds of tracklets made out of random hits

 Hadronic interactions, multiple scattering, bremsstrahlung, and fakes

 Minor inefficiency and low purity start to play a role for these types of analysis

 Red solid (dotted) shows charged (neutral) particles; thick blue represents the reconstructed

tracklet

29

JHEP 06 (2018) 022 Scattering where the extended track is missed due to the large kink Similarly for the emission of a hard photon, the extension is judged not to be part of the same track A tracklet made up of hits produced by several particles or noise hits

SLIDE 30

Disappearing Track (CMS)

 Remaining background consists of charged leptons that fail lepton identifications while the

missing pT requirements are still met

 Contribution in the signal region is estimated by creating the probability of this type of leptons

 Spurious track background is estimated from a control region enhanced with lower quality

tracks by requiring larger impact parameters

 Number of events is scaled by a transfer factor derived from tracks with three pixel hits but no hits in

the middle detector and requiring the impact parameter selections of the signal region

30

JHEP 08 (2018) 016 Estimated via tag-and-prob for 𝑎 → 𝜈𝜈 and 𝑎 → 𝑓𝑓 Probability that a single lepton event passes the missing pT selection given that the lepton isn’t identified Add extra requirement that to the Poffline that the event also passes the triggers Number of events seen in the control region

SLIDE 31

Disappearing Track (ATLAS)

 Event selection of disappearing tracks

 Expecting missing ET due to the neautralino → Rely on missing ET trigger  Require a pixel tracklet with pT > 5 GeV and no associated SCT hits  Lepton veto is applied to reduce background of V+jet and 𝑢 ҧ

𝑢 events

 Various quality requirements on the pixel tracklet to ensure good quality

 Further selection on kinematics optimised for the two channels, e.g. more jets for the strong

channel

31

JHEP 06 (2018) 022

SLIDE 32

Large ionization energy loss (ATLAS)

 Corrections for 𝑒𝐹/𝑒𝑦 applied to simulation

 Changes in the measured changes during runs

depending on run conditions and accumulated luminosity

 Effect such as radiation damange on the pixel sensors

are not taking into account in simulation and need to be corrected for  Calibrations for low momentum particles under 500

MeV corrections are applied for kaons and protons

 The pion mass is the default hypothesis in track

reconstruction and the calibrations are produced from a fit between the simulated mass and the reconstructed mass to account for this effect  Variation in the energy losses depends on the

particles indicent angle on the sensors

 After applied corrections 𝑒𝐹/𝑒𝑦 only depend on

momentum and mass of the particles

32

Phys. Lett. B 788 (2019) 96

SLIDE 33

Large 𝑒𝐹/𝑒𝑦 and TOF (ATLAS)



Fully data-driven background estimate by producing probability density functions (pdfs) of the key distributions 𝛾𝑈𝑃𝐺, (𝛾𝛿)𝑒𝐹/𝑒𝑦 for R-hadrons and momentum and 𝛾𝑈𝑃𝐺



The expected number of events in the signal regions are estimated by randomly sampling from the pdfs using 𝑛 = 𝑞/𝛾𝛿



Possible correlations between 𝛾𝑈𝑃𝐺, (𝛾𝛿)𝑒𝐹/𝑒𝑦 and the momentum are taken into account by binning the pdfs in pseudorapidity η

 𝛾𝑈𝑃𝐺, (𝛾𝛿)𝑒𝐹/𝑒𝑦 are η-dependent due to that the resolutions varies dependen on the detector region

33

arXiv:1902.01636

SLIDE 34

Large 𝑒𝐹/𝑒𝑦 and TOF (ATLAS)

 Few signal regions are defined and optimised for the different expected signatures of the three

models under consideration

 All five regions are mutually exclusive

 The lower limits on the mass are derived from mass-planes of 𝑛𝑈𝑃𝐺 and 𝑛𝑒𝐹/𝑒𝑦 for the R-hadrons

while only from 𝑛𝑈𝑃𝐺 for the stau/chargino regions

34

arXiv:1902.01636

Do not require any MS activity → region less dependent on the hadronization model Take advantage of MS information → better TOF measurements No dE/dx requirement imposed for the pair- produced stau/chargino to be effective for low- masses as well

SLIDE 35

Heavy Stable Charge Particles (CMS)

 Event selections of HSCP

 High transverse momentum single muon trigger or missing ET trigger used to select events  Track candidate with pT > 55 GeV and various quality requirements to ensure good tracks  The tracker + TOF analysis also require that the track should be matched to a reconstructed muon

and at least eight time measurements

 Data driven background estimate ABCD method using two non-correlated variables

 The regions are divided up using pT > 65 GeV and 𝐽𝑏𝑡 > 0.3  Candidates found in the control regions are used to form binned probability density functions of 𝐽ℎ

and momentum using the mass estimated for SM extrapolated to the signal region

 For the tracker+TOF analys an additional dimension is added with 1

𝛾 > 1.25 to form ABCDEFGH method

35

Phys. Rev. D 94 (2016) 112004

SLIDE 36

Heavy Stable Charge Particles (CMS)



Time of flight is measured in the muon system from the Drift Tubes (DT) and Cathode Strip Chambers (CSC)

 Slow particles can be distinguished from those traveling near

speed of light

 A relativistic particle will produce an aligned pattern of

hits in the DT while a slower particle will have a reconstructed position shifted relative to the path

 The offset of position is proportional to the delay of the

particle

 The CSC measure the delay for each hit separately

36

Phys. Rev. D 94 (2016) 112004